20 YEARS

ON THE WEB
19962016
A selection of the most read, cited,
and viewed articles on NEJM.org
 September 2016

Dear Reader,

NEJM.org celebrates a major milestone this year as we reach 20 years of publishing content on the web.
Just as digital technology has come a long way, NEJM.org is continuing to find innovative new ways to
present medical information. In 1996, NEJM.org launched and helped to redefine the nature of medical
publishing. Over the past 20 years we have added features such as Videos in Clinical Medicine, the
Image Challenge quiz, Interactive Medical Cases, and most recently, the Quick Take short video sum-
maries of major articles.

This digital collection and accompanying infographic show the progressive development over the past
20 years of the delivery of useful medical information for readers on NEJM.org. We hope that you enjoy
this collection and that you will continue to join us as we continue to explore new ways to deliver in-
sightful medical content.

Edward W. Campion, MD
Executive Editor and Online Editor
The New England Journal of Medicine

800.843.6356 | f: 781.891.1995 | nejmgroup@mms.org

860 winter street, waltham, ma 02451-1413
nejmgroup.org
TABLE OF CONTENTS

NEJM.ORG TOP CITED ARTICLES

Inflammation, Aspirin, and the Risk 1

of Cardiovascular Disease in Apparently Healthy Men

Declining Morbidity and Mortality among Patients 8

with Advanced Human Immunodeficiency Virus Infection

Overweight, Obesity, and Mortality from Cancer 16

in a Prospectively Studied Cohort of U.S. Adults

Improved Survival with Ipilimumab in Patients 30

with Metastatic Melanoma

Zika Virus Associated with Microcephaly 43

INFOGRAPHIC

FIRSTS ON NEJM.ORG 52
Audio: Classic Whooping Cough
Audio Interview: Statins and Over-the-Counter Availability
Videos in Clinical Medicine: Placement of an Arterial Line
Video Roundtable: Physicians and Execution
Highlights from a Discussion of Lethal Injection
Interactive Medical Case: The Writing on the Wall
Image Challenge: First Image Challenge
Quick Take Video Summary: In-Flight Medical Emergencies

(continued on next page)

The New England Journal of Medicine is a publication of NEJM Group, a division of the Massachusetts Medical Society.
2016 Massachusetts Medical Society, All rights reserved.
TABLE OF CONTENTS
(continued from previous page)

NEJM.ORG NOTABLE MULTIMEDIA 54

First Video Published on NEJM.org: Separation of
Conjoined Twins
Most Viewed Clinical Medicine Video: Hand Hygiene
Audio Webcast: Ebola Outbreak
Toughest Image Challenge: Dahls Sign
Image in Clinical Medicine: Femoral-Head Dislocation
Most Viewed Interactive Medical Case: A Man with Bizarre
Behavior

NEJM.ORG URGENT PUBLIC HEALTH ISSUES 56

Zika Virus
Ebola Outbreak
20 YEARS ON THE WEB
The New England
Journal of Medicine
Co pyr igh t , 1 997, by t he Mas s achus et ts Medic al So ciet y
VO LU ME 3 3 6
INFLAMMATION, ASPIRIN, AND THE RISK OF CARDIOVASCULAR DISEASE
IN APPARENTLY HEALTHY MEN
Paul M. RidkeR, M.d., MaRy CushMan, M.d., MeiR J. staMPfeR, M.d., Russell P. tRaCy, Ph.d.,
and ChaRles h. hennekens, M.d.
AbstrAct
Background Inflammation may be important in
the pathogenesis of atherothrombosis. We studied
whether inflammation increases the risk of a first
thromboticeventandwhethertreatmentwithaspi
rindecreasestherisk.
Methods WemeasuredplasmaCreactiveprotein,
a marker for systemic inflammation, in 543 appar
ently healthy men participating in the Physicians
HealthStudyinwhommyocardialinfarction,stroke,
orvenousthrombosissubsequentlydeveloped,and
in543studyparticipantswhodidnotreportvascular
disease during a followup period exceeding eight
years.Subjectswererandomlyassignedtoreceiveas
pirinorplaceboatthebeginningofthetrial.
Results Baseline plasma Creactive protein con
centrationswerehigheramongmenwhowentonto
havemyocardialinfarction(1.51vs.1.13mgperliter,
P,0.001)orischemicstroke(1.38vs.1.13mgperliter,
P50.02),butnotvenousthrombosis(1.26vs.1.13mg
perliter,P50.34),thanamongmenwithoutvascu
larevents.Themeninthequartilewiththehighest
Creactive protein values had three times the risk of
myocardialinfarction(relativerisk,2.9;P,0.001)and
two times the risk of ischemic stroke (relative risk,
1.9;P50.02)ofthemeninthelowestquartile.Risks
werestableoverlongperiods,werenotmodifiedby
smoking,andwereindependentofotherlipidrelat
edandnonlipidrelatedriskfactors.Theuseofas
pirinwasassociatedwithsignificantreductionsinthe
riskofmyocardialinfarction(55.7percentreduction,
P50.02)amongmeninthehighestquartilebutwith
onlysmall,nonsignificantreductionsamongthosein
thelowestquartile(13.9percent,P50.77).
Conclusions The baseline plasma concentration
ofCreactiveproteinpredictstheriskoffuturemyo
cardialinfarctionandstroke.Moreover,thereduction
associatedwiththeuseofaspirinintheriskofafirst
myocardialinfarctionappearstobedirectlyrelated
to the level of Creactive protein, raising the pos
sibilitythatantiinflammatoryagentsmayhaveclini
cal benefits in preventing cardiovascular disease.
(NEnglJMed1997;336:9739.)
1997,MassachusettsMedicalSociety.
Back to Table of Contents
nejm.org
A p r i l 3, 1997 NUMB ER 14
T
HROMBUS formation is the proximate
cause of myocardial infarction, but ath
erosclerosis, the chief underlying cause, is a
chronic disease that progresses over dec
ades of life.1 Laboratory and pathological data sup
port the idea that inflammation has a role in both
the initiation and the progression of atherosclerosis,
and antiinflammatory agents may have a role in the
prevention of cardiovascular disease.25 However, there
are few data to indicate whether inflammation increas
es the risk of first myocardial infarction, stroke, and
venous thrombosis or whether antiinflammatory ther
apy decreases that risk.
Creactive protein is an acutephase reactant that
is a marker for underlying systemic inflammation.
Elevated plasma concentrations of Creactive protein
have been reported in patients with acute ischemia6
or myocardial infarction7,8 and have been found to
predict recurrent ischemia among those hospitalized
with unstable angina.9 Creactive protein is also as
sociated with a risk of myocardial infarction among
patients with angina pectoris10 and with a risk of fa
tal coronary disease among smokers with multiple
risk factors for atherosclerosis.11 However, since con
centrations of Creactive protein and other acute
phase reactants increase after acute ischemia6 and are
directly related to cigarette smoking,11,12 it has been
uncertain whether associations observed in previous
studies of acutely ill patients9 or highrisk popula
From the Divisions of Preventive Medicine (P.M.R., C.H.H.) and Car
diovascular Disease (P.M.R.) and the Channing Laboratory (M.J.S.), De
partment of Medicine, Brigham and Womens Hospital; the Department of
Ambulatory Care and Prevention, Harvard Medical School (C.H.H.); and
the Departments of Epidemiology (M.J.S., C.H.H.) and Nutrition
(M.J.S.), Harvard School of Public Health all in Boston; and the Lab
oratory for Clinical Biochemistry Research, University of Vermont, Bur
lington (M.C., R.P.T.). Address reprint requests to Dr. Ridker at the Divi
sion of Preventive Medicine, Brigham and Womens Hospital, 900
Commonwealth Ave. E., Boston, MA 022151204.
1 Vol ume 336 Numbe r 14 ? 973
 T h e New Engl a nd Journa l of Me di c i ne
20 YEARS ON THE WEB
tions10,11 are causal or are due to short-term inflam-
matory changes or to interrelations with other risk
factors, in particular smoking and hyperlipidemia.
To address these issues, we measured base-line
plasma C-reactive protein concentrations in 1086
apparently healthy men participating in the Physi-
cians Health Study13,14; myocardial infarction, stroke,
or venous thrombosis subsequently developed in 543.
We hypothesized a priori that levels of C-reactive
protein would predict the risk of myocardial infarc-
tion and stroke but not of venous thrombosis an
occlusive vascular disease generally not associated
with chronic atherosclerosis. After providing base-
line blood samples, study participants were randomly
assigned to receive aspirin or placebo. Thus, we had
the unique opportunity to evaluate directly whether
aspirin, an agent with both antiplatelet and antiin-
flammatory properties, might modify any relation
between C-reactive protein and the risk of first my-
ocardial infarction.
METHODS
Study Population and Collection of Plasma Samples
The Physicians Health Study was a randomized, double-blind,
placebo-controlled two-by-two factorial trial of aspirin and beta
carotene in the primary prevention of cardiovascular disease and
cancer. A total of 22,071 U.S. male physicians 40 to 84 years of
age in 1982, with no history of myocardial infarction, stroke,
transient ischemic attack, or cancer, were assigned to one of four
treatments: 325 mg of aspirin on alternate days (Bufferin, provid-
ed by Bristol-Myers), 50 mg of beta carotene on alternate days
(Lurotin, provided by BASF Corporation), both, or neither. The
aspirin component of the study was terminated early, on January
25, 1988, primarily because of a statistically extreme 44 percent
reduction in the risk of a first infarction in the aspirin group.13
The beta carotene component continued until the studys sched-
uled termination on December 31, 1995.14
Before randomization, between August 1982 and December
1984, potential participants were asked to provide base-line blood
samples during a 16-week run-in period during which all subjects
were given aspirin and none received placebo. Blood-collection
kits, including EDTA Vacutainer tubes, were sent to participants
with instructions for taking blood. Participants were asked to
have their blood drawn into the EDTA tubes, centrifuge the
tubes, and return the plasma (accompanied by a cold pack pro-
vided to participants) by overnight courier. The specimens were
then divided into aliquots and stored at 80C. Of the 22,071
participants in the Physicians Health Study, 14,916 (68 percent)
provided base-line plasma samples. Over the 14 years of the trial,
no specimen inadvertently thawed during storage.
Confirmation of End Points and Selection of Controls
We requested hospital records (and for fatal events, death cer-
tificates and autopsy reports) for all reported cases of myocardial
infarction, stroke, and venous thrombosis. The records were re-
viewed by a committee of physicians using standardized criteria
to confirm or refute reported events. Reviewers of end points
were unaware of treatment assignments.
Reported myocardial infarction was confirmed if its symptoms
met World Health Organization (WHO) criteria and it was asso-
ciated with either elevated plasma concentrations of enzymes or
characteristic electrocardiographic changes. Silent myocardial inf-
arctions were not included, since they could not be dated accu-
rately. Deaths due to coronary disease were confirmed on the basis
of autopsy reports, symptoms, circumstances of death, and a his-
974 Apr il 3 , 1 9 9 7
2
Back to Table of Contents
nejm.org
tory of coronary disease. Reported stroke was confirmed on the
basis of medical records showing a neurologic deficit of sudden or
rapid onset that persisted for more than 24 hours or until death.
Strokes were classified as ischemic or hemorrhagic. Computed to-
mographic scans were available for more than 95 percent of the
confirmed strokes. Reported deep venous thrombosis was con-
firmed by the documentation of a positive venography study or a
positive ultrasound study; deep venous thromboses documented
only by impedance plethysmography or Doppler examination
without ultrasound were not considered confirmed. Reported pul-
monary embolism was confirmed by a positive angiogram or a
completed ventilation-perfusion scan demonstrating at least two
segmental perfusion defects with normal ventilation.
Each participant who provided an adequate base-line plasma
sample and had a confirmed myocardial infarction, stroke, or venous
thrombosis after randomization was matched with one control.
Controls were participating physicians who provided base-line plas-
ma samples and reported no cardiovascular disease at the time the
patient reported his event. Controls were selected randomly from
among study participants who met the matching criteria of age
(1 year), smoking status (smoking currently, smoked in the
past, or never smoked), and length of time since randomization
(in 6-month intervals). Using these methods, we evaluated 543
patients and 543 controls in this prospective, nested, casecontrol
study.
Laboratory Analysis
For each patient and control, plasma collected and stored at
base line was thawed and assayed for C-reactive protein by en-
zyme-linked immunosorbent assay (ELISA) based on purified
protein and polyclonal antiC-reactive protein antibodies (Cal-
biochem).15 Antibodies were used to coat microtiter-plate wells,
and biotinylated C-reactive protein, together with the patients
plasma, was diluted 1:700 in assay buffer (phosphate-buffered sa-
line with 0.1 percent Tween 20 and 1 percent bovine serum al-
bumin). The excess was then washed off and the amount of bi-
otinylated protein estimated by the addition of avidinperoxidase
(Vectastain, Vector Laboratories). Purified C-reactive protein was
used as the standard, with protein concentrations as determined
by the manufacturer. The C-reactive protein assay was standard-
ized according to the WHO First International Reference Stan-
dard and had a sensitivity of 0.08 mg per microliter, with a stan-
dard reference range of between 0.5 and 2.5 mg per liter.
Methods used to measure plasma total and high-density lipopro-
tein (HDL) cholesterol, triglyceride, lipoprotein(a), total ho-
mocysteine, fibrinogen, D-dimer, and endogenous tissue plasmi-
nogen activator (t-PA) antigen have been described elsewhere.16-20
Blood specimens were analyzed in blinded pairs, with the po-
sition of the patients specimen varied at random within the pairs
to reduce the possibility of systematic bias and decrease interassay
variability. The mean coefficient of variation for C-reactive pro-
tein across assay runs was 4.2 percent.
Statistical Analysis
Means or proportions for base-line risk factors were calculated
for patients and controls. The significance of any difference in
means was tested by using Students t-test, and the significance
of any differences in proportions was tested by using the chi-
square statistic. Because C-reactive protein values are skewed, me-
dian concentrations were computed and the significance of any
differences in median values between patients and controls was
assessed by using Wilcoxons rank-sum test. Geometric mean con-
centrations of C-reactive protein were also computed after log
transformation that resulted in nearly normal distribution. We
used tests for trend to assess any relation of increasing C-reactive
protein values with the risk of future vascular disease after divid-
ing the sample into quartiles defined by the distribution of the
control values. We obtained adjusted estimates by using condi-
tional logistic-regression models that accounted for the matching
variables and controlled for the random treatment assignment,
 I N F L A MMAT ION, ASP IRIN, AND T HE RISK OF CA RD IOVAS C UL A R D IS EASE IN A PPA R ENTLY H EA LTH Y MEN
20 YEARS ON THE WEB nejm.org

body-mass index, diabetes, history of hypertension, and parental
history of coronary artery disease. Similar models were employed
to adjust for measured base-line plasma concentrations of total
and HDL cholesterol, triglyceride, lipoprotein(a), t-PA antigen,
fibrinogen, D-dimer, and homocysteine. To evaluate whether as-
pirin affected these relations, analyses were repeated for all cases
of myocardial infarction occurring on or before January 25, 1988
the date when randomized aspirin assignment was terminated.
All P values are two-tailed, and confidence intervals were calcu-
lated at the 95 percent level.
RESULTS
Table 1 shows the base-line characteristics of the
study participants. As expected, those in whom my-
ocardial infarction subsequently developed were more
likely than those who remained free of vascular dis-
ease to have a history of hypertension or hyperlipi-
demia or a parental history of coronary artery dis-
ease. Similarly, those in whom stroke subsequently
developed were more likely to be hypertensive. Be-
cause of the matching, patients and controls were
similar in age and history of smoking.
Geometric mean and median plasma concentra-
tions of C-reactive protein at base line were signifi-
cantly higher among those in whom any vascular
event subsequently developed than among those
who remained free of vascular disease (P0.001).
The difference between patients and controls was
greatest for those in whom myocardial infarction
subsequently developed (1.51 vs. 1.13 mg per liter,
P0.001), although differences were also significant
for stroke (P0.03), particularly ischemic stroke
(P0.02). In contrast, concentrations of C-reactive
protein were not significantly higher among those in
whom venous thrombosis subsequently developed
(P0.34) (Table 2).
The relative risk of first myocardial infarction in-
creased significantly with each increasing quartile of
base-line concentrations of C-reactive protein (P for
trend across quartiles, 0.001), in such a way that
the men in the highest quartile had a risk of future
myocardial infarction almost three times that among
those in the lowest quartile (relative risk, 2.9; 95
percent confidence interval, 1.8 to 4.6; P0.001)
(Table 3). Similarly, men with the highest base-line
C-reactive protein values had twice the risk of future
ischemic stroke (relative risk, 1.9; 95 percent confi-
dence interval, 1.1 to 3.3; P0.02). No significant
associations were observed for venous thrombosis.
The findings were similar in analyses limited to non-
fatal events.
To evaluate whether increased base-line C-reactive
protein values were associated with early rather than
late thrombosis, we stratified the analysis of myocar-
dial infarction according to the number of years of
follow-up. The relative risk of future myocardial in-
farction that was associated with the highest quartile
of C-reactive protein (as compared with the lowest
quartile) ranged from 2.4 for events occurring in the
first two years of follow-up to 3.2 for events occur-
ring six or more years into follow-up (Table 4). Sim-
ilarly, the relative risk of future myocardial infarction
that was associated with a one-quartile change in the
C-reactive protein concentration was stable over long
periods (Fig. 1).
Smokers had significantly higher median concen-
trations of C-reactive protein than nonsmokers (2.20
vs. 1.19 mg per liter, P0.001). By matching pa-
tients and controls for smoking status, we mini-
mized the potential for confounding by smoking. To
assess for effect modification, however, we repeated
the analyses, limiting the cohort to nonsmokers. As
Table 3 also shows, the relative risk of future myo-
cardial infarction among nonsmokers increased sig-

TABLE 1. BASE-LINE CHARACTERISTICS OF THE STUDY PARTICIPANTS.

CHARACTERISTIC CARDIOVASCULAR DISEASE DURING FOLLOW-UP*

MYOCARDIAL VENOUS
NONE ANY INFARCTION STROKE THROMBOSIS
(N 543) (N 543) (N 246) (N 196) (N 101)

Age (yr) 599.1 599.2 588.6 629.1 579.4

Smoking status (%)
Never smoked 44 44 45 42 50
Smoked in the past 41 41 40 40 44
Currently a smoker 15 15 15 18 6
Diabetes (%) 4 7 5 12 2
Body-mass index 252.8 263.2 263.3 253.2 262.9
History of high plasma cho- 9 13 17 10 7
lesterol (%)
History of hypertension (%) 16 29 27 35 20
Parental history of coronary artery 10 13 17 11 8
disease (%)

*Plusminus values are means SD.

The body-mass index is the weight in kilograms divided by the square of the height in meters.

Vol ume 336 Numbe r 14 975

3
Back to Table of Contents
 T h e New Engl a nd Journa l of Me di c i ne
20 YEARS ON THE WEB nejm.org

nificantly with each increasing quartile of C-reac-

TABLE 2. BASE-LINE PLASMA CONCENTRATIONS OF C-REACTIVE tive protein (P for trend, 0.001). Similarly, the
PROTEIN IN STUDY PARTICIPANTS WHO REMAINED
FREE OF VASCULAR DISEASE DURING FOLLOW-UP (CONTROLS)
long-term effects of the concentration of C-reactive
AND IN THOSE IN WHOM MYOCARDIAL INFARCTION, STROKE, protein on the risk of myocardial infarction were
OR VENOUS THROMBOSIS DEVELOPED (PATIENTS). virtually identical among nonsmokers (Table 4).
Moreover, the relation between the concentration of
CARDIOVASCULAR DISEASE C-reactive protein and myocardial infarction was not
DURING FOLLOW-UP PLASMA C-REACTIVE PROTEIN significantly altered in analyses that adjusted for
GEOMETRIC P P body-mass index; the presence or absence of diabe-
MEAN VALUE MEDIAN VALUE
tes, hypertension, or a family history of premature
mg/liter mg/liter coronary artery disease; and the plasma concentra-
None (n543) 1.10 1.13 tions of total cholesterol, HDL cholesterol, triglyc-
Any vascular event (n543) 1.37 0.001 1.40 0.001 erides, lipoprotein(a), t-PA antigen, D-dimer, fibrin-
Myocardial infarction (n246) 1.48 0.001 1.51 0.001 ogen, or homocysteine (Table 5).
Any stroke (n196) 1.30 0.03 1.36 0.03
Ischemic stroke (n154) 1.36 0.01 1.38 0.02 Finally, to assess whether the beneficial effect of
Venous thrombosis (n101) 1.24 0.22 1.26 0.34 aspirin on the risk of myocardial infarction varied ac-
cording to the base-line level of C-reactive protein,
we repeated these analyses for events occurring be-
fore January 25, 1988, the date when randomized
TABLE 3. RELATIVE RISK OF FUTURE MYOCARDIAL INFARCTION,
STROKE, AND VENOUS THROMBOSIS ACCORDING TO BASE-LINE
aspirin treatment was terminated.
PLASMA CONCENTRATIONS OF C-REACTIVE PROTEIN. The risk of future myocardial infarction increased
with each increasing quartile of C-reactive protein
values for men randomly assigned to either aspirin
QUARTILE OF C-REACTIVE PROTEIN P FOR
VASCULAR EVENT* CONCENTRATION (mg/liter) TREND or placebo, and the rates of myocardial infarction were
0.55 0.561.14 1.152.10 2.11 lower in the aspirin group for all quartiles of C-reac-
Myocardial infarction
tive protein (Fig. 2). However, the magnitude of the
(total cohort) beneficial effect of aspirin in preventing myocardial
Relative risk 1.0 1.7 2.6 2.9 0.001 infarction was directly related to base-line levels of
95% CI 1.1 2.9 1.6 4.3 1.8 4.6
P value 0.03 0.001 0.001 C-reactive protein. Specifically, randomized aspirin
Myocardial infarction assignment was associated with a large and statis-
(nonsmokers) tically significant reduction in the risk of myocar-
Relative risk 1.0 1.7 2.5 2.8 0.001
95% CI 1.0 2.8 1.5 4.1 1.7 4.7 dial infarction among men with base-line levels of
P value 0.06 0.001 0.001 C-reactive protein in the highest quartile (risk re-
Ischemic stroke
Relative risk 1.0 1.7 1.9 1.9 0.03
duction, 55.7 percent; P0.02). Among those with
95% CI 0.9 2.9 1.1 3.2 1.1 3.3 base-line levels of C-reactive protein in the lowest
P value 0.07 0.02 0.02 quartile, however, the reduction in risk associated
Venous thrombosis
Relative risk 1.0 1.1 1.2 1.3 0.38 with aspirin was far smaller and no longer statistical-
95% CI 0.6 2.0 0.7 2.3 0.7 2.4 ly significant (risk reduction, 13.9 percent; P0.77).
P value 0.78 0.51 0.42 These effects were linear across quartiles, so that the
*CI denotes confidence interval. apparent benefit of aspirin diminished in magnitude
with each decreasing quartile of inflammatory risk
(Fig. 2). This finding remained essentially unchanged
after further adjustment for other coronary risk fac-
TABLE 4. RELATIVE RISK OF FIRST MYOCARDIAL INFARCTION tors, and the interaction between assignment to the
ASSOCIATED WITH THE HIGHEST QUARTILE OF BASE-LINE aspirin group and base-line levels of C-reactive pro-
PLASMA C-REACTIVE PROTEIN CONCENTRATIONS
AS COMPARED WITH THE LOWEST QUARTILE, ACCORDING TO
tein (treated as a log-transformed continuous vari-
THE YEAR OF STUDY FOLLOW-UP. able) was statistically significant (P0.048).
DISCUSSION
GROUP* FOLLOW-UP (YR)
02 2 4 4 6 6 These prospective data indicate that the base-line
Total cohort
plasma concentration of C-reactive protein in appar-
Relative risk 2.4 2.9 2.8 3.2 ently healthy men can predict the risk of first myocar-
95% CI 0.9 6.8 1.1 7.6 1.1 6.9 1.2 8.5 dial infarction and ischemic stroke. In addition, the
P value 0.09 0.03 0.03 0.02
Nonsmokers
risk of arterial thrombosis associated with the level of
Relative risk 2.8 2.9 2.7 2.9 C-reactive protein was stable over long periods and
95% CI 0.9 8.7 1.0 8.3 1.0 7.0 1.1 8.2 was not modified by other factors, including smoking
P value 0.07 0.05 0.05 0.04
status, body-mass index, blood pressure, or the plas-
*CI denotes confidence interval. ma concentration of total or HDL cholesterol, trig-

976 Ap r il 3 , 1 9 9 7

4
Back to Table of Contents
 I N F L A M MAT ION, ASP IRIN, AND T HE RISK OF CA RD IOVAS C UL A R D IS EASE IN A PPA R ENTLY H EA LTH Y MEN
20 YEARS ON THE WEB nejm.org

lyceride, lipoprotein(a), t-PA antigen, D-dimer, fibrin-

Plasma C-reactive protein level)

ogen, or homocysteine. In contrast, the benefits of

Relative Risk of Myocardial

Infarction (per quartile of
aspirin in reducing the risk of a first myocardial inf- 2
arction diminished significantly with decreasing con-
centrations of C-reactive protein an intriguing
finding, since this substance has antiinflammatory as
well as antiplatelet properties. Finally, there was no
significant association for venous thromboembolism,
1
suggesting that the relation of inflammation to vas-
cular risk may be limited to the arterial circulation.
Because blood samples were collected at base line,
we can exclude the possibility that acute ischemia af-
fected levels of C-reactive protein. Furthermore, the
0
statistically significant associations observed were 0 2 24 46 6
present among nonsmokers, indicating that the ef-
fect of C-reactive protein on vascular risk is not sim- Years of Study Follow-up
ply the result of cigarette smoking.11,12 Thus, our
Figure 1. Relative Risk (and 95 Percent Confidence Intervals) of
prospective data relating base-line levels of C-reac- a First Myocardial Infarction Associated with Each Increasing
tive protein to future risks of myocardial infarction Quartile of Base-Line C-Reactive Protein Values, According to
and stroke among apparently healthy men greatly the Year of Study Follow-up.

TABLE 5. RELATIVE RISK OF FUTURE MYOCARDIAL INFARCTION, ACCORDING TO

BASE-LINE PLASMA CONCENTRATIONS OF C-REACTIVE PROTEIN, ADJUSTED FOR LIPID
AND NONLIPID VARIABLES.*

QUARTILE OF C-REACTIVE PROTEIN P FOR

VARIABLES ADJUSTED FOR CONCENTRATION (mg/liter) TREND
0.55 0.561.14 1.152.10 2.11

Total and HDL cholesterol

Adjusted relative risk 1.0 1.8 2.2 2.3 0.002
95% CI 1.0 3.1 1.3 3.7 1.4 3.9
P value 0.05 0.004 0.002
Triglycerides
Adjusted relative risk 1.0 1.8 2.1 2.8 0.001
95% CI 1.0 3.2 1.2 3.7 1.6 4.9
P value 0.06 0.008 0.001
Lipoprotein(a)
Adjusted relative risk 1.0 2.0 2.5 2.5 0.001
95% CI 1.2 3.4 1.5 4.2 1.5 4.2
P value 0.01 0.001 0.001
t-PA antigen
Adjusted relative risk 1.0 1.7 1.9 2.9 0.002
95% CI 0.9 3.4 1.0 3.6 1.5 5.6
P value 0.13 0.06 0.002
Total homocysteine
Adjusted relative risk 1.0 1.8 2.9 3.6 0.001
95% CI 1.1 3.1 1.7 4.8 2.1 5.9
P value 0.02 0.001 0.001
D-Dimer
Adjusted relative risk 1.0 2.2 2.4 2.7 0.001
95% CI 1.2 4.1 1.3 4.2 1.5 4.7
P value 0.007 0.003 0.001
Fibrinogen
Adjusted relative risk 1.0 2.2 2.2 2.9 0.01
95% CI 1.1 4.7 1.0 4.4 1.4 5.9
P value 0.04 0.04 0.005
Body-mass index, diabetes, history
of hypertension, and family
history of coronary artery disease
Adjusted relative risk 1.0 1.5 2.4 2.6 0.001
95% CI 0.9 2.5 1.5 4.0 1.6 4.4
P value 0.14 0.001 0.001

*All models were further adjusted for random assignment of patients to receive aspirin and beta
carotene. CI denotes confidence interval.

Vol ume 336 Numbe r 14 977

5
Back to Table of Contents
 T h e New Engl a nd Journa l of Me di c i ne
20 YEARS ON THE WEB
4 of a first myocardial infarction by 44 percent.13 The
Relative Risk of Myocardial Placebo
Aspirin
3
Infarction
2.59
2.07
2 1.80
1.37
1.16
1.00
1 derlying inflammation, these data also suggest the
0 ternatively, patients with large inflammatory burdens
1 2
Quartile of Plasma C-Reactive Protein
Figure 2. Relative Risk of a First Myocardial Infarction Associ-
ated with Base-Line Plasma Concentrations of C-Reactive Pro-
tein, Stratified According to Randomized Assignment to Aspi-
rin or Placebo Therapy.
Analyses are limited to events occurring before the unblinding
of the aspirin component of the Physicians Health Study. The
reduction in the risk of myocardial infarction associated with
the use of aspirin was 13.9 percent in the first (lowest) quartile
of C-reactive protein values, 33.4 percent in the second quar-
tile, 46.3 percent in the third quartile, and 55.7 percent in the
fourth (highest) quartile.
extend previous observations from studies of acutely
ill patients,9 patients with symptomatic coronary dis-
ease,10 or those at high risk partly because of ciga-
rette smoking.11 Moreover, in these data, the effects
of C-reactive protein were independent of a large
number of lipid-related and nonlipid-related risk
factors.
The mechanism that relates the level of C-reactive
protein to atherothrombosis is unclear. Previous in-
fection with Chlamydia pneumoniae, Helicobacter py-
lori, herpes simplex virus, or cytomegalovirus may be
a source of the chronic inflammation detected by
C-reactive protein.21-27 It is also possible that C-reactive
protein is a surrogate for interleukin-6,28 a cellular cy-
tokine associated with the recruitment of macrophages
and monocytes into atherosclerotic plaques.29 In ad-
dition, C-reactive protein can induce monocytes to
express tissue factor, a membrane glycoprotein im-
portant in initiating coagulation.30 Finally, it had been
hypothesized that bronchial inflammation due to
smoking was responsible for associations seen in pre-
vious studies relating C-reactive protein to vascular
risk.11 In this regard, our observation that the effect
of C-reactive protein is present among nonsmok-
ers makes bronchial inflammation a less likely mech-
anism. Furthermore, the finding that the effects are
stable over long periods suggests that short-term ef-
fects on clotting are unlikely.
Our data regarding the interrelation of C-reactive
protein and aspirin merit careful consideration. In
978 Apr il 3 , 1 9 9 7
nejm.org
4.16 the Physicians Health Study, aspirin reduced the risk
present findings indicate that the effect of aspirin in
preventing a first myocardial infarction was greatest
among the men with the highest base-line C-reac-
tive protein concentrations and that the benefit di-
minished significantly with decreasing concentrations
of this inflammatory marker. Thus, although the an-
1.39
tiplatelet effects of aspirin may be modified by un-
possibility that the benefit of aspirin may have been
due, at least in part, to antiinflammatory effects.31 Al-
3 4 may have a distinct vascular mechanism leading to
thrombosis that is affected differently by aspirin ther-
apy. For example, the protective effect of aspirin may
differ in the setting of plaque rupture as compared
with focal endothelial erosion.32,33
The potential limitations of these data also merit
consideration. First, our analyses are based on a sin-
gle base-line determination that may not accurately
reflect inflammatory status over long periods. Fur-
thermore, although coefficients of variation were low,
misclassification due to laboratory error cannot be
ruled out. It is important to note, however, that nei-
ther of these sources of variability can account for
the observed associations, since any random mis-
classification would bias results toward the null hy-
pothesis. Since our study was limited to measures of
C-reactive protein, other prospective studies evaluat-
ing specific cytokines, cellular adhesion molecules, and
chronic infectious agents will be required to further
elucidate the role of inflammation in the initiation
and progression of atherosclerosis.
We draw four main conclusions from these data.
First, among apparently healthy men, the base-line
level of inflammation as assessed by the plasma con-
centration of C-reactive protein predicts the risk of
a first myocardial infarction and ischemic stroke, in-
dependently of other risk factors. Second, the base-
line concentration of C-reactive protein is not asso-
ciated with the risk of venous thrombosis, a vascular
event generally not associated with atherosclerosis.
Third, C-reactive protein is not simply a short-term
marker of risk, as has previously been demonstrated
in patients with unstable angina,9 but is also a long-
term marker of risk, even for events occurring six
or more years later. This observation suggests that
the effects of inflammation are probably mediated
through a chronic process and excludes the possi-
bility that undetected acute illness at base line is re-
sponsible for the observed effects. Finally, the ben-
efits of aspirin appear to be modified by underlying
inflammation an observation that raises the pos-
sibility of antiinflammatory as well as antiplatelet
effects of this agent. The latter observation also
suggests the possibility that other antiinflammatory
agents may have a role in preventing cardiovascular
6
Back to Table of Contents
 I N F L A MMAT ION, ASP IRIN, AND T HE RISK OF CA RD IOVAS C UL A R D IS EASE IN A PPA R ENTLY H EA LTH Y ME N
20 YEARS ON THE WEB
disease. Moreover, these data suggest that inflamma-
tory markers such as C-reactive protein may provide
a method of identifying people for whom aspirin is
likely to be more or less effective a hypothesis re-
quiring direct testing in randomized trials.
Supported by grants (HL-26490, HL-34595, HL-46696, CA-34944,
CA-42182, and CA-40360) from the National Institutes of Health. Dr. Rid-
ker is supported by a Clinician Scientist Award from the American Heart As-
sociation.
REFERENCES
1. Fuster V, Badimon L, Badimon JJ, Chesebro JH. The pathogenesis of
coronary artery disease and the acute coronary syndromes. N Engl J Med
1992;326:242-50.
2. Munro JM, Cotran RS. The pathogenesis of atherosclerosis: atherogen-
esis and inflammation. Lab Invest 1988;58:249-61.
3. Ross R. The pathogenesis of atherosclerosis: a perspective for the 1990s.
Nature 1993;362:801-9.
4. Alexander RW. Inflammation and coronary artery disease. N Engl J
Med 1994;331:468-9.
5. Nieminen MS, Mattila K, Valtonen V. Infection and inflammation as
risk factors for myocardial infarction. Eur Heart J 1993;14:Suppl K:12-6.
6. Berk BC, Weintraub WS, Alexander RW. Elevation of C-reactive protein
in active coronary artery disease. Am J Cardiol 1990;65:168-72.
7. de Beer FC, Hind CR, Fox KM, Allan RM, Maseri A, Pepys MB. Mea-
surement of serum C-reactive protein concentration in myocardial ischemia
and infarction. Br Heart J 1982;47:239-43.
8. Pietila K, Harmoinen A, Hermens W, Simoons ML, Van de Werf F, Ver-
straete M. Serum C-reactive protein and infarct size in myocardial infarct
patients with a closed versus an open infarct-related coronary artery after
thrombolytic therapy. Eur Heart J 1993;14:915-9.
9. Liuzzo G, Biasucci LM, Gallimore JR, et al. The prognostic value of
C-reactive protein and serum amyloid A protein in severe unstable angina.
N Engl J Med 1994;331:417-24.
10. Thompson SG, Kienast J, Pyke SDM, Haverkate F, van de Loo JCW.
Hemostatic factors and the risk of myocardial infarction or sudden death
in patients with angina pectoris. N Engl J Med 1995;332:635-41.
11. Kuller LH, Tracy RP, Shaten J, Meilahn EN. Relation of C-reactive
protein and coronary heart disease in the MRFIT nested case-control
study. Am J Epidemiol 1996;144:537-47.
12. Das I. Raised C-reactive protein levels in serum from smokers. Clin
Chim Acta 1985;153:9-13.
13. Steering Committee of the Physicians Health Study Research Group.
Final report on the aspirin component of the ongoing Physicians Health
Study. N Engl J Med 1989;321:129-35.
14. Hennekens CH, Buring JE, Manson JE, et al. Lack of effect of long-
term supplementation with beta carotene on the incidence of malignant
neoplasms and cardiovascular disease. N Engl J Med 1996;334:1145-9.
7
Back to Table of Contents
nejm.org
15. Macy EM, Hayes TE, Tracy RP. Variability in the measurement of
C-reactive protein in healthy adults: implications for reference interval and
epidemiologic methods. Clin Chem 1997;43:52-8.
16. Stampfer MJ, Sacks FM, Salvini S, Willett WC, Hennekens CH. A pro-
spective study of cholesterol, apolipoproteins, and the risk of myocardial
infarction. N Engl J Med 1991;325:373-81.
17. Ridker PM, Hennekens CH, Stampfer MJ. A prospective study of li-
poprotein(a) and the risk of myocardial infarction. JAMA 1993;270:2195-
9.
18. Ridker PM, Hennekens CH, Selhub J, Miletich JP, Malinow MR,
Stampfer MJ. Interrelation of hyperhomocyst(e)inemia, factor V Leiden,
and risks of future venous thromboembolism. Circulation (in press).
19. Ridker PM, Vaughan DE, Stampfer MJ, Manson JE, Hennekens CH.
Endogenous tissue-type plasminogen activator and risk of myocardial inf-
arction. Lancet 1993;341:1165-8.
20. Ridker PM, Hennekens CH, Cerskus A, Stampfer MJ. Plasma concen-
tration of cross-linked fibrin degradation product (D-dimer) and the risk
of future myocardial infarction among apparently healthy men. Circulation
1994;90:2236-40.
21. Buja LM. Does atherosclerosis have an infectious etiology? Circulation
1996;94:872-3.
22. Grayston JT. Chlamydia in atherosclerosis. Circulation 1993;87:1408-
9.
23. Saikku P, Leinonen M, Tenkanen L, et al. Chronic Chlamydia pneu-
moniae infection as a risk factor for coronary heart disease in the Helsinki
Heart Study. Ann Intern Med 1992;116:273-8.
24. Thom DH, Grayston JT, Siscovick DS, Wang S-P, Weiss NS, Daling
JR. Association of prior infection with Chlamydia pneumoniae and angio-
graphically demonstrated coronary artery disease. JAMA 1992;268:68-72.
25. Melnick JL, Adam E, DeBakey ME. Possible role of cytomegalovirus
in atherogenesis. JAMA 1990;263:2204-7.
26. Mendall MA, Goggin PM, Molineaux N, et al. Relation of Helico-
bacter pylori infection and coronary heart disease. Br Heart J 1994;71:437-
9.
27. Patel P, Mendall MA, Carrington D, et al. Association of Helicobacter
pylori and Chlamydia pneumoniae infections with coronary heart disease
and cardiovascular risk factors. BMJ 1995;311:711-4. [Erratum, BMJ
1995;311:985.]
28. Bataille R, Klein B. C-reactive protein levels as a direct indicator of in-
terleukin-6 levels in humans in vivo. Arthritis Rheum 1992;35:982-4.
29. Biasucci LM, Vitelli A, Liuzzo G, et al. Elevated levels of interleukin-
6 in unstable angina. Circulation 1996;94:874-7.
30. Cermak J, Key NS, Bach RR, Balla J, Jacob HS, Vercellotti GM. C-re-
active protein induces human peripheral blood monocytes to synthesize tis-
sue factor. Blood 1993;82:513-20.
31. Vane J. The evolution of non-steroidal anti-inflammatory drugs and
their mechanisms of action. Drugs 1987;33:Suppl 1:18-27.
32. van der Wal AC, Becker AE, van der Loos CM, Das PK. Site of intimal
rupture or erosion of thrombosed coronary atherosclerotic plaques is char-
acterized by an inflammatory process irrespective of the dominant plaque
morphology. Circulation 1994;89:36-44.
33. Farb A, Burke AP, Tang AL, et al. Coronary plaque erosion without
rupture into a lipid core: a frequent cause of coronary thrombosis in sud-
den coronary death. Circulation 1996;93:1354-63.
Vol ume 336 Numbe r 14  979
20 YEARS ON THE WEB nejm.org

The New England

Journal of Medicine
Co pyr igh t , 1 998, by t he Mas s achus et ts Medic al So ciet y

VOLUME 338 M A R C H 26, 1998 NUMBER 13

DECLINING MORBIDITY AND MORTALITY AMONG PATIENTS WITH ADVANCED

HUMAN IMMUNODEFICIENCY VIRUS INFECTION

FRANK J. PALELLA, JR., M.D., KATHLEEN M. DELANEY, M.S., ANNE C. MOORMAN, B.S.N., M.P.H.,
MARK O. LOVELESS, M.D., JACK FUHRER, M.D., GLEN A. SATTEN, PH.D., DIANE J. ASCHMAN, R.PH., M.S.,
SCOTT D. HOLMBERG, M.D., M.P.H., AND THE HIV OUTPATIENT STUDY INVESTIGATORS*

ABSTRACT
Background and Methods National surveillance
data show recent, marked reductions in morbidity
and mortality associated with the acquired immuno-
deficiency syndrome (AIDS). To evaluate these de-
clines, we analyzed data on 1255 patients, each of
whom had at least one CD4 count below 100 cells
per cubic millimeter, who were seen at nine clinics
specializing in the treatment of human immunodefi-
ciency virus (HIV) infection in eight U.S. cities from
January 1994 through June 1997.
Results Mortality among the patients declined
from 29.4 per 100 person-years in 1995 to 8.8 per 100
person-years in the second quarter of 1997. There
were reductions in mortality regardless of sex, race,
age, and risk factors for transmission of HIV. The in-
cidence of any of three major opportunistic infec-
tions (Pneumocystis carinii pneumonia, Mycobacte-
rium avium complex disease, and cytomegalovirus
retinitis) declined from 21.9 per 100 person-years in
1994 to 3.7 per 100 person-years by mid-1997. In a
failure-rate model, increases in the intensity of anti-
retroviral therapy (classified as none, monotherapy,
combination therapy without a protease inhibitor,
and combination therapy with a protease inhibitor)
were associated with stepwise reductions in morbid-
ity and mortality. Combination antiretroviral therapy
was associated with the most benefit; the inclusion
of protease inhibitors in such regimens conferred
additional benefit. Patients with private insurance
were more often prescribed protease inhibitors and
had lower mortality rates than those insured by Medi-
care or Medicaid.
Conclusions The recent declines in morbidity and
mortality due to AIDS are attributable to the use of
more intensive antiretroviral therapies. (N Engl J Med
1998;338:853-60.)
1998, Massachusetts Medical Society.
T
HE treatment of human immunodeficien-
cy virus (HIV) infection has undergone
considerable change.1-3 Protease inhibitors
and nonnucleoside-analogue reverse-trans-
criptase inhibitors, when used as part of combina-
tion drug regimens, can profoundly suppress viral
replication, with consequent repletion of CD4 cell
counts.4-7 Multiple clinical trials have shown the
virologic and immunologic efficacy of the newer,
highly active antiretroviral-drug combinations7,8 by
measuring the plasma load of HIV RNA and CD4
cell counts.9-16 In addition, prophylactic medications
are now being used routinely to prevent disseminat-
ed Mycobacterium avium complex infection.
Several reports have described reductions in mor-
tality and in the rate of hospitalization of HIV-
infected patients; however, such reductions have not
been clearly related to specific therapeutic regi-
mens.17-21 We analyzed data collected over 42 months
in the HIV Outpatient Study. During this period,
rates of chemoprophylaxis against opportunistic in-
fection remained relatively constant even while pat-
terns of antiretroviral therapy were changing. This
report outlines the changes in death rates and the
incidence of opportunistic infections in a large group
of HIV-infected outpatients, many of whom had
previously received extensive treatment.
From Northwestern University Medical School, Chicago (F.J.P.); the
Health Research Network of Apache Medical Systems, Chicago (K.M.D.,
D.J.A.); the Centers for Disease Control and Prevention, Atlanta (A.C.M.,
G.A.S., S.D.H.); Oregon Health Sciences University, Portland (M.O.L.);
and the State University of New York, Stony Brook (J.F.). Address reprint
requests to Dr. Palella at Northwestern University Medical School, 303
E. Superior St., Passavant Pavilion, Rm. 828, Chicago, IL 60611-0949.
*The investigators participating in the HIV Outpatient Study are listed
in the Appendix.

8 Vol ume 338 Numbe r 13  853

Back to Table of Contents
 T h e New Engl a nd Journa l of Me di c i ne
20 YEARS ON THE WEB
METHODS
The HIV Outpatient Study
The ongoing HIV Outpatient Study, into which patients are
continuously recruited, collects summaries of physicianpatient
interactions and data on the course of disease for more than 3500
HIV-infected, nonhospitalized patients who have been seen at a
total of approximately 47,000 outpatient visits since 1992. The
present analysis includes data on patients seen from January 1994
through June 1997. The study sites are nine clinics (seven private
and two public) in eight U.S. cities (Portland, Oreg.; Tampa, Fla.;
Oakland, Calif.; Washington, D.C.; Chicago; Stony Brook, N.Y.;
Atlanta; and Denver) that provide care for at least 150 HIV-
infected patients each per year. The participating physicians rou-
tinely care for hundreds of HIV-infected patients and thus have
extensive experience with HIV.22
Information in five general categories has been abstracted from
the chart for each outpatient visit and entered electronically by
trained data abstracters; the data are compiled centrally, reviewed,
and corrected before being included in the data base. Because the
study physicians are the source of primary care for these patients,
all symptoms, diagnoses, and treatments since the previous visit,
including interim changes in treatment in any setting, are noted
at each clinic visit. The categories of information abstracted are
as follows: demographic characteristics and risk factors for HIV
infection; symptoms; diagnosed diseases (both definitive and pre-
sumptive diagnoses); medications prescribed, including the dose
and duration; and laboratory values, including CD4 cell counts
and measurements of plasma HIV RNA.
Patients
Twelve hundred fifty-five study participants who had ever had
a CD4 cell count below 100 per cubic millimeter were the fo-
cus of this analysis, since our goal was to evaluate morbidity and
mortality among the persons at greatest risk for illness or death.
For 71 percent of the patients, the first CD4 cell count of less
than 100 per cubic millimeter was noted within six months before
enrollment or thereafter. For the remaining 29 percent, a CD4 ment effects for each quarter with a failure-rate model in which
cell count below 100 per cubic millimeter was documented more
than six months before study entry; for these patients, the date
of the initial study visit marked the beginning of follow-up. Pa-
tients whose CD4 cell counts later rebounded to 100 per cubic
millimeter or higher remained in the analysis.
Data from clinic visits were used to calculate the number of
days of observation per quarter for each patient in each of four
categories of prescribed antiretroviral therapy. These categories,
in increasing order of intensity, were no antiretroviral therapy,
monotherapy, combination therapy without a protease inhibitor,
and combination therapy that included a protease inhibitor. To
ensure that mortality rates were based on patients who received
therapy for an adequate time, patients were not included in cal-
culations of follow-up time or events for the first 30 days after any
change in therapy. Only deaths that occurred within 90 days after
a clinic visit were counted. Patients not seen since March 1997
contributed person-years at risk for a maximum of 90 days after
their last study visit. Because enrollment continued during the
study period, data from some new patients were included in each
quarterly analysis.
Analysis of Outcomes
Data were analyzed with SAS software (version 6.11; SAS Insti-
tute, Cary, N.C.). Morbidity (i.e., opportunistic infections) and
mortality were compared for the antiretroviral-therapy categories,
adjusted for chemoprophylaxis against opportunistic infection
and demographic factors (sex, age, race or ethnic group, and risk
factors for HIV infection [men who have sex with men, injection-
drug use, heterosexual sex, and other]), CD4 count at the first
study visit, and method of payment (i.e., insurance status). After
stratifying the data according to antiretroviral-therapy category
854  Ma r ch 2 6 , 1 9 9 8
9
Back to Table of Contents
nejm.org
and calendar quarter, we fitted Poisson failure-rate models to the
data, using the SAS Lifereg procedure. The models assume the
hazard is constant throughout each quarter but allow it to vary
between quarters. Rates of mortality and morbidity calculated in
this way are equivalent to tabulations of the number of deaths (or
events) that occurred in each quarter divided by the number of
person-years of observation during that quarter. Death rates per
100 person-years were calculated by totaling the number of per-
son-days of observation in a specified period. Deaths among ob-
served patients were counted, and observation time was standard-
ized to 100 person-years.
Acquired immunodeficiency syndrome (AIDS)defining op-
portunistic infections were analyzed in the aggregate; in addition,
separate analyses were performed for Pneumocystis carinii pneu-
monia, M. avium complex infection, and cytomegalovirus reti-
nitis. Patients with a diagnosis of cytomegalovirus retinitis or
M. avium complex disease before study entry or during the first
30 days of follow-up and patients with active P. carinii pneumo-
nia at the beginning of follow-up were excluded from the analyses
of the incidence of that opportunistic infection.
Statistical Analysis
The possible effect of demographic variables on changes in
rates of morbidity and mortality was examined. The distributions
of patients according to age, race or ethnic group, sex, and risk
factors for HIV infection were examined in each of the 42
months of observation to determine whether the study popula-
tion had changed over the study period. These demographic var-
iables, the study center, the first recorded CD4 cell count (cat-
egorized as 0 to 49, 50 to 99, or 100 per cubic millimeter), and
whether the patient received chemoprophylaxis against M. avium
complex and P. carinii were included with the antiretroviral-ther-
apy category as main effects in the failure-rate model of morbidity
and mortality in order to determine whether temporal trends
were the result of shifts in the composition of the study popula-
tion or shifts in the pattern of prophylaxis. Finally, the consistency
of the effect of treatment on mortality and morbidity over time
was tested by comparing a failure-rate model with separate treat-
treatment effects were assumed to be constant over time. Statis-
tically nonsignificant effects of the calendar quarter were excluded
from the final model, resulting in a Poisson model that assumed
a constant effect of treatment over the 42-month observation pe-
riod. Results are reported as relative risks of death or morbidity
(defined as P. carinii pneumonia, M. avium complex infection,
or cytomegalovirus retinitis), with associated 95 percent confi-
dence intervals.
The primary source of payment for medical care was docu-
mented for each patient and categorized as private insurance (in-
cluding fee-for-service care, health maintenance organizations,
and preferred-provider organizations), Medicare, Medicaid, self-
payment, or prescription programs under the Ryan White Care
Act. Mortality and rates of prescription of protease inhibitors per
100 person-years of observation in each quarter were analyzed ac-
cording to the source of payment.
RESULTS
Demographic Characteristics
Our analyses include data on 1255 HIV-1-infect-
ed persons with at least one CD4 cell count below
100 per cubic millimeter who were among the more
than 3500 HIV-infected patients seen as part of the
HIV Outpatient Study during the period of analysis
(January 1994 through June 1997). About 80 per-
cent were 30 to 49 years of age, and the age distri-
bution did not shift during the period of analysis.
We observed nonsignificant trends toward increasing
 D ECL INING MORBIDIT Y AND MORTA LIT Y A MONG PATIENTS WITH A DVA NCED H IV INFECTION
20 YEARS ON THE WEB
numbers of blacks, Hispanics, and women (making
up 20 percent, 9 percent, and 12 percent, respec-
tively, of the total by June 1997) and decreasing
proportions of men who reported same-sex sexual
activity (accounting for 65 percent of those seen by
June 1997). The proportion of patients who report-
ed injection-drug use (about 14 percent) did not
change significantly over time.
The proportion of persons whose initial CD4 chemoprophylaxis were evaluated; the patterns of
cell count at study entry was less than 50 per cubic
millimeter decreased slightly (it was 55 percent in
1994, 51 percent in 1995, 44 percent in 1996, and
42 percent in 1997). The proportion of patients
whose most recent CD4 cell count was less than
50 per cubic millimeter diminished significantly (from
67 percent in 1994 to 57 percent in 1995, 43 per-
cent in 1996, and 29 percent in 1997).
Use of Antiretroviral Agents
During the study, the pattern of antiretroviral
therapy changed dramatically among patients with
CD4 cell counts below 100 per cubic millimeter.
The proportion of patients for whom any antiretro-
viral therapy was prescribed increased, from 72 per-
cent of patients in 1994 to 95 percent by June 1997,
with marked increases in the prescription of combi-
nation regimens (from 25 percent in 1994 to 94
percent by June 1997). The most dramatic increases
were in the rate of use of regimens containing pro-
tease inhibitors, from 2 percent in mid-1995 to 82
percent by June 1997. The use of combinations in-
corporating protease inhibitors differed little accord-
ing to patients demographic characteristics, although
the study sites varied widely in their rates of use of
protease inhibitors. In the first quarter of 1996, site-
specific rates of protease-inhibitor use ranged from
6 percent to 71 percent; by the second quarter of
1997, the rates ranged from 40 percent to 95 per-
cent. Publicly funded clinics were slower to use pro-
tease inhibitors; however, the proportional increases
in use were similar among all sites.
Mortality
Mortality declined markedly in 1996 and early
1997, after remaining constant during 1994 and
1995. Death rates decreased from 29.4 per 100 per-
son-years in 1995 to 16.7 per 100 person-years in
1996 and to 8.8 per 100 by the second quarter of
1997 (Table 1 and Fig. 1).
Patterns of reduction in death rates among men
and women, white and nonwhite persons, and per-
sons 40 or above 40 years of age were similar and
declined in the same way during the 14 quarters of
observation. Although mortality decreased propor-
tionally among injection-drug users, they had con-
sistently higher mortality rates than patients who did
not report a history of injection-drug use. Although
mortality and morbidity differed among the study
10
Back to Table of Contents
nejm.org
sites, declines in both rates were noted at all clinics.
The rate of use of prophylaxis against opportunistic
infections was consistent during the 42 months of
observation: the proportion of patients receiving
prophylaxis against M. avium complex ranged from
46 percent to 55 percent; the rate of prophylaxis
against P. carinii ranged from 92 to 94 percent.
Death rates among persons receiving these types of
decreasing mortality were similar to those in the
study group as a whole.
Death rates are shown according to antiretroviral-
therapy category in Table 1. Death rates declined in
virtually every quarter, correlating inversely with the
intensity of the antiretroviral therapy prescribed
(Fig. 1), and declined most dramatically during the
last six quarters covered by the analysis (Table 1).
Differences among the patients in sex, age, race or
ethnic group, or risk category did not explain the
observed temporal trend in mortality or morbidity
when these factors were included in the preliminary
failure-rate model. The inclusion of the use of chemo-
prophylaxis against opportunistic infections in the
model also did not explain the trend; however, both
the initial CD4 cell count and the study site were
significant (P0.01) and were therefore retained in
the model.
When the effect of treatment was included in the
failure-rate model used to evaluate mortality, the ef-
fect of the calendar quarter of observation was not
significant (P0.49). The interaction of treatment
with quarter was also not significant (P0.34).
Comparisons of mortality in different antiretroviral-
therapy categories within this model (Table 2) re-
vealed that for each increase in the intensity of anti-
retroviral therapy, there was a significant additional
benefit in terms of lower mortality. Notably, mortal-
ity among patients receiving combination regimens
that did not include protease inhibitors was 1.5
times that among patients receiving combination
regimens that included a protease inhibitor.
Mortality rates are shown according to patients
primary source of payment for medical services in
Table 3. The patients whose care was funded under
the Ryan White Care Act prescription programs and
those who paid for their own care together made up
about 10 percent of the total population and had
mortality rates similar to those for patients who were
receiving Medicare. Although mortality declined over-
all among patients covered by Medicaid and those
who were privately insured, the death rates for pa-
tients insured by Medicaid were higher than the
rates in the overall study population in all but two
quarters. In 1995, mortality among those covered
by Medicaid was 46.9 per 100 person-years; among
those with private insurance, it was 24.4 per 100
person-years (data not shown). Patients with private
insurance were consistently more likely to receive a
Vol ume 338 Numbe r 13  855
 T h e New Engl a nd Journa l of Me di c i ne
20 YEARS ON THE WEB nejm.org

TABLE 1. MORTALITY AMONG PATIENTS WITH CD4 T-LYMPHOCYTE COUNTS

OF FEWER THAN 100 PER CUBIC MILLIMETER, ACCORDING TO TYPE
OF ANTIRETROVIRAL THERAPY AND CALENDAR QUARTER, 1994 THROUGH JUNE 1997.*

NUCLEOSIDE- COMBINATION
YEAR NO NUCLEOSIDE- ANALOGUE THERAPY INCLUDING
AND ALL ANTIRETROVIRAL ANALOGUE COMBINATION A PROTEASE
QUARTER PATIENTS THERAPY MONOTHERAPY THERAPY INHIBITOR

deaths/100 person-yr (no. of deaths/no. of patients)

1994
1 35.1 (16/237) 52.7 (6/74) 37.0 (7/122) 29.3 (3/59)
2 35.2 (19/261) 59.9 (8/83) 42.9 (10/130) 8.1 (1/64)
3 23.4 (14/309) 43.3 (7/104) 25.5 (6/147) 7.6 (1/76)
4 23.1 (20/429) 38.2 (9/150) 22.2 (8/219) 19.1 (3/90)
1995
1 31.2 (34/524) 66.4 (18/178) 31.8 (15/273) 4.9 (1/115) 0 (0/10)
2 27.4 (34/581) 51.6 (14/172) 33.4 (16/296) 13.4 (4/181) 0 (0/13)
3 30.8 (41/609) 62.8 (17/183) 34.9 (16/279) 12.1 (5/222) 43.2 (1/11)
4 28.5 (40/631) 42.0 (13/179) 37.5 (16/256) 23.0 (10/249) 0 (0/35)
1996
1 29.4 (41/645) 54.8 (12/150) 55.1 (13/173) 22.7 (9/256) 10.4 (3/201)
2 15.4 (22/628) 39.3 (6/110) 16.1 (2/85) 18.6 (5/199) 7.8 (5/364)
3 11.3 (16/608) 21.7 (2/64) 28.5 (2/53) 16.2 (3/123) 9.9 (9/437)
4 10.8 (15/600) 15.3 (1/47) 0 (0/30) 0 (0/110) 14.4 (14/458)
1997
1 14.9 (20/583) 16.1 (1/46) 0 (0/18) 28.0 (5/115) 13.4 (13/462)
2 8.8 (12/574) 51.6 (3/37) 0 (0/10) 5.8 (1/92) 7.8 (8/460)

*The rates shown are deaths per 100 person-years in each quarter among patients who received
the specified type of antiviral therapy for at least 30 days. Patients could be included in no more than
two categories during a quarter. The period of analysis ran from January 1994 through June 1997.
Combination therapy including protease inhibitors was not widely available until 1995.
This category includes the few patients whose observation time did not fall into any therapy cat-
egory during the quarter because of frequent changes in therapy.

Therapy with a Protease Inhibitor

40 100
Deaths per 100 Person-Years

90
80

(% of patient-days)
30
Deaths 70
60
20 50
40
30
10
20
Use of protease inhibitors 10
0 0
1994 1995 1996 1997
Figure 1. Mortality and Frequency of Use of Combination Antiretroviral Therapy Including a Protease Inhibitor among
HIV-Infected Patients with Fewer Than 100 CD4 Cells per Cubic Millimeter, According to Calendar Quarter, from Jan-
uary 1994 through June 1997.

856  Ma rch 2 6 , 1 9 9 8

11
Back to Table of Contents
 D E CL INING MORBIDIT Y AND MORTA LIT Y A MONG PATIENTS WITH A DVA NCED H IV INFECTION
20 YEARS ON THE WEB nejm.org

years by the second quarter of 1997. Again, decreas-

TABLE 2. RELATIVE RISK OF DEATH AND MORBIDITY es in mortality correlated temporally with the in-
AMONG PATIENTS WITH CD4 T-LYMPHOCYTE COUNTS
OF FEWER THAN 100 PER CUBIC MILLIMETER, ACCORDING TO
creased use of combination antiretroviral regimens,
TYPE OF ANTIRETROVIRAL THERAPY.* especially those containing protease inhibitors.
Morbidity
ADJUSTED RELATIVE RISK P
ANTIRETROVIRAL-THERAPY CATEGORY (95% CI) VALUE For the 1255 patients we studied, the incidence
of serious opportunistic infections declined marked-
Mortality
ly in 1996 and early 1997 (Fig. 2). The incidence of
None vs. monotherapy 1.5 (1.22.0) 0.002
None vs. combination 2.9 (2.14.2) 0.001 any AIDS-defining diagnosis decreased from ap-
None vs. combination  PI 4.5 (3.26.2) 0.001 proximately 50 per 100 person-years in 1994 and
Monotherapy vs. combination 1.9 (1.42.7) 0.001 1995 to 28.6 per 100 person-years in 1996; during
Monotherapy vs. combination  PI 3.0 (2.14.1) 0.001
Combination vs. combination  PI 1.5 (1.02.2) 0.03 the last two quarters of 1996, this rate fell to 13.3
Morbidity per 100 person-years, where it remained during the
None vs. monotherapy 1.9 (1.22.8) 0.003 first two quarters of 1997. To simplify the analysis,
None vs. combination 2.4 (1.53.7) 0.001 we focused on three serious common infections:
None vs. combination  PI 4.5 (2.87.2)
Monotherapy vs. combination 1.3 (0.81.9)
0.001
0.26
P. carinii pneumonia, M. avium complex disease,
Monotherapy vs. combination  PI 2.4 (1.53.8) 0.001 and cytomegalovirus retinitis. In 1994 the incidence
Combination vs. combination  PI 1.9 (1.23.1) 0.01 of these three opportunistic infections was 21.9 per
*Morbidity was defined for this analysis as a diagnosis of Pneumocystis
100 person-years; by the second quarter of 1997, it
carinii pneumonia, Mycobacterium avium complex infection, or cytomeg- was 3.7 per 100 person-years (Fig. 2). None of the
alovirus retinitis. independent demographic variables had a significant
Antiretroviral therapy was categorized as follows: no antiretroviral therapy, effect on the incidence of infection in the failure-rate
monotherapy, combination therapy without a protease inhibitor (combina-
tion), and combination therapy with a protease inhibitor (combination  PI). model, either in the group as a whole or in the var-
For mortality, the relative risks have been adjusted for study center and ious antiretroviral-therapy categories; however, chemo-
the CD4 cell count at the first study visit (0 to 49, 50 to 99, or 100 prophylaxis against M. avium complex did have a
per cubic millimeter); for morbidity, the model included these variables as significant effect (P0.001).
well as the rate of prophylaxis against M. avium complex. CI denotes con-
fidence interval. The final model for morbidity included the study
center, category of antiretroviral therapy, initial
CD4 cell count, and the use of chemoprophylaxis
against M. avium complex. The effect of time on
protease inhibitor than were patients in any other morbidity was nonsignificant (P0.13) when the
payer group, although the use of protease inhibitors first 12 quarters of the observation period were an-
increased markedly for both privately insured pa- alyzed; apparent temporal trends were explained by
tients and those whose care was publicly funded. changes in antiretroviral-therapy categories. Howev-
The vast majority of patients in all payer groups were er, when data from the last two quarters of 1997
prescribed protease inhibitors by the second quarter were included, there appeared to be a confounding
of 1997. The difference in mortality between pa- effect between the quarter (time) and the type of
tients with private insurance and those covered by antiretroviral therapy when both were entered into
public funding narrowed in later quarters; by the the model simultaneously; as a result, the effect of
second quarter of 1997, mortality among those with time appeared to be significant and the treatment
private insurance had fallen to 7.7 per 100 person- regimen appeared less so. Comparisons in which pa-
years; for those covered by Medicaid, mortality was tients were stratified according to the antiretroviral
9.2 per 100 person-years. regimen were problematic because few patients re-
In a preliminary failure-rate model with the study mained in the no-therapy or monotherapy groups
center, CD4 cell count, and payment category as and because there were too few opportunistic events
independent covariates, mortality differed signifi- in these groups for meaningful comparisons of mor-
cantly (P0.02) according to payment category. bidity among treatment groups. Hence, the fact that
However, when the type of antiretroviral therapy was time appears to have a significant effect in the later
added to this model, the effect of the payment cat- quarters is a reflection of the dramatic redistribution
egory was not significant (P0.09), suggesting that of patients to more aggressive antiretroviral-treat-
differences in mortality among payment groups were ment categories and a consequently lower number
accounted for by different patterns of treatment. of infections. After the strong correlation between
A subgroup analysis of mortality among patients these measures had been demonstrated, the calendar
with a CD4 cell count below 50 per cubic milli- quarter was removed from the model, and as expect-
meter mirrored previous findings; there was a de- ed, the observed benefit was linked to antiretroviral
cline in mortality from 39.1 per 100 person-years in therapy.
the first quarter of 1994 to 10.7 per 100 person- The most marked reductions in the overall inci-

Vol ume 338 Numbe r 13  857

12
Back to Table of Contents
 T h e New Engl a nd Journa l of Me di c i ne
20 YEARS ON THE WEB nejm.org

TABLE 3. MORTALITY AND RATES OF PRESCRIPTION OF PROTEASE INHIBITORS AMONG PATIENTS WITH CD4
T-LYMPHOCYTE COUNTS OF FEWER THAN 100 PER CUBIC MILLIMETER, ACCORDING TO CALENDAR QUARTER,
1994 THROUGH JUNE 1997.*

YEAR AND
QUARTER ALL PATIENTS PRIVATE INSURANCE MEDICARE MEDICAID
PROTEASE PROTEASE PROTEASE PROTEASE
INHIBITORS MORTALITY INHIBITORS MORTALITY INHIBITORS MORTALITY INHIBITORS MORTALITY

deaths/100 deaths/100 deaths/100 deaths/100

% person-yr % person-yr % person-yr % person-yr

1994
1 35.1 (16/237) 30.2 (8/129) 51.7 (2/20) 39.7 (5/70)
2 35.2 (19/261) 42.2 (13/147) 0 (0/16) 27.5 (4/70)
3 23.4 (14/309) 14.8 (5/166) 26.1 (1/19) 49.3 (8/90)
4 23.1 (20/429) 27.4 (13/236) 0 (0/32) 22.2 (5/110)
1995
1 2.1 31.2 (34/524) 1.4 25.0 (15/279) 0 46.8 (4/41) 1.4 47.2 (13/142)
2 2.4 27.4 (34/581) 2.0 25.6 (17/307) 0 9.8 (1/47) 0.6 38.6 (13/160)
3 4.4 30.8 (41/609) 4.4 23.5 (17/317) 0 34.4 (4/60) 3.1 49.4 (17/161)
4 17.6 28.5 (40/631) 24.8 23.7 (18/335) 9.5 14.4 (2/63) 6.5 44.7 (15/154)
1996
1 41.6 29.4 (41/645) 49.6 23.5 (18/345) 38.1 60.4 (8/53) 31.6 38.7 (13/158)
2 64.2 15.4 (22/628) 74.8 12.9 (10/330) 53.0 42.5 (6/66) 56.7 14.4 (5/157)
3 75.5 11.3 (16/608) 85.6 7.6 (6/326) 73.0 6.6 (1/63) 67.5 23.7 (8/154)
4 79.3 10.8 (15/600) 88.6 10.1 (8/336) 79.7 14.3 (2/59) 70.0 11.6 (4/150)
1997
1 81.8 14.9 (20/583) 68.6 10.2 (8/334) 78.3 45.0 (6/60) 71.2 15.8 (5/139)
2 83.6 8.8 (12/574) 89.4 7.7 (6/329) 86.0 21.9 (3/57) 71.7 9.2 (3/138)

*The mortality rates shown are deaths per 100 person-years in each quarter, with the numbers of deaths and numbers of patients in pa-
rentheses. Percentages for protease inhibitors are the percentages of patients who were ever prescribed a protease inhibitor during the quarter.
Dashes indicate that protease inhibitors were not yet available.
This category includes patients for whom the primary payer was not private insurance, Medicare, or Medicaid (e.g., self-payment, Ryan
White Care Act, or other programs).
This category includes fee-for-service care, private health maintenance organizations, preferred-provider organizations, and similar pro-
grams.

dence of opportunistic infections occurred during
the last five quarters of analysis and paralleled in-
creases in the frequency of use of protease inhibitors.
The number of patients receiving protease inhibitors
tripled from the first to the fourth quarter of 1996,
and 84 percent of all patients with fewer than 100
CD4 cells per cubic millimeter received protease
inhibitors by the second quarter of 1997.
Comparisons of the incidence of any one of the
three major opportunistic infections among the an-
tiretroviral-therapy categories in the failure-rate mod-
el produced findings consistent with the data on
mortality (Table 2); with increases in the intensity of
antiretroviral regimens, stepwise reductions in mor-
bidity were noted.
The routine use of measurements of viral load in
the participating clinics increased during the period
of analysis; by June 1997, at least one such determi-
nation had been recorded for 85 percent of patients.
Viral-load measurements, which were calculated as
the mean of the median values for each patient with-
in each antiretroviral-therapy category, were inverse-
ly related to the intensity of therapy. The group
mean for those receiving no therapy, expressed as
the log of the number of copies of HIV RNA per
milliliter of blood, was 4.10; the corresponding val-
ues were 3.66 for those receiving monotherapy, 3.43
for those receiving combination antiretroviral thera-
py without a protease inhibitor, and 2.97 for those
receiving combination regimens that included a pro-
tease inhibitor.
DISCUSSION
The data from the HIV Outpatient Study show
a dramatic reduction in morbidity and mortality
among patients with CD4 cell counts under 100
per cubic millimeter. Reductions in death and dis-
ease were clearly linked to the increasing use of com-
bination antiretroviral therapy, with the most dra-
matic reductions coinciding with increases in the use
of protease inhibitors. In this analysis, in which we
adjusted for the severity of immune compromise,
the reductions in morbidity and mortality were seen
regardless of sex, race or ethnic group, and risk fac-
tor for the transmission of HIV.
The data reflect actual patient care and outcomes
in a setting in which the effects of diverse factors,
such as access to care and physicians prescribing pref-
erences, were documented.23 January 1994 through
June 1997 is a period in which use of combinations

858  Mar ch 26 , 1 9 9 8

13
Back to Table of Contents
 D ECL INING MORBIDIT Y AND MORTA LIT Y A MONG PATIENTS WITH A DVA NCED H IV INFECTION
20 YEARS ON THE WEB
20
No. of Opportunistic Infections
Cytomegalovirus
per 100 Person-Years
15
10
5
P. carinii pneumonia
0
1994
Figure 2. Rates of Cytomegalovirus Infection, Pneumocystis carinii Pneumonia, and Mycobacterium
avium Complex Disease among HIV-Infected Patients with Fewer Than 100 CD4 Cells per Cubic Mil-
limeter, According to Calendar Quarter, from January 1994 through June 1997.
of antiretroviral agents that included protease inhib-
itors increased,24 while the rate of use of routine
prophylaxis to prevent serious opportunistic disease
remained constant.
Several types of bias might have affected the re-
sults of this study of over 1200 patients. The pa-
tients and the physicians may not have been repre-
sentative of the typical patient with HIV or the
typical clinician, although the geographic sites of
the clinics and the demographic characteristics of
the patients were diverse and roughly representative
of the HIV-infected population receiving medical
care in the United States. There were some differ-
ences in demographic factors between patients for
whom combination therapy was prescribed and
those for whom it was not prescribed (this was es-
pecially true for combination regimens that included
protease inhibitors), but demographic factors were
found to have no significant effect on morbidity or
mortality. The part of the observation period during
which the use of protease inhibitors was common
was fairly brief, and our analysis cannot address the
effects of antiretroviral-therapy combinations that in-
cluded nonnucleoside-analogue reverse-transcriptase
inhibitors.25
When we analyzed the reductions in morbidity
and mortality according to patients antiretroviral-
therapy category, it became clear that benefit was di-
rectly linked to the intensity of treatment. Antiretro-
viral-drug combinations were more beneficial than
monotherapy, and combination regimens (usually
three-drug combinations) that included protease in-
hibitors were of greater benefit than combination
regimens without these drugs. These findings are
consistent with the reported virologic and immuno-
logic benefits of protease inhibitors in previously re-
ported data.7,16
We found reductions in opportunistic infections
Back to Table of Contents
nejm.org
M. avium complex
1995 1996 1997
overall and in the proportions of patients who had
P. carinii pneumonia, M. avium complex disease,
and cytomegalovirus retinitis. The most marked re-
duction occurred at a time when the use of protease
inhibitors became widespread. The rates of use of
prophylaxis against P. carinii and M. avium complex
remained essentially constant throughout the period
of analysis, confirming the role of antiretroviral ther-
apy as the principal factor in the observed reductions
in morbidity and mortality. The results of a sub-
group analysis of patients with CD4 cell counts
under 50 per cubic millimeter were consistent with
these findings.
Patients with private insurance were more likely to
be prescribed a protease inhibitor than were those
covered by Medicare or Medicaid, perhaps reflecting
a lag in the availability of protease inhibitors in clin-
ics treating patients whose care was covered under
public programs. By June 1997, when protease in-
hibitors had been available for well over a year, there
remained a marked discrepancy in the rates of pre-
scription of protease inhibitors between patients
with private insurance and those covered by public
insurance. Despite this disparity, the rates of use of
protease inhibitors did not differ significantly when
patients were grouped according to sex, race or eth-
nic group, or age. Injection-drug users were less
likely than other patients to receive protease inhi-
bitors, but injection-drug use was not significantly
associated with morbidity (P0.70) or mortality
(P0.87) in the models. Death rates were lower for
those with private insurance than for the study pop-
ulation overall and for those in other payer catego-
ries; this effect is attributable to differences in the
rate of prescription of protease inhibitors.
Our analysis describes the rate and extent of the
adoption of new antiretroviral therapies outside the
setting of controlled clinical trials, in a group of pa-
Vol ume 338 Numbe r 13  859
14
 T h e New Engl a nd Journa l of Me di c i ne
20 YEARS ON THE WEB
tients that is reasonably representative of patients
with HIV in the United States. During the observa-
tion period, the increased routine use of quantitative
plasma measurements of HIV RNA may have affect-
ed the extent to which new therapies were adopted
by providing a timely gauge of the efficacy of treat-
ment. Stratification of patients according to meas-
urements of viral load indicated that this variable
had an inverse relation with the intensity of antiret-
roviral treatment; this finding is consistent with data
from earlier reports.
In conclusion, the routine use of increasingly in-
tensive antiretroviral therapies has resulted directly
in dramatic declines in morbidity and mortality
among HIV-infected patients with advanced immune
depletion. These declines occurred during an era in
which antiretroviral therapies became more numer-
ous and more potent. As more patients receive more
effective antiretroviral-drug combinations, analyses
to determine the optimal duration and timing of
therapy26 will become increasingly necessary. Our
data suggest that an intensive combination drug-
therapy regimen that includes a protease inhibitor
should be considered the standard of care for pa-
tients with advanced HIV infection.27
Supported by a cooperative agreement (UC64/CCU5096889-03) be-
tween the Centers for Disease Control and Prevention and the Health Re-
search Network of Apache Medical Systems.
We are indebted to Steven I. Marlowe, West Paces Ferry Medical
Clinic, Atlanta, who was instrumental in starting the project, and
to Joan Chmiel and John P. Phair, Northwestern University Medical
School, Chicago, for their many helpful comments and suggestions.
APPENDIX
The HIV Outpatient Study investigators were as follows: A.C. Moorman
and S.D. Holmberg (project officers) and J.C. Von Bargen, Division of
HIV/AIDS Prevention, National Center for HIV, STD, and TB Preven-
tion, Centers for Disease Control and Prevention, Atlanta; F.J. Palella and
C. Gardner, Northwestern University Medical School, Chicago; M.O.
Loveless and K. McKettrick, Oregon Health Sciences University, Portland;
B.G. Yangco, K.D. Halkias, and C. Lapierre, Infectious Disease Research
Institute, Tampa, Fla.; D.J. Ward and R. Deighton, Washington, D.C.; J.
Fuhrer and L. Aronson-Ryan, State University of New York, Stony Brook;
J.B. Marzouk, R.T. Phelps, P. Joseph, and M. Rachel, Adult Immunology
Clinic, Oakland, Calif.; R.E. McCabe, Fairmont Hospital, Oakland, Calif.;
W.A. Alexander and S. Lingam, Southside HealthCare, Atlanta; K.A. Lich-
tenstein, K.S. Greenberg, P. Zellner, B. Widick, and C. Stewart, Columbia
Rose Medical Center, Denver; D.J. Aschman, K.M. Delaney, and K.M.
Ragland, Health Research Network of Apache Medical Systems, Chicago.
REFERENCES
1. Ho DD, Neumann AU, Perelson AS, Chen W, Leonard JM, Markowitz
M. Rapid turnover of plasma virions and CD4 lymphocytes in HIV-1 in-
fection. Nature 1995;373:123-6.
2. Perelson AS, Neumann AU, Markowitz M, Leonard JM, Ho DD. HIV-
1 dynamics in vivo: virion clearance rate, infected cell life-span, and viral
generation time. Science 1996;271:1582-6.
3. Coffin JM. HIV population dynamics in vivo: implications for genetic
variation, pathogenesis, and therapy. Science 1995;267:483-9.
4. Hammer SM, Katzenstein DA, Hughes MD, et al. A trial comparing
nucleoside monotherapy with combination therapy in HIV-infected adults
with CD4 cell counts from 200 to 500 per cubic millimeter. N Engl J Med
1996;335:1081-90.
5. Delta Coordinating Committee. Delta: a randomised double-blind con-
860  Ma r ch 2 6 , 1 9 9 8
15
Back to Table of Contents
nejm.org
trolled trial comparing combinations of zidovudine plus didanosine or zal-
citabine with zidovudine alone in HIV-infected individuals. Lancet 1996;
348:283-91.
6. Bartlett JA, Benoit SL, Johnson VA, et al. Lamivudine plus zidovudine
compared with zalcitabine plus zidovudine in patients with HIV infection:
a randomized, double-blind, placebo-controlled trial. Ann Intern Med
1996;125:161-72.
7. Gulick R, Mellors J, Havlir D, et al. Potent and sustained antiretroviral
activity of indinavir (IDV), zidovudine (ZDV) and lamivudine (3TC). In:
Supplement to the XI International Conference on AIDS, Vancouver, B.C.,
July 712, 1996. Vancouver, B.C.: XI International Conference on AIDS
Society, 1996:19. abstract.
8. Hammer SM, Squires KE, Hughes MD, et al. A controlled trial of two
nucleoside analogues plus indinavir in persons with human immunodefi-
ciency virus infection and CD4 cell counts of 200 per cubic millimeter or
less. N Engl J Med 1997;337:725-33.
9. Ho DD. Viral counts count in HIV infection. Science 1996;272:1124-
5.
10. Hughes MD, Johnson VA, Hirsch MS, et al. Monitoring plasma
HIV-1 RNA levels in addition to CD4 lymphocyte count improves as-
sessment of antiretroviral therapeutic response. Ann Intern Med 1997;126:
929-38.
11. OBrien WA, Hartigan PM, Daar ES, Simberkoff MS, Hamilton JD.
Changes in plasma HIV RNA levels and CD4 lymphocyte counts predict
both response to antiretroviral therapy and therapeutic failure. Ann Intern
Med 1997;126:939-45.
12. OBrien WA, Hartigan PM, Martin D, et al. Changes in plasma HIV-
1 RNA and CD4 lymphocyte counts and the risk of progression to
AIDS. N Engl J Med 1996;334:426-31.
13. Mellors JW, Rinaldo CR Jr, Gupta P, White RM, Todd JA, Kingsley
LA. Prognosis in HIV-1 infection predicted by the quantity of virus in
plasma. Science 1996;272:1167-70.
14. Mellors JW, Munoz A, Giorgi JV, et al. Plasma viral load and CD4
lymphocytes as prognostic markers of HIV-1 infection. Ann Intern Med
1997;126:946-54.
15. Saag MS, Holodniy M, Kuritzkes DR, et al. HIV viral load markers in
clinical practice. Nat Med 1996;2:625-9.
16. Katzenstein DA, Hammer SM, Hughes MD, et al. The relation of vi-
rologic and immunologic markers to clinical outcomes after nucleoside
therapy in HIV-infected adults with 200 to 500 CD4 cells per cubic mil-
limeter. N Engl J Med 1996;335:1091-8.
17. Update: trends in AIDS incidence, deaths, and prevalence United
States, 1996. MMWR Morb Mortal Wkly Rep 1997;46:165-73.
18. Chaisson MA, Berenson L, Li W, Schwartz S, Mojica B, Hamburg M.
Declining AIDS mortality in New York City (NYC). In: Program and
abstracts of the Fourth Conference on Retroviruses and Opportunistic
Infections, Washington, D.C., January 2226, 1997. Washington, D.C.:
IDSA Foundation for Retrovirology and Human Health, 1997:133. ab-
stract.
19. Mouton Y, Cartier F, Dellamonica P, et al. Dramatic cut in AIDS de-
fining events and hospitalization for patients under protease inhibitors (P.I)
and tritherapies (TTT) in 9 AIDS rfrence centers (ARC) and 7,391 pa-
tients. In: Program and abstracts of the Fourth Conference on Retroviruses
and Opportunistic Infections, Washington, D.C., January 2226, 1997.
Washington, D.C.: IDSA Foundation for Retrovirology and Human
Health, 1997:208. abstract.
20. Torres RA, Barr M. Impact of combination therapy for HIV infection
on inpatient census. N Engl J Med 1997;336:1531-2.
21. Hogg RS, OShaughnessy MV, Gataric N, et al. Decline in deaths from
AIDS due to new antiretrovirals. Lancet 1997;349:1294.
22. Kitahata MM, Koepsell TD, Deyo RA, Maxwell CL, Dodge WT, Wag-
ner EH. Physicians experience with the acquired immunodeficiency syn-
drome as a factor in patients survival. N Engl J Med 1996;334:701-6.
23. Markson LE, Cosler LE, Turner BJ. Implications of generalists slow
adoption of zidovudine in clinical practice. Arch Intern Med 1994;154:
1497-504.
24. Carpenter CC, Fischl MA, Hammer SM, et al. Antiretroviral therapy
for HIV infection in 1996: recommendations of an international panel.
JAMA 1996;276:146-54.
25. DAquila RT, Hughes MD, Johnson VA, et al. Nevirapine, zidovudine,
and didanosine compared with zidovudine and didanosine in patients with
HIV-1 infection: a randomized, double-blind, placebo-controlled trial.
Ann Intern Med 1996;124:1019-30.
26. Concorde Coordinating Committee. Concorde: MRC/ANRS ran-
domised double-blind controlled trial of immediate and deferred zidovu-
dine in symptom-free HIV infection. Lancet 1994;343:871-81.
27. Carpenter CCJ, Fischl MA, Hammer SM, et al. Antiretroviral therapy
for HIV infection in 1997: updated recommendations of the International
AIDS Society USA panel. JAMA 1997;277:1962-9.
20 YEARS ON THE WEB nejm.org

new england
The
journal of medicine
established in 1812 april 24 , 2003 vol. 348 no. 17

Overweight, Obesity, and Mortality from Cancer in a Prospectively

Studied Cohort of U.S. Adults
Eugenia E. Calle, Ph.D., Carmen Rodriguez, M.D., M.P.H., Kimberly Walker-Thurmond, B.A., and Michael J. Thun, M.D.

abstract

background
The influence of excess body weight on the risk of death from cancer has not been fully From the Department of Epidemiology and
characterized. Surveillance Research, American Cancer So-
ciety, Atlanta. Address reprint requests to
Dr. Calle at the American Cancer Society,
methods 1599 Clifton Rd., NE, Atlanta, GA 30329,
In a prospectively studied population of more than 900,000 U.S. adults (404,576 men or at jcalle@cancer.org.
and 495,477 women) who were free of cancer at enrollment in 1982, there were 57,145
N Engl J Med 2003;348:1625-38.
deaths from cancer during 16 years of follow-up. We examined the relation in men and Copyright 2003 Massachusetts Medical Society.
women between the body-mass index in 1982 and the risk of death from all cancers
and from cancers at individual sites, while controlling for other risk factors in multivari-
ate proportional-hazards models. We calculated the proportion of all deaths from can-
cer that was attributable to overweight and obesity in the U.S. population on the basis of
risk estimates from the current study and national estimates of the prevalence of over-
weight and obesity in the U.S. adult population.
results
The heaviest members of this cohort (those with a body-mass index [the weight in kilo-
grams divided by the square of the height in meters] of at least 40) had death rates from
all cancers combined that were 52 percent higher (for men) and 62 percent higher (for
women) than the rates in men and women of normal weight. For men, the relative risk
of death was 1.52 (95 percent confidence interval, 1.13 to 2.05); for women, the relative
risk was 1.62 (95 percent confidence interval, 1.40 to 1.87). In both men and women,
body-mass index was also significantly associated with higher rates of death due to
cancer of the esophagus, colon and rectum, liver, gallbladder, pancreas, and kidney; the
same was true for death due to non-Hodgkins lymphoma and multiple myeloma. Signif-
icant trends of increasing risk with higher body-mass-index values were observed for
death from cancers of the stomach and prostate in men and for death from cancers of
the breast, uterus, cervix, and ovary in women. On the basis of associations observed in
this study, we estimate that current patterns of overweight and obesity in the United
States could account for 14 percent of all deaths from cancer in men and 20 percent of
those in women.
conclusions
Increased body weight was associated with increased death rates for all cancers com-
bined and for cancers at multiple specific sites.

n engl j med 348;17 www.nejm.org april 24, 2003 1625

16
Back to Table of Contents
 The new england journal
20 YEARS ON THE WEB
t he relations between excess body
weight and mortality, not only from all
causes but also from cardiovascular dis-
ease, are well established.1-6 Although we have
known for some time that excess weight is also an
important factor in death from cancer,7 our knowl-
edge of the magnitude of the relation, both for all
cancers and for cancers at individual sites, and the
public health effect of excess weight in terms of to-
tal mortality from cancer is limited.
Previous studies have consistently shown asso-
ciations between adiposity and increased risk of
cancers of the endometrium, kidney, gallbladder
(in women), breast (in postmenopausal women),
and colon (particularly in men).8-12 Adenocarci-
noma of the esophagus has been linked to obesi-
ty.11,13,14 Data on cancers of the pancreas, prostate,
liver, cervix, and ovary and on hematopoietic cancers
are scarce or inconsistent.7-11,15-17 The lack of con-
sistency may be attributable to the limited number
of studies (especially those with prospective co-
horts), the limited range and variable categorization
of overweight and obesity among studies, bias intro-
duced by reverse causality with respect to smoking-
related cancers, and possibly real differences be-
tween the effects of overweight and obesity on the
incidence of cancer and on the rates of death from
some cancers.18,19
We conducted a prospective investigation in a
large cohort of U.S. men and women to determine
the relations between body-mass index (the weight
in kilograms divided by the square of the height in
meters) and the risk of death from cancer at specific
sites. This cohort has been used previously to exam-
ine the association of body-mass index and death
from any cause.5
methods
study population
The men and women in this study were selected
from the 1,184,617 participants in the Cancer Pre-
vention Study II, a prospective mortality study begun
by the American Cancer Society in 1982.20,21 The
participants were identified and enrolled by more
than 77,000 volunteers in all 50 states, the District
of Columbia, and Puerto Rico. Families were en-
rolled if at least one household member was 45 years
of age or older and all enrolled members were 30
years of age or older. The average age of the partici-
pants at enrollment was 57 years. In 1982 they com-
pleted a confidential mailed questionnaire that in-
1626 n engl j med 348;17 www.nejm.org
17
Back to Table of Contents
of medicine
nejm.org
cluded personal identifiers and elicited information
on demographic characteristics, personal and fam-
ily history of cancer and other diseases, and various
behavioral, environmental, occupational, and di-
etary exposures.
Over 99 percent of deaths that occurred from the
month of enrollment until September 1988 were as-
certained by means of personal inquiries made by
volunteers in September 1984, 1986, and 1988.22
Approximately 93 percent of all deaths occurring
after September 1988 were ascertained by linkage
with the National Death Index.22 By December 31,
1998, 24.0 percent of the participants had died,
75.8 percent were still living, and 0.2 percent were
dropped from follow-up on September 1, 1988, be-
cause of insufficient data for linkage with the Na-
tional Death Index. Death certificates or multiple
cause-of-death codes were obtained for 98.8 percent
of all known deaths.
In the base-line questionnaire, the participants
were asked their current weight, weight one year
previously, and height (without shoes). We excluded
from the analysis participants whose values for
height or weight were missing, whose weight one
year before the interview was unknown, or who had
lost more than 10 lb (4.5 kg) in the previous year;
65,436 men and 91,282 women were excluded for
these reasons. We also excluded participants with
below-normal weight according to World Health
Organization guidelines23 as indicated by a body-
mass index of less than 18.5 (3393 men and 15,769
women). In addition, we excluded participants who
had cancer (other than nonmelanoma skin cancer)
at base line (20,839 men and 47,120 women) and
those with missing information on race or smoking
history (14,086 men and 26,639 women). The eli-
gible participants for the current analysis included
404,576 men and 495,477 women. After 16 years of
follow-up, there were a total of 32,303 deaths from
cancer in men and 24,842 deaths from cancer in
women in this population.
From the final population of 900,053 partici-
pants, we defined a subgroup of those who had nev-
er smoked (107,030 men and 276,564 women).
Within this subgroup, there were a total of 5314
deaths from cancer among men and 11,648 among
women. This subgroup provided us with an oppor-
tunity to evaluate whether the association between
body-mass index and mortality was subject to re-
sidual confounding by smoking status for smok-
ing-related cancers.
april 24, 2003
20 YEARS ON THE WEB obesity and cancer mortality nejm.org

body-mass index that was coded as unspecified. Data regarding

The body-mass index, a measure of adiposity, was cancer subsites or histologic findings were not
categorized as follows: 18.5 to 24.9, 25.0 to 29.9, available.
30.0 to 34.9, 35.0 to 39.9, and 40.0 or more. These
categories correspond to those proposed by the information on covariates
World Health Organization23 for normal range, Potential confounders included in data analyses
grade 1 overweight, grade 2 overweight (30.0 were age (in single years), race (white, black, or
to 39.9), and grade 3 overweight, respectively. other), smoking status (never smoked, formerly
For many analyses, especially for cancers in spe- smoked, or currently smokes, with three categories
cific sites and among participants who had never of cigarettes smoked per day: 0 to 19, 20, and more
smoked, the upper categories of body-mass index than 20), education (less than high school, high-
were combined, because of the small numbers. In school graduate, some college, or college graduate),
our analyses and discussion, we refer to the range physical activity (none, slight, moderate, or heavy),
of 25.0 to 29.9 as corresponding to overweight alcohol use (none, less than one drink per day, one
and to values of 30.0 or more as corresponding to drink per day, or two or more drinks per day), mar-
obesity. ital status (not married or married), current aspirin
In all primary analyses, the body-mass index use (use or nonuse), a crude index of fat consump-
category of 18.5 to 24.9 (normal range) was con- tion (estimated grams per week for 20 food items,
sidered the reference group. We also conducted with the participants divided into three roughly
analyses in which we divided this group into two equal groups),25 and vegetable consumption (the
categories of 18.5 to 22.9 and 23.0 to 24.9 and con- frequency per week of consumption of nine vege-
sidered the lower category to be the reference group. tables not including potatoes with partici-
pants divided into three roughly equal groups), and
mortality end points status with respect to estrogen-replacement therapy
The end points in our analyses were deaths from all in women (never used, currently used, or formerly
cancers (codes 140.0 to 195.8 and 199.0 to 208.9 of used).
the International Classification of Diseases, Ninth Revision
[ICD-9])24 and from cancers at selected sites. Spe- statistical analysis
cific cancers accounting for at least 150 deaths in Age-adjusted death rates were calculated for each
men and 150 deaths in women included esophageal category of body-mass index and were directly
cancer (ICD-9 codes 150.0 to 150.9), stomach can- standardized to the age distribution of the entire
cer (151.0 to 151.9), colorectal cancer (153.0 to male or female study population. Relative risks (the
154.8), liver cancer (155.0 to 155.2), gallbladder age-adjusted death rate in a specific body-mass-
cancer (156.0 to 156.9), pancreatic cancer (157.0 to index category divided by the corresponding rate in
157.9), lung cancer (162.0 to 162.9), melanoma the reference category [18.50 to 24.99]) were com-
(172.0 to 172.9), breast cancer in women (174.0 to puted; the 95 percent confidence intervals used ap-
174.9), cancer of the corpus and uterus, not other- proximate variance formulas.26,27
wise specified (179 and 182.0 to 182.8), cervical We also used Cox proportional-hazards model-
cancer (180.0 to 180.9), ovarian cancer (183.0 to ing28 to compute relative risks and to adjust for oth-
183.9), prostate cancer (185), bladder cancer er potential risk factors reported at base line. The
(188.0 to 188.9), kidney cancer (189.0 to 189.9), Cox models were stratified according to age at en-
brain cancer (191.0 to 191.9), non-Hodgkins lym- rollment by the inclusion of age (in single years) in
phoma (202.0 to 202.9), multiple myeloma (203.0 the strata statement of the Cox model. The relative
to 203.8), and leukemia (204.0 to 208.9). All other risks we report are from the multivariate Cox mod-
specific cancers that contributed to total deaths els, unless otherwise noted. Tests of linear trend
from cancer but that caused fewer than 150 deaths were performed by scoring the categories of body-
or were coded as unspecified (199.0 to 199.1) were mass index, entering the score as a continuous term
analyzed as a separate category of other cancers. in the regression model, and testing the significance
Approximately 11 percent of cancers in both men of the term by the Wald chi-square test.29
and women fell into the other category. Of these, We present results for all cancers combined and
45 percent had a specific (coded) site and caused for cancer at each site on the basis of analyses of the
fewer than 150 deaths and 55 percent had a site total populations of men and women. For most in-

n engl j med 348;17 www.nejm.org april 24, 2003 1627

18
Back to Table of Contents
20 YEARS ON THE WEB The new england journal of medicine nejm.org

Table 1. Mortality from Cancer According to Body-Mass Index among U.S. Men in the Cancer Prevention Study II, 1982 through 1998.*

Type of Cancer Body-Mass Index P for Trend

18.524.9 25.029.9 30.034.9 35.039.9 40.0

All cancers
No. of deaths 13,855 15,372 2683 350 43
Death rate 578.30 546.21 636.30 738.69 841.62
RR (95% CI) 1.00 0.97 (0.940.99) 1.09 (1.051.14) 1.20 (1.081.34) 1.52 (1.132.05) 0.001
All cancers
No. of deaths 13,855 15,372 2683 393
Death rate 578.30 546.21 636.30 749.86
RR (95% CI) 1.00 0.97 (0.940.99) 1.09 (1.051.14) 1.23 (1.111.36) 0.002
Esophageal cancer
No. of deaths 329 452 81 14
Death rate 13.97 15.74 18.07 24.18
RR (95% CI) 1.00 1.15 (0.991.32) 1.28 (1.001.63) 1.63 (0.952.80) 0.008
Stomach cancer
No. of deaths 388 455 84 18
Death rate 16.24 16.09 20.34 33.99
RR (95% CI) 1.00 1.01 (0.881.16) 1.20 (0.941.52) 1.94 (1.213.13) 0.03
Colorectal cancer
No. of deaths 1,292 1,811 337 54
Death rate 53.51 64.43 79.50 101.25
RR (95% CI) 1.00 1.20 (1.121.30) 1.47 (1.301.66) 1.84 (1.392.41) <0.001
Liver cancer
No. of deaths 222 296 78 24
Death rate 9.24 10.49 19.22 47.80
RR (95% CI) 1.00 1.13 (0.941.34) 1.90 (1.462.47) 4.52 (2.946.94) <0.001
Gallbladder cancer
No. of deaths 66 94 20
Death rate 2.68 3.37 5.16
RR (95% CI) 1.00 1.34 (0.971.84) 1.76 (1.062.94) 0.02
Pancreatic cancer
No. of deaths 740 961 182 25
Death rate 31.07 33.98 42.20 48.80
RR (95% CI) 1.00 1.13 (1.031.25) 1.41 (1.191.66) 1.49 (0.992.22) <0.001
Lung cancer
No. of deaths 4,885 4,281 681 78
Death rate 206.69 150.11 156.53 149.63
RR (95% CI) 1.00 0.78 (0.750.82) 0.79 (0.730.86) 0.67 (0.540.84) <0.001
Melanoma
No. of deaths 238 279 43
Death rate 10.02 9.77 8.09
RR (95% CI) 1.00 0.95 (0.801.13) 0.85 (0.611.18) 0.32
Prostate cancer
No. of deaths 1,681 1,971 311 41
Death rate 67.36 73.02 83.00 87.35
RR (95% CI) 1.00 1.08 (1.011.15) 1.20 (1.061.36) 1.34 (0.981.83) <0.001
Bladder cancer
No. of deaths 375 421 76
Death rate 15.19 15.47 16.69
RR (95% CI) 1.00 1.03 (0.891.18) 1.14 (0.881.46) 0.36
Kidney cancer
No. of deaths 305 437 81 14
Death rate 12.83 15.25 18.50 24.84
RR (95% CI) 1.00 1.18 (1.021.37) 1.36 (1.061.74) 1.70 (0.992.92) 0.002

1628 n engl j med 348;17 www.nejm.org april 24 , 2003

19
Back to Table of Contents
 obesity and cancer mortality
20 YEARS ON THE WEB nejm.org

Table 1. (Continued.)

Type of Cancer Body-Mass Index P for Trend

18.524.9 25.029.9 30.034.9 35.039.9 40.0

Brain cancer
No. of deaths 370 461 68
Death rate 15.98 15.86 12.76
RR (95% CI) 1.00 0.98 (0.851.13) 0.79 (0.611.03) 0.14
Non-Hodgkins lymphoma
No. of deaths 518 672 147 18
Death rate 21.51 24.04 35.25 33.22
RR (95% CI) 1.00 1.08 (0.961.21) 1.56 (1.291.87) 1.49 (0.932.39) <0.001
Multiple myeloma
No. of deaths 259 368 70 11
Death rate 10.77 13.18 16.88 20.62
RR (95% CI) 1.00 1.18 (1.011.39) 1.44 (1.101.89) 1.71 (0.933.14) 0.002
Leukemia
No. of deaths 546 720 128 20
Death rate 22.51 25.60 30.40 40.52
RR (95% CI) 1.00 1.14 (1.021.28) 1.37 (1.131.67) 1.70 (1.082.66) <0.001
All other cancers
No. of deaths 1,641 1,693 320 52
Death rate 68.72 59.81 73.29 101.88
RR (95% CI) 1.00 0.89 (0.830.95) 1.06 (0.941.20) 1.29 (0.981.71) 0.98

* Participants with any of the following features at study entry were excluded: missing data on height or current weight; unknown weight one
year before entry; weight loss at least 10 lb (4.5 kg) in the previous year; body-mass index under 18.50; existing cancer (other than nonmela-
noma skin cancer); unknown race or missing data; and missing data on smoking status. RR denotes relative risk, and CI confidence interval.
The highest body-mass-index category examined varies for cancer at different sites; higher categories have been combined when necessary
because of small numbers.
The rate per 100,000 is given, standardized to the age distribution of men in the Cancer Prevention Study II.
The Cox proportional-hazards model was adjusted for age, education, smoking status and number of cigarettes smoked, physical activity,
alcohol use, marital status, race, aspirin use, fat consumption, and vegetable consumption.
This value is for the 35.039.9 and 40.0 groups combined and is provided to facilitate comparison with the types of cancer.

dividual cancer sites, the association of body-mass
index and mortality was similar whether the analy-
sis was based on the total population or on the pop-
ulation of those who had never smoked. However,
for several cancers known to be related to smoking,
the association between body-mass index and mor-
tality was substantially different in the total popu-
lation and the population of those who had never
smoked. For these cancers (in men, all cancers, lung
cancer, esophageal cancer, pancreatic cancer, and
other cancers; in women, all cancers, lung cancer,
esophageal cancer, and other cancers), the results
from the population of those who had never smoked
are also presented.
Because weight is a modifiable risk factor, we
calculated the population attributable fraction (also
termed population attributable risk, population at-
tributablerisk proportion, and excess fraction),30
an estimate of the proportion of all cancer deaths in
the United States that might be avoided if the adult
population maintained a body-mass index in the
normal range. We used methods derived by Walter31
and presented by Kleinbaum et al.32 for a multiple-
category exposure. In this analysis, calculations
were based on the multivariate-adjusted relative
risks for the total population in the Cancer Preven-
tion Study II and for the population of those in
that study who had never smoked and on preva-
lence estimates of overweight and obesity in U.S.
men and women 50 to 69 years of age from the Na-
tional Health and Nutrition Examination Survey
for 19992000.33 This calculation assumes that
the relative-risk estimates associated with over-
weight and obesity that were observed in the cur-
rent study were causal and are generalizable to the
U.S. population.

n engl j med 348;17 www.nejm.org april 24, 2003 1629

20
Back to Table of Contents
20 YEARS ON THE WEB The new england journal of medicine nejm.org

Table 2. Mortality from Cancer According to Body-Mass Index among U.S.Women in the Cancer Prevention Study II, 1982 through 1998.*

Type of Cancer Body-Mass Index P for Trend

18.524.9 25.029.9 30.034.9 35.039.9 40.0

All cancers
No. of deaths 14,779 7107 2254 517 185
Death rate 329.30 339.75 382.62 419.59 522.51
RR (95% CI) 1.00 1.08 (1.051.11) 1.23 (1.181.29) 1.32 (1.201.44) 1.62 (1.401.87) <0.001
Esophageal cancer
No. of deaths 112 56 21
Death rate 2.56 2.68 2.90
RR (95% CI) 1.00 1.20 (0.861.66) 1.39 (0.862.25) 0.13
Stomach cancer
No. of deaths 304 134 57 13
Death rate 6.87 6.37 9.88 9.85
RR (95% CI) 1.00 0.89 (0.721.09) 1.30 (0.971.74) 1.08 (0.611.89) 0.46
Colorectal cancer
No. of deaths 1,706 906 312 67 21
Death rate 38.67 43.28 53.81 56.14 63.11
RR (95% CI) 1.00 1.10 (1.011.19) 1.33 (1.171.51) 1.36 (1.061.74) 1.46 (0.942.24) <0.001
Liver cancer
No. of deaths 200 96 37 12
Death rate 4.53 4.54 6.34 7.52
RR (95% CI) 1.00 1.02 (0.801.31) 1.40 (0.972.00) 1.68 (0.933.05) 0.04
Gallbladder cancer
No. of deaths 159 86 59
Death rate 3.57 4.15 7.79
RR (95% CI) 1.00 1.12 (0.861.47) 2.13 (1.562.90) <0.001
Pancreatic cancer
No. of deaths 952 490 154 35 19
Death rate 21.47 23.24 26.20 27.70 51.65
RR (95% CI) 1.00 1.11 (1.001.24) 1.28 (1.071.52) 1.41 (1.011.99) 2.76 (1.744.36) <0.001
Lung cancer
No. of deaths 3,693 1278 305 54 19
Death rate 81.48 60.80 51.23 43.67 52.64
RR (95% CI) 1.00 0.88 (0.830.94) 0.82 (0.720.92) 0.66 (0.500.86) 0.81 (0.521.28) <0.001
Melanoma
No. of deaths 166 61 28
Death rate 3.65 2.96 3.63
RR (95% CI) 1.00 0.85 (0.631.14) 1.10 (0.731.66) 0.95
Breast cancer
No. of deaths 1,446 908 309 68 24
Death rate 39.10 51.13 60.65 67.56 84.86
RR (95% CI) 1.00 1.34 (1.231.46) 1.63 (1.441.85) 1.70 (1.332.17) 2.12 (1.413.19) <0.001
Cancer of the corpus and
uterus, not other-
wise specified
No. of deaths 333 225 105 25 16
Death rate 10.68 15.68 26.05 30.16 60.83
RR (95% CI) 1.00 1.50 (1.261.78) 2.53 (2.023.18) 2.77 (1.834.18) 6.25 (3.7510.42) <0.001
Cervical cancer
No. of deaths 80 54 16 14
Death rate 1.73 2.63 2.73 7.81
RR (95% CI) 1.00 1.38 (0.971.96) 1.23 (0.712.13) 3.20 (1.775.78) 0.001

1630 n engl j med 348;17 www.nejm.org april 24 , 2003

21
Back to Table of Contents
20 YEARS ON THE WEB obesity and cancer mortality nejm.org

Table 2. (Continued.)

Type of Cancer Body-Mass Index P for Trend

18.524.9 25.029.9 30.034.9 35.039.9 40.0

Ovarian cancer**
No. of deaths 873 437 126 49
Death rate 27.88 31.44 31.85 44.49
RR (95% CI) 1.00 1.15 (1.021.29) 1.16 (0.961.40) 1.51 (1.122.02) 0.001
Bladder cancer
No. of deaths 180 83 34
Death rate 4.21 3.93 4.82
RR (95% CI) 1.00 1.02 (0.781.33) 1.34 (0.911.95) 0.21
Kidney cancer
No. of deaths 243 153 55 12 10
Death rate 5.43 7.35 9.24 9.56 30.14
RR (95% CI) 1.00 1.33 (1.081.63) 1.66 (1.232.24) 1.70 (0.943.05) 4.75 (2.509.04) <0.001
Brain cancer
No. of deaths 467 213 64 12
Death rate 10.26 10.27 10.68 6.35
RR (95% CI) 1.00 1.02 (0.871.21) 1.10 (0.841.44) 0.74 (0.421.32) 0.96
Non-Hodgkins lymphoma
No. of deaths 576 327 88 38
Death rate 13.02 15.48 14.99 24.09
RR (95% CI) 1.00 1.22 (1.061.40) 1.20 (0.951.51) 1.95 (1.392.72) <0.001
Multiple myeloma
No. of deaths 341 187 72 20
Death rate 7.71 8.87 12.28 12.88
RR (95% CI) 1.00 1.12 (0.931.34) 1.47 (1.131.91) 1.44 (0.912.28) 0.004
Leukemia
No. of deaths 574 282 83 18
Death rate 13.05 13.53 14.17 12.72
RR (95% CI) 1.00 1.05 (0.911.21) 1.12 (0.891.42) 0.93 (0.581.49) 0.53
All other cancers
No. of deaths 1,582 801 239 61 22
Death rate 35.70 38.15 40.61 51.99 69.19
RR (95% CI) 1.00 1.11 (1.021.21) 1.20 (1.051.38) 1.47 (1.131.90) 1.83 (1.202.80) <0.001

* Participants with any of the following features at study entry were excluded: missing data on height or current weight; unknown weight one
year before entry; weight loss of at least 10 lb (4.5 kg) in the previous year; body-mass index under 18.50; existing cancer (other than nonmel-
anoma skin cancer); unknown race or missing data; and missing data on smoking status. RR denotes relative risk, and CI confidence interval.
The highest body-mass-index category examined varies for different cancer sites; upper categories have been combined when necessary be-
cause of small numbers.
The rate per 100,000 is given, standardized to the age distribution of women in the Cancer Prevention Study II.
The Cox proportional-hazards model was adjusted for age, education, smoking status and number of cigarettes smoked, physical activity, al-
cohol use, marital status, race, aspirin use, estrogen-replacement therapy, fat consumption, and vegetable consumption.
Women who were premenopausal or perimenopausal or whose menopausal status was unknown were excluded (147,583 women, with 871
deaths).
Women who had a hysterectomy were excluded (130,717 women, 25 deaths).
**Women who had either a hysterectomy or ovarian surgery were excluded (141,924 women, 389 deaths).

examined separately for the two highest body-mass-

results
index categories of 35.0 to 39.9 and 40.0 or more.
body-mass index and mortality The relative risks of death for these categories, as
from cancer in the total population compared with the group of men of normal weight
of men and women (body-mass index, 18.5 to 24.9), were 1.20 (95 per-
The numbers of deaths among men were sufficient cent confidence interval, 1.08 to 1.34) and 1.52 (95
to permit only the death rates from all cancers to be percent confidence interval, 1.13 to 2.05), respec-

n engl j med 348;17 www.nejm.org april 24, 2003 1631

22
Back to Table of Contents
 20 YEARS ON THE WEB The new england journal
tively (Table 1). We observed significant positive lin-
ear trends in death rates with increasing body-mass
index for all cancers, esophageal cancer, stomach
cancer, colorectal cancer, liver cancer, gallbladder
cancer, pancreatic cancer, prostate cancer, kidney
cancer, non-Hodgkins lymphoma, multiple myelo-
ma, and leukemia (Table 1). As compared with men
of normal weight, men with a body-mass index of
at least 35.0 had significantly elevated relative risks
of death from cancer, which ranged from 1.23 (95
percent confidence interval, 1.11 to 1.36) for death
from any cancer to 4.52 (95 percent confidence in-
terval, 2.94 to 6.94) for death from liver cancer (Ta-
ble 1). In the total population of men, a significant
inverse association was observed between body-
mass index and death from lung cancer. We did not
find significant associations between body-mass in-
dex and death from brain cancer, bladder cancer,
melanoma, or other cancers. Among men within
the normal weight range, those with a body-mass
index of 23.0 to 24.9 were not at higher risk for
death from cancer than the leanest men (those with
a body-mass index of 18.5 to 22.9), and the observed
associations in men were not larger when a leaner
group of men was used as the reference group (data
not shown).
The results for the total population of women
were similar. Women with a body-mass index of at
least 40.0 had a relative risk of death from any can-
cer of 1.62 (95 percent confidence interval, 1.40 to
1.87), as compared with women of normal weight
(Table 2). Significant positive linear trends in death
rates were observed for colorectal cancer, liver can-
cer, gallbladder cancer, pancreatic cancer, breast
cancer, cancer of the corpus and uterus, not oth-
erwise specified, cervical cancer, ovarian cancer,
kidney cancer, non-Hodgkins lymphoma, multiple
myeloma, and other cancers (Table 2). The high-
est relative risk we observed was for death from uter-
ine cancer (relative risk, 6.25 for women with body-
mass index of at least 40.0; 95 percent confidence
interval, 3.75 to 10.42). As in men, a significant in-
verse association between body-mass index and
death from lung cancer was seen in the total popu-
lation, which included smokers. Significant asso-
ciations with body-mass index were not observed
for death from esophageal cancer, stomach cancer,
melanoma, bladder cancer, brain cancer, or leuke-
mia. Although the results for total cancer mortality
in women were virtually unchanged when a leaner
reference group was used (body-mass index, 18.5 to
22.9), there were significant differences within the
1632 n engl j med 348;17 www.nejm.org
23
Back to Table of Contents
of medicine nejm.org
normal weight range for cancers of the gallbladder,
breast, and corpus and uterus, resulting in larger
elevations in risk for these cancers throughout the
entire range of overweight and obesity as compared
with the leanest reference group (the relative risk of
death from gallbladder cancer for a body-mass in-
dex of at least 30.0 was 2.44 [95 percent confidence
interval, 1.73 to 3.44]; the relative risks of death
from breast and uterine cancers for a body-mass in-
dex of at least 40.0 were 2.32 [95 percent confidence
interval, 1.54 to 3.50] and 6.87 [95 percent confi-
dence interval, 4.09 to 11.55], respectively).
body-mass index and mortality
from cancer in men and women
who had never smoked
The association between body-mass index and death
from several smoking-related cancers changed
when the analysis was restricted to men who had
never smoked. The positive associations with death
from any cancer, esophageal cancer, pancreatic can-
cer, and other cancers were of greater magnitude
among those who had never smoked than in the
total population, and the apparent inverse associ-
ation with death from lung cancer disappeared
(Table 3).
As in men, the positive association between
body-mass index and death from any cancer, esoph-
ageal cancer, and other cancers became stronger
when the analysis was restricted to women who had
never smoked, and the seemingly protective effect
of high body-mass index on mortality from lung
cancer was attenuated (Table 3). Among women
who had never smoked, the relative risk of death
from any cancer was 1.88 (95 percent confidence
interval, 1.56 to 2.27) for those with a body-mass
index of at least 40.0, as compared with women of
normal weight.
The relative risks of cancers for which we found
significant trends of increasing death rates with in-
creasing body-mass index are summarized for the
highest categories of body-mass index that we were
able to examine in men (Fig. 1) and women (Fig. 2).
population attributable fraction
We estimated the proportion of all deaths from
cancer in the U.S. population that are attributable
to overweight and obesity to be from 4.2 percent to
14.2 percent among men and from 14.3 percent
to 19.8 percent among women (Table 4). The lower
estimates are based on relative risks for the entire
population, whereas the higher estimates are based
april 24 , 2003
 obesity and cancer mortality
20 YEARS ON THE WEB nejm.org

Table 3. Mortality from Cancer According to Body-Mass Index among U.S. Men and Women in the Cancer Prevention Study II Who Had Never
Smoked, 1982 through 1998.*

Type of Cancer Body-Mass Index P for Trend

18.524.9 25.029.9 30.034.9 35.039.9 40.0

Men
All cancers
No. of deaths 2119 2638 499 58
Death rate 303.08 346.62 442.00 421.01
RR (95% CI) 1.00 1.11 (1.051.18) 1.38 (1.241.52) 1.31 (1.011.70) <0.001
Esophageal cancer
No. of deaths 24 52 11
Death rate 3.55 6.82 7.29
RR (95% CI) 1.00 1.76 (1.082.86) 1.91 (0.923.96) 0.04
Pancreatic cancer
No. of deaths 155 212 34 8
Death rate 22.57 27.87 29.75 60.69
RR (95% CI) 1.00 1.24 (1.011.54) 1.34 (0.921.95) 2.61 (1.275.35) 0.005
Lung cancer
No. of deaths 156 179 30
Death rate 22.72 23.51 23.45
RR (95% CI) 1.00 1.00 (0.801.24) 0.93 (0.631.39) 0.78
All other cancers
No. of deaths 239 290 81
Death rate 34.65 37.99 62.18
RR (95% CI) 1.00 1.06 (0.891.26) 1.68 (1.302.18) <0.001
Women
All cancers
No. of deaths 6158 3763 1327 288 112
Death rate 241.14 277.92 330.21 356.84 485.06
RR (95% CI) 1.00 1.14 (1.091.18) 1.33 (1.251.41) 1.40 (1.251.58) 1.88 (1.562.27) <0.001
Esophageal cancer
No. of deaths 29 23 14
Death rate 1.08 1.62 2.82
RR (95% CI) 1.00 1.49 (0.852.59) 2.64 (1.365.12) 0.004
Lung cancer
No. of deaths 476 224 78 17
Death rate 18.71 16.40 19.18 17.51
RR (95% CI) 1.00 0.85 (0.731.00) 0.99 (0.771.26) 0.81 (0.491.31) 0.21
All other cancers
No. of deaths 689 440 146 34 16
Death rate 26.69 31.63 36.24 42.88 72.92
RR (95% CI) 1.00 1.17 (1.041.32) 1.30 (1.081.56) 1.54 (1.082.17) 2.51 (1.524.14) <0.001

* Participants with any of the following features at study entry were excluded: missing height or current weight; unknown weight one year before
entry; weight loss of at least 10 lb (4.5 kg) in the previous year; body-mass index under 18.50; existing cancer (other than nonmelanoma skin
cancer); and missing data on smoking status. RR denotes relative risk, and CI confidence interval.
The highest body-mass-index category examined varies for different cancer sites; upper categories have been combined when necessary be-
cause of small numbers.
The rate per 100,000 is given, standardized to the age distribution of men in the Cancer Prevention Study II.
The Cox proportional-hazards model was adjusted for age, education, physical activity, alcohol use, marital status, race, aspirin use, estrogen-
replacement therapy (in women), fat consumption, and vegetable consumption.
The rate per 100,000 is given, standardized to the age distribution of women in the Cancer Prevention Study II.

n engl j med 348;17 www.nejm.org april 24, 2003 1633

24
Back to Table of Contents
 20 YEARS ON THE WEB The new england journal of medicine nejm.org

Men
Prostate (35) 1.34

Non-Hodgkins lymphoma (35) 1.49

Type of Cancer (highest BMI category)

All cancers (40) 1.52

All other cancers (30) 1.68*

Kidney (35) 1.70

Multiple myeloma (35) 1.71

Gallbladder (30) 1.76

Colon and rectum (35) 1.84

Esophagus (30) 1.91*

Stomach (35) 1.94

Pancreas (35) 2.61*

Liver (35) 4.52

0 1 2 3 4 5 6 7
Relative Risk of Death (95% Confidence Interval)

Figure 1. Summary of Mortality from Cancer According to Body-Mass Index for U.S. Men in the Cancer Prevention
Study II, 1982 through 1998.
For each relative risk, the comparison was between men in the highest body-mass-index (BMI) category (indicated in pa-
rentheses) and men in the reference category (body-mass index, 18.5 to 24.9). Asterisks indicate relative risks for men
who never smoked. Results of the linear test for trend were significant (P0.05) for all cancer sites.

on relative risks for those who never smoked. The seen in our study reflect a greater effect of body-
estimates based on relative risks among men and mass index on mortality than on incidence of cancer
women who never smoked (Table 4) do not describe at some sites.18,19 These risk estimates are proba-
the fraction of deaths attributable to overweight and bly conservative, since they are based on the total
obesity among this population only. Rather, they population, including current and former smokers.
are estimates of the fraction of deaths attributable Among women who never smoked, the relative risk
to overweight and obesity in the total U.S. popula- associated with a body-mass index of at least 40.0
tion, on the assumption that the relative risks was 88 percent; however, there were not enough
among those who never smoked offer the most deaths among men in this category for us to deter-
valid estimates of the true effect of overweight and mine the relative risk.
obesity on mortality from cancer. The proportion of all deaths from cancer that is
attributable to overweight and obesity in U.S. adults
discussion 50 years of age or older may be as high as 14 percent
in men and 20 percent in women. These estimates
The heaviest men and women (those with a body- are based on the relative risks in our study and the
mass index of at least 40.0) in the large cohort we current patterns of overweight and obesity in the
studied prospectively had death rates from all can- United States. Under the assumption that these re-
cers that were 52 percent and 62 percent higher, re- lations are causal, the public health implications for
spectively, than the rates in men and women of nor- the United States are profound: more than 90,000
mal weight. This finding is consistent with those of deaths per year from cancer might be avoided if ev-
previous studies, but the magnitude of the effect is eryone in the adult population could maintain a
somewhat larger.7,16,17 The stronger associations body-mass index under 25.0 throughout life.
we found probably reflect our ability to examine The International Agency for Research on Cancer
deaths from cancer across a wider range of over- (IARC) has concluded that there is sufficient evi-
weight and obesity than has been possible previ- dence of a cancer-preventive effect of avoidance of
ously. It is also likely that the stronger associations weight gain for cancers of the colon, breast (in post-

1634 n engl j med 348;17 www.nejm.org april 24 , 2003

25
Back to Table of Contents
 obesity and cancer mortality
20 YEARS ON THE WEB nejm.org

Women
Multiple myeloma (35) 1.44

Colon and rectum (40) 1.46

Ovary (35) 1.51

Type of Cancer (highest BMI category) 1.68
Liver (35)
All cancers (40) 1.88*

Non-Hodgkins lymphoma (35) 1.95

Breast (40) 2.12

Gallbladder (30) 2.13

All other cancers (40) 2.51*

Esophagus (30) 2.64*

Pancreas (40) 2.76

Cervix (35) 3.20

Kidney (40) 4.75

Uterus (40) 6.25

0 1 2 3 4 5 6 7 8 9 10 11
Relative Risk of Death (95% Confidence Interval)

Figure 2. Summary of Mortality from Cancer According to Body-Mass Index for U.S. Women in the Cancer Prevention
Study II, 1982 through 1998.
For each relative risk, the comparison was between women in the highest body-mass-index (BMI) category (indicated in
parentheses) and women in the reference category (body-mass index, 18.5 to 24.9). Asterisks indicate relative risks for
women who never smoked. Results of the linear test for trend were significant (P0.05) for all cancer sites.

menopausal women), endometrium, kidney (renal- esophageal cancer according to subsite, the strong-
cell carcinoma), and esophagus (adenocarcino- er association observed in participants who had nev-
ma).11 Potential biologic mechanisms include er smoked may be partly explained by the greater
increased levels of endogenous hormones (sex ster- contribution of adenocarcinoma to all esophageal
oids, insulin, and insulin-like growth factor I) as- cancer in nonsmokers than in smokers.14
sociated with overweight and obesity and the contri- Recent studies of gallbladder cancer have con-
bution of abdominal obesity to gastroesophageal sistently found elevated risks for women with a high
reflux and esophageal adenocarcinoma.11 Our study body-mass index (by a factor of about two) but gen-
supports the conclusion of the IARC. Moderate rel- erally have involved too few cases for the associa-
ative risks (less than 2.0) associated with overweight tion to be evaluated in men.7,16,17,38,39 Obesity may
and obesity both for colon cancer and for breast operate indirectly by increasing the risk of gall-
cancer in postmenopausal women have been doc- stones, which, in turn, increase the risk of gallblad-
umented consistently in casecontrol and cohort der cancer.8
studies.8,34,35 Much higher relative risks have been Studies suggest that high body-mass index is
observed for uterine cancer (2 to 10) and kidney associated with approximately a doubling of the
cancer (1.5 to 4), and the increased risk of kidney risk of pancreatic cancer in both men15,40,41 and
cancer associated with excess weight is higher in women15,41 a result similar to our findings. In
women than in men in this and most previous contrast, there is no strong support for an associa-
studies.8,36,37 Increases by a factor of two to three tion between body-mass index and prostate can-
in the risk of adenocarcinoma of the esophagus in cer.42-44 However, some data suggest a slight in-
association with high body-mass index have been crease in the risk of advanced prostate cancer or
reported,13,14 and the magnitude of this association death among patients with a high body-mass in-
has been found by other investigators to be great- dex.19,45,46 Positive associations of ovarian cancer
er in nonsmokers.13 Because we could not examine with body-mass index have been found, with relative

n engl j med 348;17 www.nejm.org april 24, 2003 1635

26
Back to Table of Contents
 20 YEARS ON THE WEB The new england journal of medicine nejm.org

Table 4. Estimated Population Attributable Fraction According to Body-Mass Index for Mortality from Cancer
in U.S. Men and Women.*

Body-Mass Index Men Women

Population Population
Prevalence Relative Attributable Prevalence Relative Attributable
of Exposure Risk Fraction of Exposure Risk Fraction
% % % %
All subjects
25.029.9 42.1 0.97 1.2 28.8 1.08 2.0
30.034.9 21.0 1.09 1.8 22.5 1.23 4.5
35.039.9 9.2 1.20 1.8 10.7 1.32 3.0
40.0 3.6 1.52 1.9 7.9 1.62 4.9
Total population attributable fraction 4.2 14.3
Subjects who never smoked
25.029.9 42.1 1.11 4.0 28.8 1.14 3.3
30.034.9 21.0 1.38 6.8 22.5 1.33 6.1
35.039.9 12.8 1.31 3.4 10.7 1.40 3.5
40.0 7.9 1.88 7.0
Total population attributable fraction 14.2 19.8

* Data on prevalence of exposure among men are from the National Health and Nutrition Examination Survey (NHANES)
(19992000) for U.S. men 50 to 69 years of age. Data on prevalence of exposure among women are from NHANES
(19992000) for U.S. women 50 to 69 years of age. Data on relative risk are from the Cancer Prevention Study II (Table 1
for data for all men, Table 2 for data for all women, and Table 3 for data for men and women who never smoked). The
population attributable fraction was calculated with the use of equation 9.6 in Kleinbaum et al.32
Values for men are applicable to men with a body-mass index of 35.0 or higher.

risks in the range of 1.5 to 2.0 for the highest body-
mass-index categories studied7,47-49; however, sev-
eral studies have not shown an association.16,17,50,51
Two studies that examined obesity and liver can-
cer found an excess risk in both men and women,
with relative risks in the range of 2.0 to 4.016,17 a
result similar to our findings. Our results and those
of a prospective study in Sweden16 suggest that this
excess risk is higher among men than among wom-
en. Obesity also increases the risk of adenocarcino-
ma of the gastric cardia,13,14,52 but the data are lim-
ited and inconsistent for noncardia cancers of the
stomach.13,52 In an earlier American Cancer Society
cohort, as in our study, mortality from stomach can-
cer was associated with body-mass index among
men but not among women.7 This difference may
reflect the greater contribution of the cardia to all
cases of gastric cancer in men than in women. Our
results for cervical cancer are also similar to those in
the earlier American Cancer Society cohort,7 where-
as the increased risks observed in two cohorts of
hospitalized patients with a diagnosis of obesity,
as compared with the general population, were
much smaller than those observed in our study.16,17
Data are scarce on the relation between hematopoi-
etic cancers and body-mass index, and the findings
have not been consistent.7,16,17,53
Our results are based on data on mortality and
reflect the combined influence of body-mass index
both on the incidence of cancer and on survival,
whereas most of the available literature on site-spe-
cific cancers is based on incidence data. Our results
may be influenced by adiposity-related differences
in the diagnosis or treatment of cancer, as well as by
true biologic effects of adiposity on survival. For ex-
ample, adiposity has been shown to be adversely as-
sociated with the incidence of breast cancer, survival
among women with the disease,54 and stage at di-
agnosis.55,56 These combined effects may explain
why the association of body-mass index with mor-

1636 n engl j med 348;17 www.nejm.org april 24 , 2003

27
Back to Table of Contents
20 YEARS ON THE WEB obesity and cancer mortality nejm.org

tality from breast cancer in our study cohort is some-
what stronger than those in previous studies of in-
cident breast cancer.18
Smoking profoundly alters the relation between
body-mass index and many causes of death. We be-
lieve that public health recommendations regarding
optimal body mass are most valid when they are
based on data from studies of persons who have
never smoked.5,57,58 For smoking-related cancers,
the prospective effects of body-mass index on the
risk of death among smokers cannot be separated
from the prospective effects of smoking namely,
decreased body mass and increased risk of death.
Previous analyses of the Cancer Prevention Study II
cohort demonstrated that the apparent inverse as-
sociation of body-mass index and mortality due to
lung cancer was incrementally attenuated with in-
creasingly complex statistical control for smoking
in multivariate models, and it disappeared entirely
when the analysis was restricted to those who had
never smoked.59 Thus, for smoking-related can-
cers, we believe that the estimates of relative risk
and population attributable fraction presented for
the total population (Tables 1, 2, and 4) are likely
to be underestimates, whereas those presented for
the population of those who never smoked (Tables 3
and 4) offer the most valid estimates of the true ef-
fect of overweight and obesity on mortality from
these cancers.
We used self-reported weight and height at study
entry to calculate the body-mass index, a widely used
index of weight adjusted for height.60,61 Although
self-reported weight and height are highly correlat-
ed with measured weight and height,62-64 the small
error that exists is generally systematic, with an
overestimation of height and an underestimation
of weight, especially at higher weights.62-64 Thus,
our measure of body-mass index probably under-
estimated the true body-mass-index values among
overweight persons. We had no direct measure of
adiposity or of lean body mass and no measure of
central adiposity, such as the waist-to-hip ratio. We
also could not evaluate the effect of weight change
or weight cycling throughout the follow-up period,
and we could not estimate the extent of misclassifi-
cation that weight change might introduce. The as-
sociations observed in this study were not changed
in models that excluded deaths in the first two years
of follow-up.
The large size of our cohort allowed us to inves-
tigate the effect of overweight and obesity on the
occurrence of 57,000 deaths from cancer among
900,000 men and women who were free of cancer
at base line. Overweight and obesity are associated
with the risk of death from all cancers and with
death from cancers at many specific sites. From our
results, we estimate that 90,000 deaths due to cancer
could be prevented each year in the United States if
men and women could maintain normal weight. It
is unlikely that this goal can be achieved without
concerted effort and substantial investment on the
part of policymakers, educators, clinicians, employ-
ers, and schools to promote physical activity and
healthful dietary practices as a cultural norm.

referen c es
1. Manson JE, Willett WC, Stampfer MJ, et
al. Body weight and mortality among wom-
en. N Engl J Med 1995;333:677-85.
2. Willett WC, Manson JE, Stampfer MJ,
et al. Weight, weight change, and coronary
heart disease in women: risk within the nor-
mal weight range. JAMA 1995;273:461-5.
3. Stevens J, Plankey MW, Williamson DF,
et al. The body mass index-mortality relation-
ship in white and African American women.
Obes Res 1998;6:268-77.
4. Lindsted KD, Singh PN. Body mass and
26 y risk of mortality among men who never
smoked: a re-analysis of men from the Ad-
ventist Mortality Study. Int J Obes Relat
Metab Disord 1998;22:544-8.
5. Calle EE, Thun MJ, Petrelli JM, Rodriguez
C, Heath CW Jr. Body-mass index and mor-
tality in a prospective cohort of U.S. adults.
N Engl J Med 1999;341:1097-105.
6. Kopelman P. Obesity as a medical prob-
lem. Nature 2000;404:635-43.
7. Lew EA, Garfinkel L. Variations in mor-
tality by weight among 750,000 men and
women. J Chronic Dis 1979;32:563-76.
8. World Cancer Research Fund. Food, nu-
trition and the prevention of cancer: a global
perspective. Washington, D.C.: American In-
stitute for Cancer Research, 1997:371-3.
9. Carroll K. Obesity as a risk factor for cer-
tain types of cancer. Lipids 1998;33:1055-9.
10. Bergstrom A, Pisani P, Tenet V, Wolk A,
Adami H-O. Overweight as an avoidable
cause of cancer in Europe. Int J Cancer 2001;
91:421-30. [Erratum, Int J Cancer 2001;92:
927.]
11. IARC handbooks of cancer prevention.
Vol. 6. Weight control and physical activity.
Lyons, France: International Agency for Re-
search on Cancer, 2002.
12. Peto J. Cancer epidemiology in the last
century and the next decade. Nature 2001;
411:390-5.
13. Chow W-H, Blot WJ, Vaughan TL, et al.
Body mass index and risk of adenocarcino-
mas of the esophagus and gastric cardia.
J Natl Cancer Inst 1998;90:150-5.
14. Vaughan TL, Davis S, Kristal A, Thomas
DB. Obesity, alcohol, and tobacco as risk
factors for cancers of the esophagus and gas-
tric cardia: adenocarcinoma versus squa-
mous cell carcinoma. Cancer Epidemiol
Biomarkers Prev 1995;4:85-92.
15. Michaud DS, Giovannucci E, Willett WC,
Colditz G, Stampfer M, Fuchs C. Physical
activity, obesity, height, and the risk of pan-
creatic cancer. JAMA 2001;286:921-9.
16. Wolk A, Gridley G, Svensson M, et al.
A prospective study of obesity and cancer
risk (Sweden). Cancer Causes Control 2001;
12:13-21.
17. Moller H, Mellemgaard A, Lindvig K,
Olsen J. Obesity and cancer risk: a Danish
record-linkage study. Eur J Cancer 1994;
30A:344-50.
18. Petrelli JM, Calle EE, Rodriguez C, Thun

n engl j med 348;17 www.nejm.org april 24, 2003 1637

28
Back to Table of Contents
 20 YEARS ON THE WEB
MJ. Body mass index, height, and postmeno-
pausal breast cancer mortality in a prospec-
tive cohort of US women. Cancer Causes
Control 2002;13:325-32.
19. Rodriguez C, Patel AV, Calle EE, Jacobs
EJ, Chao A, Thun MJ. Body mass index,
height, and prostate cancer mortality in two
large cohorts of adult men in the United
States. Cancer Epidemiol Biomarkers Prev
2001;10:345-53.
20. Garfinkel L. Selection, follow-up, and
analysis in the American Cancer Society pro-
spective studies: In: Selection, follow-up, and
analysis in prospective studies: a workshop.
National Cancer Institute monograph 67. Be-
thesda, Md.: National Cancer Institute, 1985:
49-52. (NIH publication no. 85-2713.)
21. Stellman SD, Garfinkel L. Smoking hab-
its and tar levels in a new American Cancer
Society prospective study of 1.2 million men
and women. J Natl Cancer Inst 1986;76:
1057-63.
22. Calle EE, Terrell DD. Utility of the Na-
tional Death Index for ascertainment of mor-
tality among Cancer Prevention Study II par-
ticipants. Am J Epidemiol 1993;137:235-41.
23. Physical status: the use and interpreta-
tion of anthropometry: report of a WHO Ex-
pert Committee. World Health Organ 1995;
854:1-452.
24. International classification of diseases:
manual of the international statistical classi-
fication of diseases, injuries, and causes of
death. Vol. 1. Geneva: World Health Organi-
zation, 1977.
25. Thun MJ, Calle EE, Namboodiri MM, et
al. Risk factors for fatal colon cancer in a
large prospective study. J Natl Cancer Inst
1992;84:1491-500.
26. Flanders WD. Approximate variance for-
mulas for standardized rate ratios. J Chronic
Dis 1984;37:449-53.
27. Rothman KJ, Greenland S. Modern epi-
demiology. 2nd ed. Philadelphia: Lippincott
Raven, 1998.
28. Cox DR. Regression models and life-
tables. J R Stat Soc [B] 1972;34:187-220.
29. Cox DR, Oakes D. Analysis of survival
data. London: Chapman & Hall, 1984.
30. Rockhill B, Newman B, Weinberg C. Use
and misuse of population attributable frac-
tions. Am J Public Health 1998;88:15-9.
31. Walter SD. The estimation and interpre-
tation of attributable risk in health research.
Biometrics 1976;32:829-49.
32. Kleinbaum DG, Kupper LL, Morgen-
stern H. Epidemiologic research: principles
and quantitative methods. Belmont, Calif.:
Lifetime Learning, 1982:163.
33. National Health and Nutrition Exami-
nation Survey 1999-2000. Atlanta: Centers
for Disease Control and Prevention, 2002.
(Accessed April 1, 2003, at http://
1638 n engl j med 348;17
obesity and cancer mortality
www.cdc.gov/nchs/about/major/nhanes/
NHANES99_00.htm.)
34. Hunter DJ, Willett WC. Nutrition and
breast cancer. Cancer Causes Control 1996;
7:56-68.
35. Potter JD. Nutrition and colorectal can-
cer. Cancer Causes Control 1996;7:127-46.
36. Hill HA, Austin H. Nutrition and en-
dometrial cancer. Cancer Causes Control
1996;7:19-32.
37. Wolk A, Lindblad P, Adami H-O. Nutri-
tion and renal cell cancer. Cancer Causes
Control 1996;7:5-18.
38. Strom BL, Soloway RD, Rios-Dalenz JL,
et al. Risk factors for gallbladder cancer:
an international collaborative case-control
study. Cancer 1995;76:1747-56.
39. Zatonski WA, Lowenfels AB, Boyle P, et
al. Epidemiologic aspects of gallbladder can-
cer: a case-control study of the SEARCH Pro-
gram of the International Agency for Re-
search on Cancer. J Natl Cancer Inst 1997;
89:1132-8.
40. Gapstur SM, Gann PH, Lowe W, Liu K,
Colangelo L, Dyer A. Abnormal glucose me-
tabolism and pancreatic cancer mortality.
JAMA 2000;283:2552-8.
41. Silverman DT, Swanson CA, Gridley G,
et al. Dietary and nutritional factors and
pancreatic cancer: a case-control study
based on direct interviews. J Natl Cancer
Inst 1998;90:1710-9.
42. Kolonel LN. Nutrition and prostate can-
cer. Cancer Causes Control 1996;7:83-94.
43. Calle E. Do anthropometric measures
predict risk of prostate cancer? Am J Epide-
miol 2000;151:550-3.
44. Nomura A. Body size and prostate can-
cer. Epidemiol Rev 2001;23:126-31.
45. Giovannucci E, Rimm EB, Stampfer MJ,
Colditz GA, Willett WC. Height, body
weight, and risk of prostate cancer. Cancer
Epidemiol Biomarkers Prev 1997;6:557-63.
46. Andersson SO, Wolk A, Bergstrom R, et
al. Body size and prostate cancer: a 20-year
follow-up study among 135006 Swedish con-
struction workers. J Natl Cancer Inst 1997;
89:385-9.
47. Farrow DC, Weiss NS, Lyon JL, Daling
JR. Association of obesity and ovarian cancer
in a case-control study. Am J Epidemiol 1989;
129:1300-4.
48. Cramer DW, Welch WR, Hutchinson GB,
Willett W, Scully RE. Dietary animal fat in re-
lation to ovarian cancer risk. Obstet Gynecol
1984;63:833-8.
49. Purdie DM, Bain CJ, Webb PM, White-
man DC, Pirozzo S, Green AC. Body size and
ovarian cancer: case-control study and sys-
tematic review (Australia). Cancer Causes
Control 2001;12:855-63.
50. Mink PJ, Folsom AR, Sellers TA, Kushi
LH. Physical activity, waist-to-hip ratio, and
www.nejm.org april 24, 2003
29
Back to Table of Contents
nejm.org
other risk factors for ovarian cancer: a fol-
low-up study of older women. Epidemiolo-
gy 1996;7:38-45.
51. Slattery ML, Schuman KL, West DW,
French TK, Robison LM. Nutrient intake
and ovarian cancer. Am J Epidemiol 1989;
130:497-502.
52. Ji B-T, Chow W-H, Yang G, et al. Body
mass index and the risk of cancers of the
gastric cardia and distal stomach in Shang-
hai, China. Cancer Epidemiol Biomarkers
Prev 1997;6:481-5.
53. Holly EA, Lele C, Bracci PM, McGrath
MS. Case-control study of non-Hodgkins
lymphoma among women and heterosexual
men in the San Francisco Bay area, Califor-
nia. Am J Epidemiol 1999;150:375-89.
54. Boyd NF, Campbell JE, Germanson T,
Thomson DB, Sutherland DJ, Meakin JW.
Body weight and prognosis in breast cancer.
J Natl Cancer Inst 1981;67:785-9.
55. Reeves MJ, Newcomb PA, Remington
PL, Marcus PM, MacKenzie WR. Body mass
and breast cancer: relationship between
method of detection and stage of disease.
Cancer 1996;77:301-7.
56. Wee CC, McCarthy EP, Davis RB, Phil-
lips RS. Screening for cervical and breast
cancer: is obesity an unrecognized barrier to
preventive care? Ann Intern Med 2000;132:
697-704.
57. Willett WC, Dietz WH, Colditz GA.
Guidelines for healthy weight. N Engl J Med
1999;341:427-34.
58. Allison DB, Fontaine KR, Manson JE,
Stevens J, VanItallie TB. Annual deaths at-
tributable to obesity in the United States.
JAMA 1999;282:1530-8.
59. Henley SJ, Flanders WD, Manatunga A,
Thun MJ. Leanness and lung cancer risk:
fact or artifact? Epidemiology 2002;13:268-
76.
60. Kuczmarski RJ, Carroll MD, Flegal KM,
Troiano RP. Varying body mass index cutoff
points to describe overweight prevalence
among U.S. adults: NHANES III (1988 to
1994). Obes Res 1997;5:542-8.
61. Willett WC. Nutritional epidemiology.
2nd ed. New York: Oxford University Press,
1998.
62. Palta M, Prineas RJ, Berman R, Hannan
P. Comparison of self-reported and meas-
ured height and weight. Am J Epidemiol
1982;115:223-30.
63. Stewart AW, Jackson RT, Ford MA, Bea-
glehole R. Underestimation of relative weight
by use of self-reported height and weight.
Am J Epidemiol 1987;125:122-6.
64. Stevens J, Keil JE, Waid R, Gazes PC. Ac-
curacy of current, 4-year, and 28-year self-
reported body weight in an elderly popula-
tion. Am J Epidemiol 1990;132:1156-63.
Copyright 2003 Massachusetts Medical Society.
20 YEARS ON THE WEB nejm.org

new england
The
journal of medicine
established in 1812 august 19, 2010 vol. 363 no. 8

Improved Survival with Ipilimumab in Patients

with Metastatic Melanoma
F. Stephen Hodi, M.D., Steven J. ODay, M.D., David F. McDermott, M.D., Robert W. Weber, M.D.,
Jeffrey A. Sosman, M.D., John B. Haanen, M.D., Rene Gonzalez, M.D., Caroline Robert, M.D., Ph.D.,
Dirk Schadendorf, M.D., Jessica C. Hassel, M.D., Wallace Akerley, M.D., Alfons J.M. van den Eertwegh, M.D., Ph.D.,
Jose Lutzky, M.D., Paul Lorigan, M.D., Julia M. Vaubel, M.D., Gerald P. Linette, M.D., Ph.D., David Hogg, M.D.,
Christian H. Ottensmeier, M.D., Ph.D., Celeste Lebb, M.D., Christian Peschel, M.D., Ian Quirt, M.D.,
Joseph I. Clark, M.D., Jedd D. Wolchok, M.D., Ph.D., Jeffrey S. Weber, M.D., Ph.D., Jason Tian, Ph.D.,
Michael J. Yellin, M.D., Geoffrey M. Nichol, M.B., Ch.B., Axel Hoos, M.D., Ph.D., and Walter J. Urba, M.D., Ph.D.

A bs t r ac t

Background
An improvement in overall survival among patients with metastatic melanoma has Drs. Hodi and ODay contributed equally
to this article.
been an elusive goal. In this phase 3 study, ipilimumab which blocks cytotoxic
T-lymphocyteassociated antigen 4 to potentiate an antitumor T-cell response The authors affiliations and participating
administered with or without a glycoprotein 100 (gp100) peptide vaccine was com- investigators are listed in the Appendix. Ad-
dress reprint requests to Dr. Hodi at the
pared with gp100 alone in patients with previously treated metastatic melanoma. DanaFarber Cancer Institute, 44 Binney
St., Boston, MA 02115, or at stephen_
Methods hodi@dfci.harvard.edu.
A total of 676 HLA-A*0201positive patients with unresectable stage III or IV mela-
noma, whose disease had progressed while they were receiving therapy for meta- This article (10.1056/NEJMoa1003466) was
published on June 5, 2010, and last updated
static disease, were randomly assigned, in a 3:1:1 ratio, to receive ipilimumab plus on September 1, 2010, at NEJM.org.
gp100 (403 patients), ipilimumab alone (137), or gp100 alone (136). Ipilimumab, at a
N Engl J Med 2010;363:711-23.
dose of 3 mg per kilogram of body weight, was administered with or without gp100
Copyright 2010 Massachusetts Medical Society.
every 3 weeks for up to four treatments (induction). Eligible patients could receive
reinduction therapy. The primary end point was overall survival.
Results
The median overall survival was 10.0 months among patients receiving ipilimumab
plus gp100, as compared with 6.4 months among patients receiving gp100 alone
(hazard ratio for death, 0.68; P<0.001). The median overall survival with ipilimumab
alone was 10.1 months (hazard ratio for death in the comparison with gp100 alone,
0.66; P = 0.003). No difference in overall survival was detected between the ipili-
mumab groups (hazard ratio with ipilimumab plus gp100, 1.04; P = 0.76). Grade 3
or 4 immune-related adverse events occurred in 10 to 15% of patients treated with
ipilimumab and in 3% treated with gp100 alone. There were 14 deaths related to
the study drugs (2.1%), and 7 were associated with immune-related adverse events.
Conclusions
Ipilimumab, with or without a gp100 peptide vaccine, as compared with gp100 alone,
improved overall survival in patients with previously treated metastatic melanoma.
Adverse events can be severe, long-lasting, or both, but most are reversible with ap-
propriate treatment. (Funded by Medarex and Bristol-Myers Squibb; ClinicalTrials.gov
number, NCT00094653.)
n engl j med 363;8 nejm.org august 19, 2010 711
30
Back to Table of Contents
 The n e w e ng l a n d j o u r na l of m e dic i n e
20 YEARS ON THE WEB nejm.org

T
he incidence of metastatic mela
noma has increased over the past three de-
cades,1,2 and the death rate continues to
rise faster than the rate with most cancers.3 The
World Health Organization (WHO) estimates that
worldwide there are 66,000 deaths annually from
skin cancer, with approximately 80% due to mel-
anoma.4 In the United States alone, an estimated
8600 persons died from melanoma in 2009.1 The
median survival of patients with melanoma who
have distant metastases (American Joint Com-
mittee on Cancer stage IV) is less than 1 year.5,6
No therapy is approved beyond the first-line ther-
apy for metastatic melanoma, and enrollment in
a clinical trial is the standard of care. No therapy
has been shown in a phase 3, randomized, con-
trolled trial to improve overall survival in patients
with metastatic melanoma.69
Regulatory pathways that limit the immune
response to cancer are becoming increasingly
well characterized. Cytotoxic T-lymphocyteasso-
ciated antigen 4 (CTLA-4) is an immune check-
point molecule that down-regulates pathways of
T-cell activation.10 Ipilimumab, a fully human
monoclonal antibody (IgG1) that blocks CTLA-4
to promote antitumor immunity,1114 has shown
activity in patients with metastatic melanoma
when it has been used as monotherapy in phase 2
studies.1517 Ipilimumab has also shown activity
when combined with other agents,18,19 including
cancer vaccines.20,21 One well-studied cancer vac-
cine comprises HLA-A*0201restricted peptides
derived from the melanosomal protein, glyco-
protein 100 (gp100). Monotherapy with this vac-
cine induces immune responses but has limited
antitumor activity.22 However, the results of a
recent study suggest that gp100 may improve the
efficacy of high-dose interleukin-2 in patients
with metastatic melanoma.23 With no accepted
standard of care, gp100 was used as an active
control for our phase 3 study, which evaluated
whether ipilimumab with or without gp100 im-
proves overall survival, as compared with gp100
alone, among patients with metastatic melano-
ma who had undergone previous treatment.
Me thods
Patients
Patients were eligible for inclusion in the study if ments, and the final protocol, as well as the sta-
they had a diagnosis of unresectable stage III or tistical analysis plan, are available with the full
IV melanoma and had received a previous thera- text of this article at NEJM.org.
peutic regimen containing one or more of the
following: dacarbazine, temozolomide, fotemus-
tine, carboplatin, or interleukin-2. Other inclu-
sion criteria were age of at least 18 years; life ex-
pectancy of at least 4 months; Eastern Cooperative
Oncology Group (ECOG) performance status of 0
(fully active, able to carry on all predisease per-
formance without restriction) or 1 (restricted in
physically strenuous activity but ambulatory and
able to carry out work of a light or sedentary na-
ture, such as light housework or office work)24;
positive status for HLA-A*0201; normal hemato-
logic, hepatic, and renal function; and no system-
ic treatment in the previous 28 days. Exclusion
criteria were any other cancer from which the
patient had been disease-free for less than 5 years
(except treated and cured basal-cell or squamous-
cell skin cancer, superficial bladder cancer, or
treated carcinoma in situ of the cervix, breast, or
bladder); primary ocular melanoma; previous re-
ceipt of antiCTLA-4 antibody or cancer vaccine;
autoimmune disease; active, untreated metastases
in the central nervous system; pregnancy or lac-
tation; concomitant treatment with any nonstudy
anticancer therapy or immunosuppressive agent;
or long-term use of systemic corticosteroids.
The protocol was approved by the institution-
al review board at each participating institution
and was conducted in accordance with the ethi-
cal principles originating from the Declaration
of Helsinki and with Good Clinical Practice as
defined by the International Conference on Har-
monization. All patients (or their legal represen-
tatives) gave written informed consent before
enrollment.
Study Design and Treatment
In this randomized, double-blind, phase 3 study,
we enrolled patients at 125 centers in 13 coun-
tries in North America, South America, Europe,
and Africa. Between September 2004 and August
2008, patients were randomly assigned to one of
three study groups, with stratification according
to baseline metastasis stage (M0, M1a, or M1b
vs. M1c, classified according to the tumornode
metastasis [TNM] categorization for melanoma
of the American Joint Committee on Cancer), and
receipt or nonreceipt of previous interleukin-2
therapy. The full original protocol, a list of amend-

712 n engl j med 363;8 nejm.org august 19, 2010

31
Back to Table of Contents
 Ipilimumab for metastatic melanoma
20 YEARS ON THE WEB
Patients were randomly assigned, in a 3:1:1
ratio, to treatment with an induction course of
ipilimumab, at a dose of 3 mg per kilogram of
body weight, plus a gp100 peptide vaccine; ipi-
limumab plus gp100 placebo; or gp100 plus ipi-
limumab placebo all administered once every
3 weeks for four treatments. In the vaccine
groups, patients received two modified HLA-
A*0201restricted peptides, injected subcutane-
ously as an emulsion with incomplete Freunds
adjuvant (Montanide ISA-51): a gp100:209-
217(210M) peptide, 1 mg injected in the right
anterior thigh, and a gp100:280-288(288V) pep-
tide, 1 mg injected in the left anterior thigh.
Peptide injections were given immediately after
a 90-minute intravenous infusion of ipilimumab
or placebo. Treatment began on day 1 of week 1,
and if there were no toxic effects that could not
be tolerated, no rapidly progressive disease, and
no significant decline in performance status,
patients received an additional treatment during
weeks 4, 7, and 10. Patients in whom new lesions
developed or baseline lesions grew were allowed
to receive additional treatments to complete induc-
tion. Patients with stable disease for 3 months
duration after week 12 or a confirmed partial or
complete response were offered additional courses
of therapy (reinduction) with their assigned treat-
ment regimen if they had disease progression.
The original primary end point was the best
overall response rate (i.e., the proportion of pa-
tients with a partial or complete response). The
primary end point was amended to overall sur-
vival (with the amendment formally approved on
January 15, 2009) in the ongoing blinded study,
on the basis of phase 2 data and in alignment
with another ongoing phase 3 trial of ipilimu-
mab involving patients with metastatic melano-
ma.25 The primary comparison in overall sur-
vival was between the ipilimumab-plus-gp100
group and the gp100-alone group. Prespecified
secondary end points included a comparison of
overall survival between the ipilimumab-alone
and the gp100-alone groups and between the
two ipilimumab groups, the best overall response
rate, the duration of response, and progression-
free survival. Subgroup comparisons of overall
survival were performed across five prespecified
categories: metastasis stage (M0, M1a, or M1b
vs. M1c), receipt or nonreceipt of previous inter-
leukin-2 therapy, baseline levels of serum lactate
dehydrogenase (less than or equal to the upper
limit of the normal range vs. higher than the
n engl j med 363;8
Back to Table of Contents
nejm.org
upper limit of the normal range), age (<65 years
vs. 65 years), and sex.
The trial was designed jointly by the senior
academic authors and the sponsors, Medarex
and Bristol-Myers Squibb. Data were collected by
the sponsors and analyzed in collaboration with
the senior academic authors, who vouch for the
completeness and accuracy of the data and
analyses and for the conformance of this report
to the protocol, as amended. An initial draft of
the manuscript was prepared by six of the aca-
demic authors in collaboration with the sponsor
and a professional medical writer paid by the
sponsor. All the authors contributed to subse-
quent drafts and made the decision to submit
the manuscript for publication. All the authors
signed a confidentiality disclosure agreement
with the sponsor.
Assessments
For the assessment of a patients eligibility, each
patients HLA-A*0201 status was determined at a
central laboratory. Patients who met the study
criteria were assigned to receive treatment within
35 days after HLA typing and within 28 days af-
ter diagnostic imaging. Computed tomography
with contrast material or magnetic resonance
imaging of the brain, chest, abdomen, pelvis,
and other anatomical regions, as clinically indi-
cated, was performed. Cutaneous lesions were
photographed. Tumor assessments were per-
formed at baseline, and all patients who did not
have documented early disease progression and
who had stable disease or better at week 12 had
confirmatory scans at weeks 16 and 24 and every
3 months thereafter. Tumor responses were de-
termined by the investigators with the use of
modified WHO criteria to evaluate bidimension-
ally measurable lesions.26
Adverse events were graded according to the
National Cancer Institutes Common Terminol-
ogy Criteria for Adverse Events, version 3.0. An
immune-related adverse event was defined as an
adverse event that was associated with exposure
to the study drug and that was consistent with
an immune phenomenon. Protocol guidelines
for the management of immune-related adverse
events included the administration of cortico-
steroids (orally or intravenously), a delay in a
scheduled dose, or discontinuation of thera-
py.1517 Assigned doses were delayed in the case
of nondermatologic immune-related adverse
events of grade 2 or higher until the event im-
nejm.org august 19, 2010 713
32
 The n e w e ng l a n d j o u r na l
20 YEARS ON THE WEB
proved to grade 1 or lower; if the event did not
improve to grade 1 or lower, treatment was dis-
continued permanently. Monitoring of adverse
events continued for at least 70 days after the
last dose of study drugs had been administered
or until any ongoing event resolved or stabilized.
All patients, including those with low-grade
changes in bowel frequency or stool consistency,
were followed closely. A data and safety moni-
toring committee provided independent over-
sight of safety and the riskbenefit ratio.
During the study enrollment, the following
stopping rule was in place: if 10% or more of the
patients in any study treatment group, evaluated
cumulatively every 3 months, had a nondermato-
logic-related toxic adverse event of grade 3 or
higher that was attributable to the investigational
agents and that could not be alleviated or con-
trolled by appropriate care or corticosteroid ther-
apy within 14 days after the initiation of sup-
portive care or corticosteroid therapy, assignment
of patients to that study group would be sus-
pended until the sponsor and the data and safety
monitoring committee had reviewed the events
and determined the appropriate course of action.
Statistical Analysis
The original study sample size of 750 patients
was determined on the basis of the primary end
point of best overall response rate but was re-
vised with the new primary end point of overall
survival. We estimated that with 385 events
(deaths) among a total of 500 patients randomly
assigned to the ipilimumab-plus-gp100 and the
gp100-alone groups, the study would have at
least 90% power to detect a difference in overall
survival, at a two-sided alpha level of 0.05, with
the use of a log-rank test. A total of 481 events
were required in all three groups (assuming that
the events were distributed in a 3:1:1 ratio in the
ipilimumab-plus-gp100, ipilimumab-alone, and
gp100-alone groups, respectively). Therefore, all
patients who were randomly assigned in the
study were to be followed until at least 481 events
had occurred in the study. Enrollment was com-
pleted on July 25, 2008, when more than 650 pa-
tients had been enrolled. A post hoc power anal-
ysis showed that the 219 events observed among
a total of 273 patients randomly assigned to the
ipilimumab-alone and gp100-alone groups pro- Efficacy
vided at least 80% power to detect a difference in All the analyses of the efficacy end points re-
overall survival between the two groups, at a ported here were prespecified as per protocol.
714 n engl j med 363;8
Back to Table of Contents
of m e dic i n e
nejm.org
two-sided alpha level of 0.05, with the assump-
tion that ipilimumab alone has the same treat-
ment effect as the combination regimen of ipili-
mumab plus gp100.
Survival was defined as the time from ran-
domization to death from any cause, and pro-
gression-free survival as the time from random-
ization to documented disease progression or
death. Event-time distributions were estimated
with the use of the KaplanMeier method. Cox
proportional-hazards models, stratified accord-
ing to metastasis status and receipt or nonre-
ceipt of previous interleukin therapy, were used
to estimate hazard ratios and to test for sig-
nificance of the timing of events. All reported
P values are two-sided, and confidence intervals
are at the 95% level. Survival rates were based on
KaplanMeier estimation, and confidence inter-
vals were calculated with the use of the boot-
strap method. Descriptive statistics were used
for adverse events.
R e sult s
Patients and Treatment
Among 676 patients enrolled in the study, 403
were randomly assigned to receive ipilimumab
plus gp100, 137 to receive ipilimumab alone, and
136 to receive gp100 alone (control group) (Fig. 1
in the Supplementary Appendix, available at
NEJM.org). Included among these patients were
82 patients who had metastases in the central
nervous system at baseline, of whom 77 received
the study drug. The baseline characteristics of
the patients are shown in Table 1. Efficacy analy-
ses were performed on the intention-to-treat
population, which included all patients who had
undergone randomization (676 patients). The
safety population included all patients who had
undergone randomization and who had received
any amount of study drug (643 patients). A total
of 242 of 403 patients in the ipilimumab-plus-
gp100 group (60.0%), 88 of 137 in the ipilimu-
mab-alone group (64.2%), and 78 of 136 in the
gp100-alone group (57.4%) received all four ipi-
limumab doses or placebo infusions. The most
frequent reason for discontinuation of therapy
was disease progression.
nejm.org august 19, 2010
33
 Ipilimumab for metastatic melanoma
20 YEARS ON THE WEB nejm.org

Table 1. Baseline Characteristics of the Patients.*

Ipilimumab Ipilimumab
plus gp100 Alone gp100 Alone Total
Variable (N = 403) (N = 137) (N = 136) (N = 676)
Mean age yr 55.6 56.8 57.4 56.2
Sex no. (%)
Male 247 (61.3) 81 (59.1) 73 (53.7) 401 (59.3)
Female 156 (38.7) 56 (40.9) 63 (46.3) 275 (40.7)
ECOG performance status no. (%)
0 232 (57.6) 72 (52.6) 70 (51.5) 374 (55.3)
1 166 (41.2) 64 (46.7) 61 (44.9) 291 (43.0)
2 4 (1.0) 1 (0.7) 4 (2.9) 9 (1.3)
3 1 (0.2) 0 0 1 (0.1)
Unknown 0 0 1 (0.7) 1 (0.1)
M stage no. (%)
M0 5 (1.2) 1 (0.7) 4 (2.9) 10 (1.5)
M1a 37 (9.2) 14 (10.2) 11 (8.1) 62 (9.2)
M1b 76 (18.9) 22 (16.1) 23 (16.9) 121 (17.9)
M1c 285 (70.7) 100 (73.0) 98 (72.1) 483 (71.4)
Lactate dehydrogenase level no. (%)
Upper limit of the normal range 252 (62.5) 84 (61.3) 81 (59.6) 417 (61.7)
>Upper limit of the normal range 149 (37.0) 53 (38.7) 52 (38.2) 254 (37.6)
Unknown 2 (0.5) 0 3 (2.2) 5 (0.7)
CNS metastases at baseline no. (%) 46 (11.4) 15 (10.9) 21 (15.4) 82 (12.1)
Received study drug 42 (10.4) 15 (10.9) 20 (14.7) 77 (11.4)
Had had previous treatment for CNS 39 (9.7) 15 (10.9) 19 (14.0) 73 (10.8)
metastases
Previous systemic therapy for metastatic 403 (100.0) 137 (100.0) 136 (100.0) 676 (100.0)
disease no. (%)
Previous interleukin-2 therapy no. (%) 89 (22.1) 32 (23.4) 33 (24.3) 154 (22.8)

* Percentages may not total 100 because of rounding. CNS denotes central nervous system.
The Eastern Cooperative Oncology Group (ECOG) status ranges from 0 to 5, with higher scores indicating greater im-
pairment (5 indicates death).
The metastasis (M) stage was classified according to the tumornodemetastasis (TNM) categorization for melanoma
of the American Joint Committee on Cancer.

Patients were followed for up to 55 months, with
median follow-up times for survival of 21.0
months in the ipilimumab-plus-gp100 group,
27.8 months in the ipilimumab-alone group, and
17.2 months in the gp100-alone group. The me-
dian overall survival in the ipilimumab-plus-
gp100 group was 10.0 months (95% confidence
interval [CI], 8.5 to 11.5), as compared with 6.4
months (95% CI, 5.5 to 8.7) in the gp100-alone
group (hazard ratio for death, 0.68; P<0.001).
The median overall survival in the ipilimumab-
alone group was 10.1 months (95% CI, 8.0 to
13.8) (hazard ratio for death with ipilimumab
alone as compared with gp100 alone, 0.66;
P = 0.003). No difference in overall survival was
detected between the two ipilimumab groups
(hazard ratio for death with ipilimumab plus
gp100, 1.04; P = 0.76) (Fig. 1). Analyses of sur-
vival showed that the rates of overall survival in
the ipilimumab-plus-gp100 group, the ipilimu-
mab-alone group, and the gp100-alone group,
respectively, were 43.6%, 45.6%, and 25.3% at 12
months, 30.0%, 33.2%, and 16.3% at 18 months,
and 21.6%, 23.5%, and 13.7% at 24 months. The

n engl j med 363;8 nejm.org august 19, 2010 715

34
Back to Table of Contents
20 YEARS ON THE WEB The n e w e ng l a n d j o u r na l of m e dic i n e nejm.org

effect of ipilimumab on overall survival was inde-

Ipi plus gp100 Ipi gp100
Censored Censored Censored
pendent of age, sex, baseline serum lactate dehy-
drogenase levels, metastasis stage of disease,
A Overall Survival and receipt or nonreceipt of previous interleu-
100 kin-2 therapy (Fig. 2).
90 A 19% reduction in the risk of progression
80
was noted with ipilimumab plus gp100, as com-
Overall Survival (%)

70
pared with gp100 alone (hazard ratio, 0.81;
60
P<0.05), and a 36% reduction in risk of progres-
50
sion was seen with ipilimumab alone as com-
40
30
pared with gp100 alone (hazard ratio, 0.64;
20
P<0.001). The reduction in risk with ipilimumab
10 plus gp100 was less than that with ipilimumab
0 alone (hazard ratio with ipilimumab plus gp100,
0 4 8 12 16 20 24 28 32 36 40 44 48 52 56 1.25; P = 0.04). The median values for progres-
Months sion-free survival were similar in all groups at
No. at Risk the time of the first assessment of progression
Ipi plus gp100 403 297 223 163 115 81 54 42 33 24 17 7 6 4 0 (week 12), after which there was a separation
Ipi 137 106 79 56 38 30 24 18 13 13 8 5 2 1 0
gp100 136 93 58 32 23 17 16 7 5 5 3 1 0 0 0 between the curves (Fig. 1B).
The highest percentage of patients with an
B Progression-free Survival objective response or stable disease was in the
100 ipilimumab-alone group (Table 2); this group
90 had a best overall response rate of 10.9% and a
Progression-free Survival (%)

80 disease control rate (the proportion of patients

70 with a partial or complete response or stable
60
disease) of 28.5%. In the ipilimumab-alone group,
50
9 of 15 patients (60.0%) maintained an objective
40
response for at least 2 years (26.5 to 44.2 months
30
[ongoing]), and in the ipilimumab-plus-gp100
20
10
group, 4 of 23 patients (17.4%) maintained the
0
response for at least 2 years (27.9 to 44.4 months
0 4 8 12 16 20 24 28 32 36 40 44 48 52 [ongoing]). Neither of the two patients in the
Months gp100-alone group who had a partial response
No. at Risk maintained the response for 2 years. Responses
Ipi plus gp100 403 85 52 27 17 14 10 8 5 4 2 2 1 0 to ipilimumab continued to improve beyond week
Ipi 137 37 26 17 13 10 10 9 6 4 2 1 0 0
gp100 136 18 7 5 3 2 2 2 1 0 0 0 0 0 24: in the ipilimumab-plus-gp100 group, 3 pa-
tients with disease progression improved to stable
Figure 1. KaplanMeier
AUTHOR: Hodi Curves for Overall SurvivalRETAKE: and Progression-free
1st
disease, 3 with stable disease improved to a par-
Survival in the Intention-to-Treat Population. 2nd tial response, and 1 with a partial response im-
FIGURE: 1 of 2
The median follow-up for overall survival (Panel A) in the ipilimumab
Revised
3rd
(Ipi)- proved to a complete response; in the ipilimu-
plus-glycoprotein
ARTIST:100
ts (gp100) group was 21.0 months, and the median mab-alone group, 2 patients with stable disease
SIZE
overall survival was 10.0 months (95% CI, 8.5 to 11.5); in4 col the ipilimumab-
TYPE: Line Combo 4-C H/T 22p3
improved to a partial response and 3 with a
alone group, the median follow-up was 27.8 months, and the median over-
all survival, 10.1 months (95% CI, 8.0
AUTHOR, to 13.8);
PLEASE NOTE:and in the gp100-alone partial response improved to a complete re-
group, the median Figure has beenwas
follow-up redrawn
17.2 and type has
months, been
and thereset.
median overall sur- sponse. Among 31 patients given reinduction
Please check carefully.
vival, 6.4 months (95% CI, 5.5 to 8.7). The median progression-free survival therapy with ipilimumab, a partial or complete
(Panel B) was
JOB: 2.76
362xxmonths (95% CI, 2.73 to 2.79) inISSUE: the ipilimumab-plus-
xx-xx-10 response or stable disease was achieved by 21
gp100 group, 2.86 months (95% CI, 2.76 to 3.02) in the ipilimumab-alone
(Table 2).
group, and 2.76 months (95% CI, 2.73 to 2.83) in the gp100-alone group.
The rates of progression-free survival at week 12 were 49.1% (95% CI, 44.1
to 53.9) in the ipilimumab-plus-gp100 group, 57.7% (95% CI, 48.9 to 65.5) Adverse Events
in the ipilimumab-alone group, and 48.5% (95% CI, 39.6 to 56.7) in the The adverse events reported in the safety popu-
gp100-alone group. lation are listed in Table 3. The most common
adverse events related to the study drugs were

716 n engl j med 363;8 nejm.org august 19, 2010

35
Back to Table of Contents
 Ipilimumab for metastatic melanoma
20 YEARS ON THE WEB nejm.org

A
Subgroup Ipi plus gp100 gp100 Hazard Ratio (95% CI)
no. of deaths/no. randomized
All patients 306/403 119/136 0.69 (0.560.85)
Sex
Male 191/247 66/73 0.66 (0.500.87)
Female 115/156 53/63 0.72 (0.520.99)
Age
<65 yr 219/291 81/94 0.70 (0.540.90)
65 yr 87/112 38/42 0.69 (0.471.01)
M stage at study entry
M0, M1a, M1b 78/118 31/38 0.57 (0.380.87)
M1c 228/285 88/98 0.74 (0.580.95)
Baseline LDH
ULN 178/252 66/81 0.70 (0.530.93)
>ULN 127/149 50/52 0.71 (0.510.98)
Prior use of interleukin-2
Yes 68/89 25/33 0.78 (0.491.24)
No 238/314 94/103 0.66 (0.520.84)
0.5 1.0 1.5

Ipi plus gp100 gp100

Better Better

B
Subgroup Ipi gp100 Hazard Ratio (95% CI)
no. of deaths/no. randomized
All patients 100/137 119/136 0.64 (0.490.84)
Sex
Male 53/81 66/73 0.54 (0.370.77)
Female 47/56 53/63 0.81 (0.551.20)
Age
<65 yr 69/95 81/94 0.65 (0.470.90)
65 yr 31/42 38/42 0.61 (0.380.99)
M stage at study entry
M0, M1a, M1b 21/37 31/38 0.47 (0.270.82)
M1c 79/100 88/98 0.72 (0.530.97)
Baseline LDH
ULN 52/84 66/81 0.56 (0.390.81)
>ULN 48/53 50/52 0.76 (0.511.13)
Prior use of interleukin-2
Yes 19/32 25/33 0.50 (0.280.91)
No 81/105 94/103 0.69 (0.510.93)
0.5 1.0 1.5

Ipi gp100
Better Better

Figure 2. Subgroup Analyses of Overall Survival.

The prespecified analyses of overall survival among subgroups of patients, as defined by baseline demographic
characteristics and stratification factors (metastasis [M] stage, classified according to the tumornodemetastasis
[TNM] categorization for melanoma of the American Joint Committee on Cancer; and receipt or nonreceipt of inter-
leukin-2 therapy), showed that hazard ratios were lower than 1 (indicating a lower risk of death) for each subgroup
in the ipilimumab (Ipi)-plus-glycoprotein 100 (gp100) group as compared with the gp100-alone group (Panel A) and
for each subgroup in the ipilimumab-alone group as compared with the gp100-alone group (Panel B). Hazard ratios
were estimated with the use of unstratified Cox proportional-hazards models. Horizontal lines represent 95% confi-
dence intervals. LDH denotes lactate dehydrogenase, and ULN the upper limit of the normal range.

n engl j med 363;8 nejm.org august 19, 2010 717

36
Back to Table of Contents
 The n e w e ng l a n d j o u r na l of m e dic i n e
20 YEARS ON THE WEB nejm.org

Table 2. Best Response to Treatment and Time-to-Event Data.*

Ipilimumab Ipilimumab
plus gp100 Alone gp100 Alone
Response and Time to Event (N = 403) (N = 137) (N = 136)
Overall survival
Total no. of deaths 306 100 119
Comparison with gp100 alone
Hazard ratio (95% CI) 0.68 (0.550.85) 0.66 (0.510.87)
P value by log-rank test <0.001 0.003
Comparison with ipilimumab alone
Hazard ratio (95% CI) 1.04 (0.831.30)
P value by log-rank test 0.76
Evaluation of therapy
Induction
Best overall response no. (%)
Complete response 1 (0.2) 2 (1.5) 0
Partial response 22 (5.5) 13 (9.5) 2 (1.5)
Stable disease 58 (14.4) 24 (17.5) 13 (9.6)
Progressive disease 239 (59.3) 70 (51.1) 89 (65.4)
Not evaluated 83 (20.6) 28 (20.4) 32 (23.5)
Best overall response rate % (95% CI) 5.7 (3.78.4) 10.9 (6.317.4) 1.5 (0.25.2)
P value for comparison with gp100 alone 0.04 0.001
P value for comparison with ipilimumab alone 0.04
Disease control rate % (95% CI) 20.1 (16.324.3) 28.5 (21.136.8) 11.0 (6.317.5)
P value for comparison with gp100 alone 0.02 <0.001
P value for comparison with ipilimumab alone 0.04
Time to event mo
Time to progression median (95% CI) 2.76 (2.732.79) 2.86 (2.763.02) 2.76 (2.732.83)
Time to response mean (95% CI) 3.32 (2.913.74) 3.18 (2.753.60) 2.74 (2.123.37)
Duration of response median (95% CI) 11.5 (5.4NR) NR (28.1NR) NR (2.0NR)
Reinduction
Best overall response no./total no. (%)
Complete response 0 1/8 (12.5) 0
Partial response 3/23 (13.0) 2/8 (25.0) 0
Stable disease 12/23 (52.2) 3/8 (37.5) 0
Progressive disease 8/23 (34.8) 2/8 (25.0) 1/1 (100.0)

* Of the 143 patients who could not be evaluated for a response, 33 patients did not receive any study drug and 110 pa-
tients did not have baseline or week-12 tumor assessments (or both). Percentages may not total 100 because of round-
ing. NR denotes not reached.
The disease control rate is the percentage of patients with a partial or complete response or stable disease.
A total of 40 patients (29 in the ipilimumab-plus-gp100 group; 9 in the ipilimumab-alone group, and 2 in the gp100-
alone group) were given reinduction therapy, but 8 were not included in the efficacy analyses: 3 had major protocol vio-
lations and 5 were not eligible owing to the fact that they had had a best overall response of progressive disease during
induction and were given reinduction therapy inadvertently.

718 n engl j med 363;8 nejm.org august 19, 2010

37
Back to Table of Contents
20 YEARS ON THE WEB
Ipilimumab for metastatic melanoma
immune-related events, which occurred in approx-
imately 60% of the patients treated with ipilimu-
mab and 32% of the patients treated with gp100.
The frequency of grade 3 or 4 immune-related
adverse events was 10 to 15% in the ipilimumab
groups and 3.0% in the gp100-alone group. All
immune-related events occurred during the in-
duction and reinduction periods; the immune-
related adverse events most often affected the
skin and gastrointestinal tract. The median time
to the resolution of immune-related adverse
events of grade 2, 3, or 4 was 6.3 weeks (95% CI,
4.3 to 8.4) in the ipilimumab-plus-gp100 group,
4.9 weeks (95% CI, 3.1 to 6.4) in the ipilimumab-
alone group, and 3.1 weeks (95% CI, 1.1 to not
reached) in the gp100-alone group.
The most common immune-related adverse
event was diarrhea, which occurred at any grade
in 27 to 31% of the patients in the ipilimumab
groups. After the administration of corticoste-
roids, the median time to the resolution of diar-
rhea of grade 2 or higher was 2.0 weeks for 40
of 44 patients in the ipilimumab-plus-gp100
group and 2.3 weeks for 14 of 15 patients in the
ipilimumab-alone group. In addition to cortico-
steroids, 4 patients received infliximab (anti
tumor necrosis factor antibody) for diarrhea of
grade 3 or higher or colitis. Among the 94 per-
sons who survived for 2 years, residual effects of
adverse events included those related to injection-
site reactions (16 patients), vitiligo (12), diarrhea
or colitis (e.g., proctocolitis with rectal pain) (4),
and endocrine immune-related adverse events (e.g.,
inflammation of the pituitary) that required
hormone-replacement therapy (8). Ongoing events
in the persons who survived for 2 years included
rash, pruritus, diarrhea, anorexia, and fatigue,
generally of grade 1 or 2 (in 5 to 15% of the
patients) and grade 3 leukocytosis (in one pa-
tient). There were 14 deaths related to the study
drugs (2.1%), of which 7 were associated with
immune-related adverse events.
Discussion
This phase 3 study showed that ipilimumab, ei-
ther alone or with gp100, improved overall sur-
vival as compared with gp100 alone in patients
with metastatic melanoma who had undergone
previous treatment. More than 70% of the pa-
tients had M1c disease (presence of visceral me-
n engl j med 363;8
Back to Table of Contents
nejm.org
tastases) and more than 36% had elevated lactate
dehydrogenase levels, both of which are associ-
ated with very poor survival.27,28 The eligibility
criteria for patients in this study included HLA-
A*0201positive status, on the basis of the mech-
anism of action of gp100. However, CTLA-4
blockade by ipilimumab is independent of HLA
status, as indicated by efficacy and safety out-
comes in earlier clinical trials that were similar
between HLA-A*0201positive and HLA-A*0201
negative patients21 (and unpublished data).
In our study, the efficacy of ipilimumab was
not improved by the addition of gp100. It is un-
likely that this is due to a lack of gp100 expres-
sion in the tumors, because differentiation anti-
gens have been shown to be strongly expressed
in more than 90% of melanoma tumors, regard-
less of stage.29 Some studies of adjuvant therapy
for melanoma showed that patients who were ad-
ministered nongp100 vaccines had shorter sur-
vival than did patients in the control groups.30,31
In contrast, phase 3 trials showed that in sub-
groups of patients with melanoma, vaccines had
clinical activity when used as either adjuvant
therapy or therapy for metastatic disease.32,33
Cumulative data show that gp100-based vaccines
have immunologic activity, although clinical ac-
tivity is minimal when gp100 vaccines are ad-
ministered as monotherapy.22 In a randomized,
phase 3 study involving patients with metastatic
melanoma, a significant improvement in pro-
gression-free survival and response rate, and a
nonsignificant improvement in overall survival,
were seen with gp100-plus-high-dose interleu-
kin-2, as compared with interleukin-2 alone.23
Although gp100 appeared to attenuate ipilimu-
mab responses in our study, it is important to
consider the fact that some radiographic re-
sponses of immunotherapeutic agents are not
captured by standard response criteria.34 Regard-
less, such effects of gp100 did not translate into
a difference in overall survival between the two
ipilimumab groups.
The data in this study are consistent with the
results of phase 2 trials of ipilimumab mono-
therapy in the same patient population.1517 The
data from phase 2 studies suggest that there is
a long-term survival effect of ipilimumab mono-
therapy; ipilimumab monotherapy at a dose of
3 mg per kilogram resulted in 1-year and 2-year
survival rates of 39.3% and 24.2%, respectively.16
nejm.org august 19, 2010 719
38
 720
Table 3. Adverse Events in the Safety Population.*

Adverse Event Ipilimumab plus gp100 (N = 380) Ipilimumab Alone (N = 131) gp100 Alone (N = 132)
Total Grade 3 Grade 4 Total Grade 3 Grade 4 Total Grade 3 Grade 4
number of patients (percent)
Any event 374 (98.4) 147 (38.7) 26 (6.8) 127 (96.9) 49 (37.4) 11 (8.4) 128 (97.0) 54 (40.9) 8 (6.1)
20 YEARS ON THE WEB

Any drug-related event 338 (88.9) 62 (16.3) 4 (1.1) 105 (80.2) 25 (19.1) 5 (3.8) 104 (78.8) 15 (11.4) 0
Gastrointestinal disorders
Diarrhea 146 (38.4) 16 (4.2) 1 (0.3) 43 (32.8) 7 (5.3) 0 26 (19.7) 1 (0.8) 0
Nausea 129 (33.9) 5 (1.3) 1 (0.3) 46 (35.1) 3 (2.3) 0 52 (39.4) 3 (2.3) 0
Constipation 81 (21.3) 3 (0.8) 0 27 (20.6) 3 (2.3) 0 34 (25.8) 1 (0.8) 0
Vomiting 75 (19.7) 6 (1.6) 1 (0.3) 31 (23.7) 3 (2.3) 0 29 (22.0) 3 (2.3) 0
Abdominal pain 67 (17.6) 6 (1.6) 0 20 (15.3) 2 (1.5) 0 22 (16.7) 6 (4.5) 1 (0.8)
The

Other
Fatigue 137 (36.1) 19 (5.0) 0 55 (42.0) 9 (6.9) 0 41 (31.1) 4 (3.0) 0
Decreased appetite 88 (23.2) 5 (1.3) 1 (0.3) 35 (26.7) 2 (1.5) 0 29 (22.0) 3 (2.3) 1 (0.8)
Pyrexia 78 (20.5) 2 (0.5) 0 16 (12.2) 0 0 23 (17.4) 2 (1.5) 0

39
Headache 65 (17.1) 4 (1.1) 0 19 (14.5) 3 (2.3) 0 19 (14.4) 3 (2.3) 0

n engl j med 363;8

Cough 55 (14.5) 1 (0.3) 0 21 (16.0) 0 0 18 (13.6) 0 0
Dyspnea 46 (12.1) 12 (3.2) 2 (0.5) 19 (14.5) 4 (3.1) 1 (0.8) 25 (18.9) 6 (4.5) 0

Back to Table of Contents

Anemia 41 (10.8) 11 (2.9) 0 15 (11.5) 4 (3.1) 0 23 (17.4) 11 (8.3) 0

nejm.org
Any immune-related event 221 (58.2) 37 (9.7) 2 (0.5) 80 (61.1) 16 (12.2) 3 (2.3) 42 (31.8) 4 (3.0) 0
n e w e ng l a n d j o u r na l

Dermatologic 152 (40.0) 8 (2.1) 1 (0.3) 57 (43.5) 2 (1.5) 0 22 (16.7) 0 0

of

Pruritus 67 (17.6) 1 (0.3) 0 32 (24.4) 0 0 14 (10.6) 0 0

Rash 67 (17.6) 5 (1.3) 0 25 (19.1) 1 (0.8) 0 6 (4.5) 0 0

august 19, 2010

Vitiligo 14 (3.7) 0 0 3 (2.3) 0 0 1 (0.8) 0 0
Gastrointestinal 122 (32.1) 20 (5.3) 2 (0.5) 38 (29.0) 10 (7.6) 0 19 (14.4) 1 (0.8) 0
m e dic i n e

Diarrhea 115 (30.3) 14 (3.7) 0 36 (27.5) 6 (4.6) 0 18 (13.6) 1 (0.8) 0

Colitis 20 (5.3) 11 (2.9) 1 (0.3) 10 (7.6) 7 (5.3) 0 1 (0.8) 0 0
Endocrine 15 (3.9) 4 (1.1) 0 10 (7.6) 3 (2.3) 2 (1.5) 2 (1.5) 0 0
Hypothyroidism 6 (1.6) 1 (0.3) 0 2 (1.5) 0 0 2 (1.5) 0 0
Hypopituitarism 3 (0.8) 2 (0.5) 0 3 (2.3) 1 (0.8) 1 (0.8) 0 0 0
Hypophysitis 2 (0.5) 2 (0.5) 0 2 (1.5) 2 (1.5) 0 0 0 0
Adrenal insufficiency 3 (0.8) 2 (0.5) 0 2 (1.5) 0 0 0 0 0
nejm.org
 Ipilimumab for metastatic melanoma
20 YEARS ON THE WEB nejm.org

The long-term effect of ipilimumab in our study

related to the study drug were associated with immune-related adverse events: 5 in the ipilimumab-plus-gp100 group (1 patient had grade 3 colitis and septicemia; 3 patients had bow-

were not associated with immune-related adverse events included deaths from sepsis, myelofibrosis, and acute respiratory distress syndrome (3 patients in the ipilimumab-plus-gp100
el perforationinflammatory colitis, bowel perforation, or multiorgan failureperitonitis; and 1 patient had GuillainBarr syndrome, which is considered to be consistent with a neuro-
Patients could have more than one adverse event. Included are all patients who received at least one dose of a study drug (643 patients). A total of 14 deaths (2.2%) were determined

logic immune-related adverse event) and 2 in the ipilimumab-alone group (1 patient had colic bowel perforation and the other had liver failure). Deaths related to the study drug that
by the investigators to be related to the study drug (8 in the ipilimumab-plus-gp100 group, 4 in the ipilimumab-alone group, and 2 in the gp100-alone group). Seven of the 14 deaths
is shown by survival analyses at late time points,

* The adverse events listed here were reported in at least 15% of patients. The most common immune-related adverse events and those of particular clinical relevance are also listed.

group); severe infectionrenal failureseptic shock, and vascular leak syndrome (2 patients in the ipilimumab-alone group), and cachexia and septic shock (2 patients in the gp100-
which showed 1-year and 2-year survival rates of

0
0
0
0
0
0
0
45.6% and 23.5%, respectively. In recent, ran-
domized, phase 3 trials involving patients with

3 (2.3)

1 (0.8)
unresectable stage III or IV melanoma who had
received previous treatment, 1-year survival rates
0
0

0
0
0
were reported to be 22% to 38% with various
treatment regimens.35,36 The median overall sur-
6 (4.5)
3 (2.3)
2 (1.5)

3 (2.3)
vival in these studies ranged from 5.9 to 9.7
months. Neither these nor other randomized,
0
0

0
controlled trials had shown a significant im-
provement in overall survival.
The adverse-event profile of ipilimumab in
this study is consistent with that reported in
phase 2 trials,1517 with the majority of adverse
1 (0.8)

1 (0.8)

events being immune-related and consistent

0

0
0
0
0

with the proposed mechanism of action of ipi-

limumab.1114 As shown in phase 2 studies,
2 (1.5)

prompt medical attention and early administra-

0
0
0
0
0
0

tion of corticosteroids are critical to the man-

agement of immune-related adverse events.1517
1 (0.8)
2 (1.5)
5 (3.8)
2 (1.5)
1 (0.8)
1 (0.8)
6 (4.6)

Management guidelines (algorithms) for immune-

related adverse events involve close patient follow-
up and the administration of high-dose systemic
corticosteroids which were used as necessary
in our study for grade 3 or 4 events.37,38
In conclusion, this randomized, controlled
trial showed that there was a significant improve-
ment in overall survival among patients with
0
0
0
0
0
0
0

metastatic melanoma. In some patients, side ef-

fects can be life-threatening and may be treat-
4 (1.1)
2 (0.5)
1 (0.3)
1 (0.3)
5 (1.3)

ment-limiting. Reinduction with ipilimumab at

the time of disease progression can result in fur-
0
0

ther clinical benefit. Overall, our findings suggest

that the T-cell potentiator ipilimumab may be use-
2 (0.5)

8 (2.1)
3 (0.8)
4 (1.1)
2 (0.5)
12 (3.2)

ful as a treatment for patients with metastatic

0

melanoma whose disease progressed while they

were receiving one or more previous therapies.
Increase in aspartate aminotransferase
Decrease in serum corticotropin level

Increase in alanine aminotransferase

Supported by Medarex and Bristol-Myers Squibb.

Increase in serum thyrotropin level

All study sites and institutions received funding from Med-

arex or Bristol-Myers Squibb to cover the expenses of the inves-
tigators for undertaking this trial.
Dr. Hodi reports receiving consulting fees from Bristol-Myers
SquibbMedarex, Novartis, and Genentech; Dr. ODay, receiving
consulting fees, grants, honoraria, and fees for participation in
speakers bureaus from Bristol-Myers Squibb; Dr. McDermott,
receiving consulting fees from Bristol-Myers SquibbMedarex;
Dr. Gonzalez, receiving honoraria from Bristol-Myers Squibb; Dr.
Hepatitis

Schadendorf, serving on a board for and receiving consulting

fees, fees for expert testimony, and fees for participation in speak-
alone group).
Hepatic

ers bureaus from Bristol-Myers Squibb; Dr. van den Eertwegh,

Other

receiving consulting fees from and serving on a board for Bristol-

Myers Squibb; Dr. Lutzky, receiving consulting fees and honoraria
from Bristol-Myers Squibb; Dr. Lorigan, receiving consulting fees
from Bristol-Myers Squibb; Dr. Hogg, serving on a board for

n engl j med 363;8 nejm.org august 19, 2010 721

40
Back to Table of Contents
 20 YEARS ON THE WEB The n e w e ng l a n d j o u r na l of m e dic i n e nejm.org

Bristol-Myers Squibb (pending); Dr. Ottensmeier, receiving hono-
raria and grant funding from Bristol-Myers Squibb; Dr. Lebb,
serving on a board for Bristol-Myers Squibb; Dr. Wolchok, serving
on a board for Bristol-Myers Squibb; Dr. J.S. Weber, receiving
consulting fees from Bristol-Myers Squibb; Drs. Tian, Yellin, and
Nichol being former employees of Medarex; Dr. Hoos being cur-
rently employed by Bristol-Myers Squibb with stock or stock op-
tions; and Dr. Urba, receiving consulting fees from Bristol-Myers
SquibbMedarex. No other potential conflicts of interest relevant
to this article were reported.
Disclosure forms provided by the authors are available with
the full text of this article at NEJM.org
We thank the patients who volunteered to participate in this
study and staff members at the study sites who cared for them;
the members of the data and safety monitoring committee; and
representatives of the sponsors who were involved in data collec-
tion and analyses (in particular, Tai-Tsang Chen, Xiaoping Zhu,
Marianne Messina, and Helena Brett-Smith). Editorial and writ-
ing assistance was provided by Ward A. Pedersen of StemScien-
tific, funded by Bristol-Myers Squibb.

Appendix
The authors affiliations are as follows: The DanaFarber Cancer Institute (F.S.H.) and Beth Israel Deaconess Medical Center
(D.F.M.) both in Boston; the Angeles Clinic and Research Institute, Los Angeles (S.J.O.); St. Marys Medical Center, San Fran-
cisco (R.W.W.); Vanderbilt University Medical Center, Nashville (J.A.S.); Netherlands Cancer Institute (J.B.H.) and VU University
Medical Center (A.J.M.E.) both in Amsterdam; University of Colorado Cancer Center, Aurora (R.G.); Institut Gustave Roussy,
Villejuif, France (C.R.); University Hospital Essen, Essen (D.S., J.M.V.), German Cancer Research Center, University of Mannheim,
Mannheim (J.C.H.), and Technical University Munich, Munich (C.P.) all in Germany; Huntsman Cancer Institute, Salt Lake City
(W.A.); Mount Sinai Comprehensive Cancer Center, Miami (J.L.); Christie Hospital NHS Trust, Manchester (P.L.), and Southampton
University Hospitals, Southampton (C.H.O.) both in the United Kingdom; Washington University School of Medicine, St. Louis
(G.P.L.); Princess Margaret Hospital, Toronto (D.H., I.Q.); Saint Louis Hospital, Paris (C.L.); Loyola University Medical Center,
Maywood, IL (J.I.C.); Memorial Sloan-Kettering Cancer Center, New York (J.D.W.); H. Lee Moffitt Cancer Center, Tampa, FL
(J.S.W.); Medarex, Bloomsbury, NJ (J.T., M.J.Y., G.M.N.); Bristol-Myers Squibb, Wallingford, CT (A.H.); and the Earle A. Chiles
Research Institute, Portland, OR (W.J.U.).
In addition to the authors, the following investigators (listed by country in alphabetical order) participated in the study: Argentina:
M. Chacn, L. Koliren, G.L. Lerzo, R.L. Santos all in Buenos Aires; Belgium: A. Awada (Brussels), V. Cocquyt (Ghent), J. Kerger
(Yvoir), J. Thomas (Leuven), T. Velu (Brussels); Brazil: C. Barrios (Porto Alegre), C. Dzik (So Paulo), M. Federico (So Paulo), J. Ho-
hmann (Barretos), M. Liberrati (Londrina), A. Lima (Santo Andr), G. Schwartsmann (Porto Alegre), J. Segalla (Ja); Canada: T. Baetz
(Kingston, ON), T. Cheng (Calgary, AB), W. Miller (Montreal), S. Rorke (St. Johns, NL), S. Verma (Ottawa), R. Wong (Winnipeg, MB);
Chile: H. Harbst (Santiago), P. Gonzalez-Mella (Via del Mar), P. Salman (Santiago); France: F. Cambazard (Saint-Etienne), O. Dereure
(Montpellier), B. Dreno (Nantes), L. Geoffrois (Vandoeuvre-ls-Nancy), J-J. Grob (Marseille), T. Lesimple (Dunkerque), S. Ngrier
(Lyon), N. Penel (Lille), A. Thyss (Nice); Germany: J.C. Becker (Wrzburg), C. Garbe (Tbingen), S. Grabbe (Molnz), U. Keilholz (Ber-
lin), C. Loquai (Mainz), H. Naeher (Heidelberg), G. Shuler (Erlangen), U. Trefzer (Berlin), J. Welzel (Augsburg); Hungary: Z. Karolyi
(Miskolc); Netherlands: R.L.H. Jansen (Maastricht); South Africa: G.L. Cohen (Pretoria), J.I. Raats (Panorama), D.A. Vorobiof (Morning-
side); Switzerland: R. Dummer (Zurich), O. Michielin (Lausanne); United Kingdom: J. Barber (Cardiff), S. Danson (Sheffield), M. Gore
(London), S. Houston (Surrey), C.G. Kelly (Newcastle-upon-Tyne), M. Middleton (Oxford), P.M. Patel (Nottingham), E. Rankin (Dundee,
Scotland); United States: M. Adler (Vista, CA), T. Amatruda (Robbinsdale, MN), A. Amin (Charlotte, NC), C. Anderson (Columbia, MO),
L. Blakely (Memphis, TN), E. Borden (Cleveland), S. Burdette-Radoux (Burlington, VT), R. Chapman (Detroit), J. Chesney (Louisville,
KY), A. Cohn (Denver), F.A. Collichio (Chapel Hill, NC), G. Daniels (La Jolla, CA), J. Drabick (Hershey, PA), J.A. Figueroa (Lubbock, TX),
J. Fleagle (Boulder, CO), J. Goydos (New Brunswick, NJ), N. Haas (Philadelphia), E. Hersh (Tucson, AZ), H.L. Kaufman (New York), K.D.
Khan (Indianapolis), A. Khurshid (Arlington, TX), J.M. Kirkwood (Pittsburgh), J.J. Kirshner (East Syracuse, NY), H. Kluger (New Haven,
CT), D. Lawrence (Boston), D. Lawson (Atlanta), P.D. Leming (Cincinnati), K. Margolin (Seattle), M. Mastrangelo (Philadelphia), B.
Mirtsching (Dallas), W. Paroly (San Diego, CA), A.L. Pecora (Hackensack, NJ), D. Pham (Jacksonville, FL), R. Rangineni (St. Joseph, MO),
N. Rothschild (West Palm Beach, FL), W.E. Samlowski (Las Vegas), D. Schwartzentruber (Goshen, IN), M. Scola (Morristown, NJ), W.H.
Sharfman (Lutherville, MD), J.J. Stephenson (Greenville, SC), N.S. Tchekmedyian (Long Beach, CA), J. Wade (Decatur, IL), M. Wax (Berke-
ley Heights, NJ), A. Weeks (Collierville, TN), J.L. Zapas (Baltimore).

References

1. American Cancer Society. Cancer facts
& figures 2009. (Accessed June 4, 2010, at
http://www.cancer.org/downloads/STT/
500809web.pdf.)
2. Gray-Schopfer V, Wellbrock C, Marais
R. Melanoma biology and new targeted
therapy. Nature 2007;445:851-7.
3. Lens MB, Dawes M. Global perspec-
tives of contemporary epidemiological
trends of cutaneous malignant melano-
ma. Br J Dermatol 2004;150:179-85.
4. World Health Organization. Skin can-
cers. (Accessed June 4, 2010, at http://www
.who.int/uv/faq/skincancer/en/index1.html.)
5. Tsao H, Atkins MB, Sober AJ. Man-
agement of cutaneous melanoma. N Engl J
Med 2004;351:998-1012. [Erratum, N Engl
J Med 2004;351:2461.]
6. Agarwala SS. Current systemic therapy
for metastatic melanoma. Expert Rev An-
ticancer Ther 2009;9:587-95.
7. Eggermont AM, Kirkwood JM. Re-
evaluating the role of dacarbazine in meta-
static melanoma: what have we learned in
30 years? Eur J Cancer 2004;40:1825-36.
8. Petrella T, Quirt I, Verma S, Haynes
AE, Charette M, Bak K. Single-agent inter-
leukin-2 in the treatment of metastatic
melanoma: a systematic review. Cancer
Treat Rev 2007;33:484-96.
9. Trinh VA. Current management of
metastatic melanoma. Am J Health Syst
Pharm 2008;65:Suppl 9:S3-S8.
10. Melero I, Hervas-Stubbs S, Glennie M,
Pardoll DM, Chen L. Immunostimulatory
monoclonal antibodies for cancer thera-
py. Nat Rev Cancer 2007;7:95-106.
11. ODay SJ, Hamid O, Urba WJ. Target-
ing cytotoxic T-lymphocyte antigen-4
(CTLA-4): a novel strategy for the treat-
ment of melanoma and other malignan-
cies. Cancer 2007;110:2614-27.
12. Fong L, Small EJ. Anti-cytotoxic T-lym-
phocyte antigen-4 antibody: the first in an
emerging class of immunomodulatory
antibodies for cancer treatment. J Clin
Oncol 2008;26:5275-83.
13. Robert C, Ghiringhelli F. What is the
role of cytotoxic T lymphocyte-associated
antigen 4 blockade in patients with meta-
static melanoma? Oncologist 2009;14:
848-61.
14. Weber J. Ipilimumab: controversies in
its development, utility and autoimmune
adverse events. Cancer Immunol Immu-
nother 2009;58:823-30.
15. Weber J, Thompson JA, Hamid O, et al.

722 n engl j med 363;8 nejm.org august 19, 2010

41
Back to Table of Contents
 Ipilimumab for metastatic melanoma
20 YEARS ON THE WEB nejm.org

A randomized, double-blind, placebo-
controlled, phase II study comparing the
tolerability and efficacy of ipilimumab
administered with or without prophylac-
tic budesonide in patients with unresect-
able stage III or IV melanoma. Clin Can-
cer Res 2009;15:5591-8.
16. Wolchok JD, Neyns B, Linette G, et al.
Ipilimumab monotherapy in patients with
pretreated advanced melanoma: a random-
ised, double-blind, multicentre, phase 2,
dose-ranging study. Lancet Oncol 2010;
11:155-64.
17. ODay SJ, Maio M, Chiarion-Sileni V,
et al. Efficacy and safety of ipilimumab
monotherapy in patients with previously
treated, advanced melanoma: a multicenter,
single-arm phase II study. Ann Oncol
2010 February 10 (Epub ahead of print).
18. Maker AV, Phan GQ, Attia P, et al.
Tumor regression and autoimmunity in
patients treated with cytotoxic T lympho-
cyteassociated antigen 4 blockade and
interleukin 2: a phase I/II study. Ann Surg
Oncol 2005;12:1005-16.
19. Agarwala SS. Novel immunotherapies
as potential therapeutic partners for tradi-
tional or targeted agents: cytotoxic T-lym-
phocyte antigen-4 blockade in advanced
melanoma. Melanoma Res 2010;20:1-10.
20. Attia P, Phan GQ, Maker AV, et al.
Autoimmunity correlates with tumor re-
gression in patients with metastatic mela-
noma treated with anticytotoxic T-lym-
phocyte antigen-4. J Clin Oncol 2005;23:
6043-53.
21. Downey SG, Klapper JA, Smith FO, et
al. Prognostic factors related to clinical
response in patients with metastatic mela-
noma treated by CTL-associated antigen-4
blockade. Clin Cancer Res 2007;13:6681-8.
22. Rosenberg SA, Yang JC, Restifo NP.
Cancer immunotherapy: moving beyond
current vaccines. Nat Med 2004;10:909-15.
23. Schwartzentruber DJ, Lawson D,
Richards J, et al. A phase III multi-institu-
tional randomized study of immunization
with the gp100:209217(210M) peptide
followed by high-dose IL-2 compared with
high-dose IL-2 alone in patients with met-
astatic melanoma. J Clin Oncol 2009;27:
Suppl:463s. abstract.
24. Oken MM, Creech RH, Tormey DC, et
al. Toxicity and response criteria of the
Eastern Cooperative Oncology Group. Am
J Clin Oncol 1982;5:649-55.
25. ClinicalTrials.gov. Dacarbazine and
ipilimumab vs. dacarbazine with placebo
in untreated unresectable stage III or IV
melanoma. (Accessed June 4, 2010, at http://
www.clinicaltrials.gov/ct/show/
NCT00324155.)
26. James K, Eisenhauer E, Christian M,
et al. Measuring response in solid tumors:
unidimensional versus bidimensional mea-
surement. J Natl Cancer Inst 1999;91:
523-8.
27. Korn EL, Liu PY, Lee SJ, et al. Meta-
analysis of phase II cooperative group tri-
als in metastatic stage IV melanoma to
determine progression-free and overall
survival benchmarks for future phase II
trials. J Clin Oncol 2008;26:527-34.
28. Bedikian AY, Johnson MM, Warneke
CL, et al. Prognostic factors that deter-
mine the long-term survival of patients
with unresectable metastatic melanoma.
Cancer Invest 2008;26:624-33.
29. Barrow C, Browning J, MacGregor D,
et al. Tumor antigen expression in mela-
noma varies according to antigen and
stage. Clin Cancer Res 2006;12:764-71.
30. Morton DL, Mozzillo N, Thompson
JF, et al. An international, randomized,
phase III trial of bacillus Calmette-Guerin
(BCG) plus allogeneic melanoma vaccine
(MCV) or placebo after complete resec-
tion of melanoma metastatic to regional
or distant sites. J Clin Oncol 2007;25:
Suppl:474s. abstract.
31. Eggermont AM, Suciu S, Ruka W, et
al. EORTC 18961: post-operative adjuvant
ganglioside GM2-KLH21 vaccination treat-
ment vs observation in stage II (T3-
T4N0M0) melanoma: 2nd interim analy-
sis led to an early disclosure of the results.
J Clin Oncol 2008;26:Suppl:484s. abstract.
32. Testori A, Richards J, Whitman E, et al.
Phase III comparison of vitespen, an autol-
ogous tumor-derived heat shock protein
gp96 peptide complex vaccine, with phy-
sicians choice of treatment for stage IV
melanoma. J Clin Oncol 2008;26:955-62.
33. Sosman JA, Unger JM, Liu PY, et al.
Adjuvant immunotherapy of resected, inter-
mediate-thickness, node-negative mela-
noma with an allogeneic tumor vaccine:
impact of HLA class I antigen expression
on outcome. J Clin Oncol 2002;20:2067-
75.
34. Wolchok JD, Hoos A, ODay S, et al.
Guidelines for the evaluation of immune
therapy activity in solid tumors: immune-
related response criteria. Clin Cancer Res
2009;15:7412-20.
35. Eisen T, Trefzer U, Hamilton A, et al.
Results of a multicenter, randomized,
double-blind phase 2/3 study of lenalido-
mide in the treatment of pretreated re-
lapsed or refractory metastatic malignant
melanoma. Cancer 2010;116:146-54.
36. Hauschild A, Agarwala SS, Trefzer U,
et al. Results of a phase III, randomized,
placebo-controlled study of sorafenib in
combination with carboplatin and pacli-
taxel as second-line treatment in patients
with unresectable stage III or stage IV
melanoma. J Clin Oncol 2009;27:2823-30.
37. Weber J. Anti-CTLA-4 antibody ipilim-
umab: case studies of clinical response
and immune-related adverse events. On-
cologist 2007;12:864-72.
38. Lin R, Yellin MJ, Lowy I, Safferman A,
Chin K, Ibrahim R. An analysis of the ef-
fectiveness of specific guidelines for the
management of ipilimumab-mediated di-
arrhea/colitis: prevention of gastrointesti-
nal perforation and/or colectomy. J Clin
Oncol 2008;26:Suppl:497s. abstract.
Copyright 2010 Massachusetts Medical Society.

full text of all journal articles on the world wide web

Access to the complete contents of the Journal on the Internet is free to all subscribers. To use this Web site, subscribers should
go to the Journals home page (NEJM.org) and register by entering their names and subscriber numbers as they appear on their
mailing labels. After this one-time registration, subscribers can use their passwords to log on for electronic access to the entire
Journal from any computer that is connected to the Internet. Features include a library of all issues since January 1993 and
abstracts since January 1975, a full-text search capacity, and a personal archive for saving articles and search results of interest.
All articles can be printed in a format that is virtually identical to that of the typeset pages. Beginning 6 months after
publication, the full text of all Original Articles and Special Articles is available free to nonsubscribers.

n engl j med 363;8 nejm.org august 19, 2010 723

42
Back to Table of Contents
 T h e n e w e ng l a n d j o u r na l
20 YEARS ON THE WEB of m e dic i n e nejm.org

Brief Report

Zika Virus Associated with Microcephaly

Jernej Mlakar, M.D., Misa Korva, Ph.D., Nataa Tul, M.D., Ph.D.,
Mara Popovi, M.D., Ph.D., Mateja Poljak-Prijatelj, Ph.D., Jerica Mraz, M.Sc.,
Marko Kolenc, M.Sc., Katarina Resman Rus, M.Sc., Tina Vesnaver Vipotnik, M.D.,
Vesna Fabjan Voduek, M.D., Alenka Vizjak, Ph.D., Joe Piem, M.D., Ph.D.,
Miroslav Petrovec, M.D., Ph.D., and Tatjana Avi upanc, Ph.D.

Sum m a r y

A widespread epidemic of Zika virus (ZIKV) infection was reported in 2015 in

South and Central America and the Caribbean. A major concern associated with
this infection is the apparent increased incidence of microcephaly in fetuses born
to mothers infected with ZIKV. In this report, we describe the case of an expectant
mother who had a febrile illness with rash at the end of the first trimester of
pregnancy while she was living in Brazil. Ultrasonography performed at 29 weeks
of gestation revealed microcephaly with calcifications in the fetal brain and pla-
centa. After the mother requested termination of the pregnancy, a fetal autopsy
was performed. Micrencephaly (an abnormally small brain) was observed, with
almost complete agyria, hydrocephalus, and multifocal dystrophic calcifications in
the cortex and subcortical white matter, with associated cortical displacement and
mild focal inflammation. ZIKV was found in the fetal brain tissue on reverse-
transcriptasepolymerase-chain-reaction (RT-PCR) assay, with consistent findings
on electron microscopy. The complete genome of ZIKV was recovered from the
fetal brain.

Z
IKV, an emerging mosquito-borne flavivirus, was initially iso- From the Institute of Pathology, Faculty
lated from a rhesus monkey in the Zika forest in Uganda in 1947.1 It is of Medicine (J. Mlakar, M. Popovi, J. Mraz,
A.V., J.P.), and the Institute of Microbiol-
transmitted by various species of aedes mosquitoes. After the first human ogy and Immunology, Faculty of Medicine
ZIKV infection, sporadic cases were reported in Southeast Asia and sub-Saharan (M. Korva, M.P.-P., M. Kolenc, K.R.R.,
Africa.2 ZIKV was responsible for the outbreak in Yap Island of Micronesia in 2007 M. Petrovec, T.A.Z.), University of Ljubljana,
and the Department of Perinatology, Di-
and for major epidemics in French Polynesia, New Caledonia, the Cook Islands, vision of Gynecology and Obstetrics
and Easter Island in 2013 and 2014.3,4 In 2015, there was a dramatic increase in (N.T., V.F.V.), and the Institute of Radiol-
reports of ZIKV infection in the Americas. Brazil is the most affected country, with ogy (T.V.V.), University Medical Center
Ljubljana all in Ljubljana, Slovenia. Ad-
preliminary estimates of 440,000 to 1.3 million cases of autochthonous ZIKV dress reprint requests to Dr. Avi upanc
infection reported through December 2015.5 at the Institute of Microbiology and Im-
The classic clinical picture of ZIKV infection resembles that of dengue fever and munology, Faculty of Medicine, Univer-
sity of Ljubljana, Zaloka 4, Ljubljana 1000,
chikungunya and is manifested by fever, headache, arthralgia, myalgia, and macu- Slovenia, or at tatjana.avsic@mf.uni-lj.si.
lopapular rash, a complex of symptoms that hampers differential diagnosis.
This article was published on February 10,
Although the disease is self-limiting, cases of neurologic manifestations and the 2016, at NEJM.org.
GuillainBarr syndrome were described in French Polynesia and in Brazil during
N Engl J Med 2016;374:951-8.
ZIKV epidemics.5,6 Recent reports from the Ministry of Health of Brazil suggest DOI: 10.1056/NEJMoa1600651
that cases of microcephaly have increased by a factor of approximately 20 among Copyright 2016 Massachusetts Medical Society.

newborns in the northeast region of the country, which indicates a possible asso-
ciation between ZIKV infection in pregnancy and fetal malformations.5
We present a case of vertical transmission of ZIKV in a woman who was prob-

n engl j med 374;10 nejm.org March 10, 2016 951

43
Back to Table of Contents
 The n e w e ng l a n d j o u r na l
20 YEARS ON THE WEB
ably infected with ZIKV in northeastern Brazil at
the end of the first trimester of pregnancy. Our
discussion includes details of fetal imaging and
pathological and virologic analyses.
C a se R ep or t
In mid-October 2015, a 25-year-old previously
healthy European woman came to the Depart-
ment of Perinatology at the University Medical
Center in Ljubljana, Slovenia, because of assumed
fetal anomalies. Since December 2013, she had
lived and worked as a volunteer in Natal, the
capital of Rio Grande do Norte state. She had
become pregnant at the end of February 2015.
During the 13th week of gestation, she had be-
come ill with high fever, which was followed by
severe musculoskeletal and retroocular pain and
an itching, generalized maculopapular rash. Since
there was a ZIKV epidemic in the community,
infection with the virus was suspected, but no
virologic diagnostic testing was performed.
Ultrasonography that was performed at 14 and
20 weeks of gestation showed normal fetal
growth and anatomy.
The patient returned to Europe at 28 weeks of
gestation. Ultrasonographic examination that
was performed at 29 weeks of gestation showed
the first signs of fetal anomalies, and she was
referred to the Department of Perinatology. At
that time, she also noticed reduced fetal move-
ments. Ultrasonography that was performed at
32 weeks of gestation confirmed intrauterine
growth retardation (estimated third percentile
of fetal weight) with normal amniotic fluid, a
placenta measuring 3.5 cm in thickness (normal
size) with numerous calcifications, a head cir-
cumference below the second percentile for gesta-
tion (microcephaly), moderate ventriculomegaly,
and a transcerebellar diameter below the second
percentile. Brain structures were blurred, and
there were numerous calcifications in various
parts of the brain (Fig. 1A and 1B). There were
no other obvious fetal structural abnormalities.
Fetal, umbilical, and uterine blood flows were
normal on Doppler ultrasonography.
The clinical presentation raised suspicion of
fetal viral infection. Because of severe brain dis-
ease and microcephaly, the fetus was given a
poor prognosis for neonatal health. The mother
requested that the pregnancy be terminated, and
952 n engl j med 374;10
Back to Table of Contents
of m e dic i n e
nejm.org
the procedure was subsequently approved by
national and hospital ethics committees. Medi-
cal termination of the pregnancy was performed
at 32 weeks of gestation. At the delivery, the only
morphologic anomaly was the prominent micro-
cephaly. Genetic consultation that included a de-
tailed maternal family history revealed no suspi-
cion of genetic syndromes or diseases. An autopsy
was performed, as is mandatory in all cases of
termination of pregnancy. The mother provided
written informed consent for the publication of
this case report.
Me thods
Autopsy and Central Nervous System (CNS)
Examination
An autopsy of the fetus and placenta was per-
formed 3 days after termination of the pregnancy,
with an extensive sampling of all organs, pla-
centa, and umbilical cord. Samples were fixed in
10% buffered formalin and embedded in paraf-
fin. Fresh tissue samples were collected for micro-
biologic investigations. Brain and spinal cord were
fixed in 27% buffered formalin for 3 weeks, after
which a neuropathological examination was per-
formed with extensive sampling of the brain and
spinal cord. Sections of all tissue samples were
stained with hematoxylin and eosin. Immunos-
taining for glial fibrillary acid protein, neurofila-
ment, human leukocyte antigen DR (HLA-DR),
CD3 (to highlight T cells), and CD20 (to high-
light B cells) was performed on representative
CNS samples.
Electron Microscopy
Tissue was collected from formalin-fixed brain
and underwent fixation in 1% osmium tetroxide
and dehydration in increasing concentrations of
ethanol. The sample was then embedded in Epon.
Semithin sections (1.4 m) were made, stained
with Azur II, and analyzed by means of light
microscopy. Ultrathin sections (60 nm) were
stained with uranyl acetate and lead citrate. In
addition, a small piece of brain (5 mm3) was
homogenized in buffer. The suspension was then
cleared by low-speed centrifugation, and the ob-
tained supernatant was ultracentrifuged directly
onto an electron microscopic grid with the use
of an Airfuge (Beckman Coulter). Negative stain-
ing was performed with 1% phosphotungstic
nejm.org March 10, 2016
44
 Brief Report
20 YEARS ON THE WEB nejm.org

A B

C D

* *
*
*
*
*

Figure 1. Prenatal Ultrasonographic Images and Photographs of Coronal Slices of Brain.

Panel A shows numerous calcifications in various parts of the brain (some marked with arrows) and the dilated
occipital horn of the lateral ventricle (Vp, marked with a measurement bar) as seen on transverse ultrasonography.
Panel B shows numerous calcifications in the placenta. Panel C shows multifocal cortical and subcortical white cal-
cifications (arrows) and almost complete loss of gyration of the cortex. The basal ganglia are developed but poorly
delineated (black asterisks), and the sylvian fissures are widely open on both sides (arrowheads on the left). The
third ventricle is not dilated (white asterisk). Panel D shows dilated body of the lateral ventricles (white arrowheads);
the left is collapsed. Temporal horns of the lateral ventricles (black arrowheads) are also dilated. The thalami (black
asterisks) and the left hippocampus (white asterisk) are well developed, whereas the contralateral structure is not
recognizable owing to autolysis.

acid. Imaging of the ultrathin sections and brain Microbiologic Investigation

homogenate was performed with the use of a
120-kV JEM-1400Plus transmission electron mi-
croscope (JEOL).
Indirect Immunofluorescence
Paraffin-embedded sections of the fetal brain
tissue and brain tissue of an autopsied man as a
negative control were incubated with serum ob-
tained from the mother of the fetus (dilution,
1:10), followed by antihuman IgG antibodies
labeled with fluorescein isothiocyanate (FITC)
(dilution, 1:50). In addition, fetal brain tissue
was incubated with a serum obtained from a
healthy blood donor, as well as with FITC-labeled
antihuman IgG antibodies only.
RNA was extracted from 10 mg of the placenta,
lungs, heart, skin, spleen, thymus, liver, kidneys,
and cerebral cortex with the use of a TRIzol Plus
RNA purification kit (Thermo Fisher Scientific).
Real-time RT-PCR for the detection of ZIKV RNA
(NS5) and one-step RT-PCR for the detection of
the envelope-protein coding region (360 bp) were
performed as described previously.7,8 In addition,
next-generation sequencing was performed in
samples of fetal brain tissue with the use of Ion
Torrent (Thermo Fisher Scientific) and Geneious
software, version 9.0.6. Reads from both runs
were combined and mapped to the reference
sequence (ZIKV MR766; LC002520) with the use
of default measures. For phylogenetic analysis,

n engl j med 374;10 nejm.org March 10, 2016 953

45
Back to Table of Contents
 The n e w e ng l a n d j o u r na l of m e dic i n e
20 YEARS ON THE WEB nejm.org

complete-genome ZIKV sequences were used, and phages expressing HLA-DR were present through-
multiple sequence alignments (ClustalW) were out most of the cerebral gray and white matter
performed. A neighbor-joining phylogenetic tree (Fig. 2D). Scattered mild perivascular infiltrates
(GTR+G+I model) was constructed, with the use composed of T cells and some B cells were pres-
of the MEGA6 software system,9 to show the ent in the subcortical white matter (Fig. S1 in
phylogenetic relationships. The nucleotide se- the Supplementary Appendix). The cerebellum,
quence of ZIKV that was obtained in this study brain stem, and spinal cord showed neither in-
has been deposited in GenBank under accession flammation nor dystrophic calcifications. The
number KU527068. A detailed description of the brain stem and spinal cord showed Wallerian
molecular methods is provided in the Supple- degeneration of the long descending tracts, es-
mentary Appendix, available with the full text of pecially the lateral corticospinal tract, whereas
this article at NEJM.org. The results of compre- ascending dorsal columns were well preserved
hensive serologic analyses of maternal serum (Fig. 2E). Indirect immunofluorescence revealed
and a description of the molecular differential granular intracytoplasmic reaction in destroyed
diagnostic procedures used with fetal tissue neuronal structures, which pointed to a possible
samples are provided in Tables S1 and S2 in the location of the virus in neurons (Fig. 2F, and Fig.
Supplementary Appendix. All the authors vouch S1 in the Supplementary Appendix). Histologic
for the completeness and accuracy of the data examination of the placenta confirmed focal
and analyses presented. calcifications in villi and decidua, but no inflam-
mation was found. There were no relevant patho-
logical changes in other fetal organs or in the
R e sult s
umbilical cord or fetal membranes. Fetal karyo-
Autopsy and Neuropathological Findings typing with the use of microarray technology
The fetal body weight was 1470 g (5th percentile), showed a normal 46XY (male) profile.
the length 42 cm (10th percentile), and the head
circumference 26 cm (1st percentile). The only Electron Microscopy
external anomaly that was noted was micro- Although analysis of the ultrathin sections of
cephaly. The placenta weighed 200 g, resulting the brain showed poorly preserved brain tissue
in a placentalfetal weight ratio of 0.136 (<3rd with ruptured and lysed cells, clusters of dense
percentile). Macroscopic examination of the CNS virus-like particles of approximately 50 nm in
revealed micrencephaly with a whole-brain weight size were found in damaged cytoplasmic vesi-
of 84 g (4 SD below average), widely open sylvian cles. Groups of enveloped structures with a bright
fissures, and a small cerebellum and brain stem. interior were also detected. At the periphery of
Almost complete agyria and internal hydro- such groups, the remains of membranes could
cephalus of the lateral ventricles were observed. be seen. Negative staining of homogenized brain
There were numerous variable-sized calcifica- revealed spherical virus particles measuring
tions in the cortex and subcortical white matter 42 to 54 nm with morphologic characteristics
in the frontal, parietal, and occipital lobes. The consistent with viruses of the Flaviviridae fam-
subcortical nuclei were quite well developed (Fig. ily (Fig. 3).
1C and 1D). In spite of some autolysis, micro-
scopic examination revealed appropriate cyto- Microbiologic Investigation
architecture of the fetal brain. The most promi- Positive results for ZIKV were obtained on RT-PCR
nent histopathological features were multifocal assay only in the fetal brain sample, where
collections of filamentous, granular, and neuron- 6.5107 viral RNA copies per milligram of tissue
shaped calcifications in the cortex and subcorti- were detected. In addition, all autopsy samples
cal white matter with focal involvement of the were tested on PCR assay and were found to be
whole cortical ribbon, occasionally associated negative for other flaviviruses (dengue virus, yel-
with cortical displacement (Fig. 2A and 2B). Dif- low fever virus, West Nile virus, and tick-borne
fuse astrogliosis was present with focal astrocytic encephalitis virus), along with chikungunya
outburst into the subarachnoid space, mostly on virus, lymphocytic choriomeningitis, cytomegalo-
the convexity of the cerebral hemispheres (Fig. virus, rubella virus, varicellazoster virus, herpes
2C). Activated microglial cells and some macro- simplex virus, parvovirus B19, enteroviruses, and

954 n engl j med 374;10 nejm.org March 10, 2016

46
Back to Table of Contents
 Brief Report
20 YEARS ON THE WEB nejm.org

A B

C D

E F

* * *

Figure 2. Microscopic Analysis of Brain Tissue.

Panel A shows thickened leptomeninges (black arrowhead) and irregular cortical and subcortical calcifications (asterisks)
associated with cortical displacement (arrows), with preserved germinative matrix (white arrowhead); gyration is absent.
Panel B shows higher magnification of calcifications with filamentous structures (arrow), possibly representing encrust-
ed, damaged axons and dendrites, and oval and polygonal structures (arrowheads), possibly representing encrusted,
damaged neuronal-cell bodies (hematoxylin and eosin staining in Panels A and B). Panel C shows immunohistochemical
labeling of proliferated reactive astrocytes that extend into the subarachnoid space (asterisk) (glial fibrillary acid protein,
clone 6F2 [Dako]). Panel D shows immunohistochemical labeling of numerous activated microglial cells and macro-
phages in the cortex (full thickness marked with a line) and subcortical white matter (lower part of the figure). Nonspecif-
ic staining of the calcifications is present (arrow). Focal leptomeningeal infiltrates of macrophages are seen (arrowhead)
(HLA-DR, clone TAL 1B5 [Dako]). Panel E shows neurofilament immunohistochemical staining of axons in a cross-sec-
tion of the lumbar spinal cord with severe Wallerian degeneration of the lateral corticospinal tracts (black asterisks), mod-
erate involvement of other descending tracts (arrows), and well-preserved ascending tracts in the dorsal columns (white
asterisk) (neurofilament, clone 2F11 [Dako]). Panel F shows indirect immunofluorescence of fetal brain tissue, revealing a
green granular intracytoplasmic reaction (see also inset). The yellow signals adjacent to the green granules indicate auto-
fluorescence of lipofuscin, suggesting that viral particles are located in the cytoplasm of neurons.

n engl j med 374;10 nejm.org March 10, 2016 955

47
Back to Table of Contents
 The n e w e ng l a n d j o u r na l of m e dic i n e
20 YEARS ON THE WEB nejm.org

A B

1 m 100 nm

C D

100 nm 100 nm

Figure 3. Electron Microscopy of Ultrathin Sections of Fetal Brain and Staining of a Flavivirus-like Particle.
Panel A shows a damaged brain cell with a cluster of dense virions located in the disrupted endoplasmic reticulum.
Remains of membranes derived from different cellular compartments and filamentous structures are also seen.
A magnified view of the boxed area with virions clearly visible (arrows) is shown in Panel B. Panel C shows a group
of enveloped structures with a bright interior, presumably indicating viral replication (arrow). Panel D shows a nega-
tively stained viral particle with morphologic characteristics consistent with those of Flaviviridae viruses (arrow).

Toxoplasma gondii (Table S2 in the Supplementary
Appendix).
A complete ZIKV genome sequence (10,808
nucelotides) was recovered from brain tissue.
Phylogenetic analysis showed the highest iden-
tity (99.7%) with the ZIKV strain isolated from a
patient from French Polynesia in 2013 (KJ776791)
and ZIKV detected in Sao Paolo, Brazil, in 2015
(KU321639), followed by a strain isolated in Cam-
bodia in 2010 (JN860885, with 98.3% identity)
and with a strain from the outbreak in Microne-
sia in 2007 (EU545988, with 98% identity) (Fig. 4).
In the ZIKV polyprotein, 23 polymorphisms were
detected in comparison with the strain from
Micronesia and 5 polymorphisms in comparison
with the isolate from French Polynesia; three
amino acid changes were found in the NS1 re-
gion (K940E, T1027A, and M1143V), one in the
NS4B region (T2509I), and one in the FtsJ-like
methyltransferase region (M2634V).
Discussion
This case shows severe fetal brain injury associ-
ated with ZIKV infection with vertical transmis-
sion. Recently, ZIKV was found in amniotic fluid
of two fetuses that were found to have micro-
cephaly, which was consistent with intrauterine
transmission of the virus.10 Described cases are
similar to the case presented here and were
characterized by severely affected CNS and gross
intrauterine growth retardation. Calcifications
in the placenta and a low placentalfetal weight
ratio,11 which were seen in this case, indicate
potential damage to the placenta by the virus.
Among the few reports of teratogenic effects of

956 n engl j med 374;10 nejm.org March 10, 2016

48
Back to Table of Contents
 Brief Report
20 YEARS ON THE WEB nejm.org

KU365777 BeH818995
92
100 KU365780 BeH815744

99 KU365779 BeH819966

KU365778 BeH819015
99 99
KU312312 Suriname

KU321639 Sao Paulo

100
99
KU527068 Bahia, Brazil
93
KJ776791 2013 French Polynesia
100
JN860885 2010 Cambodia
100
EU545988 2007 Micronesia

HQ234499 1996 Malaysia

HQ234500 1968 Nigeria

LC002520 1947 Uganda

0.01

Figure 4. Phylogenetic Analysis of the Complete Genome of Zika Virus.

The evolutionary history was inferred by means of the neighbor-joining method under a GTR+G+I substitution mod-
el. The percentage of replicate trees in which the associated taxa clustered together in the bootstrap test (2000 rep-
licates) is shown next to the branches. The GenBank accession number, year of isolation, and country of origin are
indicated on the ZIKV branches for all strains except for those identified in 2015 and 2016. ZIKV strain Bahia, Brazil
(KU527068), was obtained in this study. The complete genome sequence was recovered from fetal brain tissue. The
0.01 scale bar denotes the genetic distance in nucleotide substitutions per site.

flaviviruses, investigators described the brain
and eyes as the main targets.12,13 No presence of
virus and no pathological changes were detected
in any other fetal organs apart from the brain,
which suggests a strong neurotropism of the
virus.
The localization of immunofluorescence sig-
nal and the morphologic appearance of the cal-
cifications, which resembled destroyed neuronal
structures, indicate a possible location of the
virus in neurons. The consequent damage might
cause arrested development of the cerebral cor-
tex at the embryonic age of approximately 20
weeks.14 The mechanism involved in the neurot-
ropism of ZIKV is currently not clear. The asso-
ciation between ZIKV infection and fetal brain
anomalies was also noted by findings on electron
microscopy that were consistent with ZIKV de-
tection in the fetal brain. Dense particles consis-
tent with ZIKV were seen in damaged endoplas-
mic reticulum. Groups of enveloped structures
with a bright interior resembling the remains of
replication complexes that are characteristic of
flaviviruses15,16 indicate viral replication in the
brain. The findings on electron microscopy sug-
gest a possible persistence of ZIKV in the fetal
brain, possibly because of the immunologically
secure milieu for the virus. The number of viral
copies that were detected in the fetal brain
were substantially higher than those reported in
the serum obtained from adult ZIKV-infected
patients17 but similar to those reported in semen
samples.18
The complete genome sequence of ZIKV that
was recovered in this study is consistent with the
observation that the present strain in Brazil has
emerged from the Asian lineage.19 The presence
of two major amino acid substitutions posi-
tioned in nonstructural proteins NS1 and NS4B
probably represents an accidental event or indi-
cates a process of eventual adaptation of the vi-
rus to a new environment. Further research is
needed to better understand the potential impli-
cations of these observations. It is likely that the
rapid spread of ZIKV around the globe will be a
strong impetus for collaborative research on the
biologic properties of the virus, particularly
since the risk of neurotropic and teratogenic vi-
rus infections places a high emotional and eco-
nomic burden on society.

n engl j med 374;10 nejm.org March 10, 2016 957

49
Back to Table of Contents
 Brief Report
20 YEARS ON THE WEB nejm.org

Disclosure forms provided by the authors are available with
the full text of this article at NEJM.org.
We thank the patient in this case for her willingness to pro-
vide detailed medical and immunologic data; Miha Juvan for
processing of brain photographs; Peter trafela for his assis-
tance with the neuropathological analyses; Martin Sagadin,
Tina Uri, Nataa Toplak, Simon Koren, and Andrej Steyer for
their assistance in virus detection, sequencing, and analysis of
next-generation sequencing data; Mateja Jelovek for her assis-
tance in comprehensive serologic investigations, and Luca
Lovrei and Marija Volk for their assistance in molecular karyo-
typing with microarray testing.

References
1. Dick GW, Kitchen SF, Haddow AJ.
Zika virus. I. Isolations and serological
specificity. Trans R Soc Trop Med Hyg
1952;46:509-20.
2. Hayes EB. Zika virus outside Africa.
Emerg Infect Dis 2009;15:1347-50.
3. Duffy MR, Chen TH, Hancock WT, et
al. Zika virus outbreak on Yap Island, Fed-
erated States of Micronesia. N Engl J Med
2009;360:2536-43.
4. Cao-Lormeau VM, Roche C, Teissier A,
et al. Zika virus, French Polynesia, South
Pacific, 2013. Emerg Infect Dis 2014;20:
1085-6.
5. Rapid risk assessment: Zika virus epi-
demic in the Americas: potential associa-
tion with microcephaly and Guillain-Barr
syndrome. Stockholm: European Centre
for Disease Prevention and Control, De-
cember 10, 2015 (http://ecdc.europa.eu/
en/publications/Publications/zika-virus
-americas-association-with-microcephaly
-rapid-risk-assessment.pdf).
6. Ioos S, Mallet HP, Leparc Goffart I,
Gauthier V, Cardoso T, Herida M. Current
Zika virus epidemiology and recent epi-
demics. Med Mal Infect 2014;44:302-7.
7. Faye O, Faye O, Diallo D, Diallo M,
Weidmann M, Sall AA. Quantitative real-
time PCR detection of Zika virus and
evaluation with field-caught mosquitoes.
Virol J 2013;10:311.
8. Faye O, Faye O, Dupressoir A, Weid-
mann M, Ndiaye M, Alpha Sall A. One-step
RT-PCR for detection of Zika virus. J Clin
Virol 2008;43:96-101.
9. Tamura K, Stecher G, Peterson D,
Filipski A, Kumar S. MEGA6: Molecular
Evolutionary Genetics Analysis version 6.0.
Mol Biol Evol 2013;30:2725-29.
10. Oliveira Melo AS, Malinger G,
Ximenes R, Szejnfeld PO, Alves Sampaio
S, Bispo de Filippis AM. Zika virus intra-
uterine infection causes fetal brain ab-
normality and microcephaly: tip of the
iceberg? Ultrasound Obstet Gynecol 2016;
47:6-7.
11. Macdonald EM, Koval JJ, Natale R,
Regnault T, Campbell MK. Population-
based placental weight ratio distributions.
Int J Pediatr 2014;2014:291846.
12. Alpert SG, Fergerson J, Nol LP. Intra-
uterine West Nile virus: ocular and sys-
temic findings. Am J Ophthalmol 2003;
136:733-5.
13. Tsai TF. Congenital arboviral infec-
tions: something new, something old.
Pediatrics 2006;117:936-9.
14. Chi JG, Dooling EC, Gilles FH. Gyral
development of the human brain. Ann
Neurol 1977;1:86-93.
15. Goldsmith CS, Ksiazek TG, Rollin PE,
et al. Cell culture and electron microscopy
for identifying viruses in diseases of un-
known cause. Emerg Infect Dis 2013;19:
886-91.
16. Gillespie LK, Hoenen A, Morgan G,
Mackenzie JM. The endoplasmic reticu-
lum provides the membrane platform for
biogenesis of the flavivirus replication
complex. J Virol 2010;84:10438-47.
17. Lanciotti RS, Kosoy OL, Laven JJ, et al.
Genetic and serologic properties of Zika
virus associated with an epidemic, Yap
State, Micronesia, 2007. Emerg Infect Dis
2008;14:1232-9.
18. Musso D, Roche C, Robin E, Nhan T,
Teissier A, Cao-Lormeau VM. Potential
sexual transmission of Zika virus. Emerg
Infect Dis 2015;21:359-61.
19. Faye O, Freire CC, Iamarino A, et al.
Molecular evolution of Zika virus during
its emergence in the 20th century. PLoS
Negl Trop Dis 2014;8(1):e2636.
Copyright 2016 Massachusetts Medical Society.

SHARING DATA IN A PUBLIC HEALTH EMERGENCY

The case for sharing data, and the consequences of not doing so, have been brought into stark relief by the Ebola and Zika
outbreaks. In response, the New England Journal of Medicine has become a journal signatory to the following statement.
In the context of a public health emergency of international concern, it is imperative that all parties make available any information
that might have value in combatting the crisis. As research funders and journals, we are committed to working in partnership to
ensure that the global response to public health emergencies is informed by the best available research evidence and data.
Journal signatories will make all content concerning the Zika virus free access. Any data or preprint deposited
for unrestricted dissemination ahead of submission of any paper will not preempt later publication in these journals.
Funder signatories will require researchers undertaking work relevant to public health emergencies to establish mechanisms
to share quality-assured interim and final data as rapidly and widely as possible, including with public health and research
communities and the World Health Organization.
We urge other journals and research funders to make the same commitments.

958 n engl j med 374;10 nejm.org March 10, 2016

50
Back to Table of Contents
20 YEARS ON THE WEB nejm.org

INFOGRAPHIC

Back to Table of Contents

 20 YEARS ON THE WEB nejm.org

96
M.org launched
edefined the
ional medical
cation

2013 FIRSTS 2004

Quick Take Video
Summary: In-Flight
Medical Emergencies
ON NEJM.ORG Audio: Classic
Whooping Cough

7ries in which NEJM

d each week.
M is the most widely
2005
2009 Audio Interview:
cited, and influential Image Challenge: Statins and
ral medicine periodical Whats the diagnosis? Over-the-Counter
world Availability

2009 2006
Interactive Medical Case: Videos in Clinical Medicine:

3 Million + The Writing on the Wall Placement of an Arterial Line

2008
Video Roundtable:
Physicians and Execution Highlights
e global visitors
from a Discussion of Lethal Injection
month

FIRSTS ON NEJM.ORG
NEJM.ORG
ER
NOTABLE MULTIMEDIA
2004 Audio: Classic Whooping Cough 2006 V
 ideos in Clinical Medicine: Placement of
A three-year-old girl with severe, fully devel- an Arterial Line

Million
e have taken the
oped whooping cough.
Klass P. Index Case. Classic Whooping
Cough. N Engl J Med 2004;350:2023-2026.
This video demonstrates the insertion of an
arterial line and considers indications, con
traindications, and potential complications.
M Image Challenge,
May 13, 2004. Perspective. Tegtmeyer K, Brady G, Lai S, Hodo R, and
at way to hone your
and diagnostic skills Braner D. Placement of an Arterial Line.
viewing images of 2005 A
 udio Interview: Statins and Over-the-
N Engl J Med 2006;354:e13. April 13, 2006.
mon conditions Counter Availability
Videos in Clinical Medicine.
The motivation for making statins available
MOST VIEWED MOST VIEWED INTERACTIVE
overIMAGE
the counter is understandable: to in-
IN CLINICAL MEDICINE
(continued
MEDICAL CASE
on next page)
creaseFemoral-Head
access to an effective and underused
Dislocation A Man with Bizarre Behavior
to the Scrotum 72,333+ VIEWS
therapy. Dr. Brian Strom
1.12M+ VIEWS
writes that it isMOST VIEWED CLINICAL
MEDICINE VIDEO
unclear that such a switch would help to Hand Hygiene

385,000+
achieve that goal. 317,796+ VIEWS

Strom BL. Statins and Over-the-Counter

Availability. N Engl J Med 2005;352:1403-
M Facebook
followers 1405. April 7, 2005. Perspective.
52
Back to Table of Contents
AUDIO WEBCAST TOUGHEST IMAGE CHALLENGE
Ebola Outbreak Whats the Diagnosis?
20 YEARS ON THE WEB nejm.org

FIRSTS ON NEJM.ORG (continued)

2008 Video Roundtable: Physicians and 2009 Image Challenge: First Image Challenge
Execution Highlights from a Discussion The image illustrates a left sixth nerve palsy
of Lethal Injection with herpes zoster ophthalmicus.
On January 14, 2008, the Journal hosted a Jude E, Chakraborty A. Left Sixth Cranial
videotaped roundtable discussion of the Nerve Palsy with Herpes Zoster Ophthal-
issues raised by Baze v. Rees, currently micus. N Engl J Med 2005;353:e14. October
before the Supreme Court, that asks 20, 2005. Image Challenge.
whether the three-drug protocol used to
carry out the death penalty by lethal 2013 Quick Take Video Summary: In-Flight
injection causes unnecessary pain and Medical Emergencies
suffering in violation of the Constitutional These investigators analyzed consultations
ban on cruel and unusual punishment. from five major airlines to a physician-di-
Gawande A, Denno DW, Truog RD, Waisel rected communications center to study the
D. Physicians and Execution Highlights epidemiology of 11,920 in-flight medical
from a Discussion of Lethal Injection. emergencies.
N Engl J Med 2008;358:448-451. January 31, Peterson DC, Martin Gill C, Guyette FX, et
2008. Perspective Roundtable. al. Outcomes of Medical Emergencies on
Commercial Airline Flights. N Engl J Med
2009 I nteractive Medical Case: The Writing on
2013;368;2075-83. Quick Take Video.
the Wall
A 52-year-old man with diabetes mellitus
was admitted to the hospital after present-
ing to the emergency department with
abdominal discomfort. His symptoms
began 5 weeks earlier, when epigastric
discomfort developed that did not radiate
elsewhere.
Vaishnava P, Ross JJ, Miller AL, Wolpin
BM. The Writing on the Wall. N Engl J Med
2009;361:e19. September 10, 2009. Interac-
tive Medical Case.

53
Back to Table of Contents
3 Million+
e global visitors
2008
Video Roundtable:
Physicians and Execution Highlights
from a Discussion of Lethal Injection
month 20 YEARS ON THE WEB nejm.org

NEJM.ORG
ER
Million
NOTABLE MULTIMEDIA
e have taken the
M Image Challenge,
at way to hone your
and diagnostic skills
viewing images of
mon conditions

MOST VIEWED MOST VIEWED INTERACTIVE

IMAGE IN CLINICAL MEDICINE MEDICAL CASE
Femoral-Head Dislocation A Man with Bizarre Behavior
to the Scrotum 72,333+ VIEWS
MOST VIEWED CLINICAL
1.12M+ VIEWS
MEDICINE VIDEO
Hand Hygiene

385,000+
317,796+ VIEWS

M Facebook
followers

AUDIO WEBCAST TOUGHEST IMAGE CHALLENGE

Ebola Outbreak Whats the Diagnosis?
OCTOBER 22, 2014 ONLY 18.72%
ANSWERED CORRECTLY

46,200+ THE FIRST VIDEO PUBLISHED ON NEJM.ORG

Separation of Conjoined Twins
M Twitter followers AUGUST 10, 2000

NEJM.ORG NOTABLE MULTIMEDIA

NEJM.ORG
TOP CITED ARTICLES
2000 F irst Video Published on NEJM.org: 2011 Most Viewed Clinical Medicine Video:
Separation of Conjoined Twins Hand Hygiene
ODUCT There are two congenital anomalies specific Infections associated with health care are
UNCHES to multifetal pregnancies: twin reversed-
arterial-perfusion sequence and conjoined
the most common adverse events resulting
from hospital stays; 5 to 10% of patients
twinning. in the developed world acquire such infec
Norwitz ER, Hoyte tions, and the burden is even higher in
ed as the 1997LPJ, Jenkins KJ, et al. 1998 2003
ngland Journal
Separation of Conjoined
Inflammation, Twins
Aspirin, with the Twin
Declining Morbidity developing countries.
Overweight, This video demon-
Obesity,
dicine and Surgery and the Risk of and Mortality among Patients strates proper
and Mortality from Cancer
e Collateral Branches Reversed-Arterial-Perfusion Sequence after hand-hygiene techniques.
Cardiovascular Disease in with Advanced Human in a Prospectively Studied
ence; renamed the Apparently
ngland Journal of
Prenatal PlanningHealthy Men Immunodeficiency Virus Infection
with Three-Dimensional Longtin Y, Sax Cohort of US AdultsB, et al. Hand
H, Allegranzi
ne in 1928 Modeling. N Engl J Med 2000;343:399-402. Hygiene. N Engl J Med 2011;364:e24.
August 10, 2000. Original Article March 31, 2011. Videos in Clinical
Brief Report. Medicine.
h of NEJM Journal Watch 2011 2016
Improved Survival with Zika Virus Associated (continued on next page)
Ipilimumab in Patients with Microcephaly
with Metastatic Melanoma
54
Back to Table of Contents
Group, a division
Massachusetts
20 YEARS ON THE WEB
NEJM.ORG NOTABLE MULTIMEDIA (continued)
2014 Audio Webcast: Ebola Outbreak
Experts Discuss Virology Epidemiology, and
Clinical Care.
Christopher Dye (World Health Organi-
zation), Jeremy Farrar (Wellcome Trust),
Armand Sprecher (Mdecins Sans Fron-
tires), Arjun Srinivasan and Matthew
Arduino (Centers for Disease Control and
Prevention), and Paul Farmer (Partners in
Health) joined moderators Jeffrey Drazen
and Lindsey Baden (New England Journal of
Medicine) in a live audio webcast to discuss
virology, epidemiology, and clinical care.
This international panel with hands-on ex-
perience provided the latest information on
the Ebola outbreak, including protection for
health care professionals and the general
public and projections for the future.
NEJM Live Audio Webcast: Ebola Outbreak.
October 22, 2014.
2014 Toughest Image Challenge: Dahls Sign
A 76-year-old woman was admitted to the
hospital with shortness of breath. She had
tachypnea and was breathing through
pursed lips. The patient had a long history
of advanced COPD. Symmetric, slanting
regions of hyperpigmentation were noted
on both thighs.
Miller PE, Houston BA. Dahls Sign. N Engl
J Med 2014;371:357. July 24, 2014. Images
in Clinical Medicine.
55
Back to Table of Contents
nejm.org
2015 I mage in Clinical Medicine: Femoral-Head
Dislocation
A 33-year-old man was admitted to the
emergency department after a motorcycle
accident. Examination of the intubated
patient showed a hard, swollen, bluish
scrotum and an externally rotated and
slightly shortened left leg.
Schicho A, Riepl C. Femoral-Head Dis-
location to the Scrotum. N Engl J Med
2015;372:863. February 26, 2015. Images in
Clinical Medicine.
2015 M
 ost Viewed Interactive Medical Case:
A Man with Bizarre Behavior
This interactive feature presents the case
of a 24-year-old man with abnormal and
agitated behavior. Test your diagnostic and
therapeutic skills at NEJM.org.
Rariy CM, Schram A, Ruan DT, et al. A
Man with Bizarre Behavior. N Engl J Med
2015;373:e8. August 13, 2015. Interactive
Medical Case.
 Improved Survival with Zika Virus Associated
Ipilimumab in Patients with Microcephaly
with Metastatic Melanoma

20 YEARS ON THE WEB nejm.org

Group, a division
Massachusetts
al Society, was formed

NEJM.ORG
h of NEJM Knowledge+

URGENT PUBLIC HEALTH ISSUES

h of NEJM Catalyst

h of NEJM Resident 360

EBOLA OUTBREAK ZIKA VIRUS

A S WE
AS W CELEBRAT
CELEBR
CELEBRA
CELEB
CELEBRATE
CE
CEL
ELEBRATE
LEBRATE
BRATE
RATE
ATE
TE OUR
O R PAST,
PA
PAS
AST,
ST,
T, WE
W ALSO
ALS
ALLSO
SOO LOOK
NEJM.ORG URGENT PUBLIC HEALTH ISSUES
LOOKK TOO THE
LOO TH
T HEE FUTURE.
FUTURE
FUTUR
FUTU
FU
F UTURE.
URE.
RE.
We
e wil
will continue
conti
co tinue
ue to evolve
evolv
ev ve and
a d look
loo
ok fo
for new
n w and
nd exciting
excit
ex iting
g ways
wa s to
o deliver
deli
deliver
er
 Zika
trusted
trus Virus
ed clinical
usted cl icall research
clinic researc
rese rch and
r search a d insightful
ins
insigh
nsightfu
ghtfulful commentary
commen
co
comm
mmentar ary to the
entary cal community.
th medical
m
medica
mededical Ebola
commun
comm
co Outbreak
mmunity
unity.
ty.
A collection of articles and other resources A collection of articles and other resources
on the Zika virus outbreak, including on the Ebola outbreak, including clinical
clinical reports, management guidelines, reports, management guidelines, and
and commentary. commentary.
2016 Copyright Massachusetts Medical Society. NEJM Group is a division of the Massachusetts Medical Society.

To view infographic in its entirety, click here.

56
Back to Table of Contents
346,200+
20 YEARS ON THE WEB
THE FIRST VIDEO PUBLISHED ON NEJM.ORG
Separation of Conjoined Twins
nejm.org

NEJM Twitter followers AUGUST 10, 2000

NEJM.ORG
PRODUCT TOP CITED ARTICLES
LAUNCHES
1812
Founded as the 1997 1998 2003
New England Journal Inflammation, Aspirin, Declining Morbidity Overweight, Obesity,
of Medicine and Surgery and the Risk of and Mortality among Patients and Mortality from Cancer
and the Collateral Branches Cardiovascular Disease in with Advanced Human in a Prospectively Studied
of Science; renamed the Apparently Healthy Men Immunodeficiency Virus Infection Cohort of US Adults
New England Journal of
Medicine in 1928

1987
Launch of NEJM Journal Watch 2011 2016
Improved Survival with Zika Virus Associated
Ipilimumab in Patients with Microcephaly
with Metastatic Melanoma
2012
NEJM Group, a division
of the Massachusetts
Medical Society, was formed

2014 NEJM.ORG
Launch of NEJM Knowledge+

URGENT PUBLIC HEALTH ISSUES

2015
Launch of NEJM Catalyst

2016
Launch of NEJM Resident 360

EBOLA OUTBREAK ZIKA VIRUS

A S WE
AS W CELEBRAT
CELEBR
CELEBRA
CELEB
CELEBRATE
CE
CEL
ELEBRATE
LEBRATE
BRATE
RATE
ATE
TE OUR
O R PAST,
PA
PAS
AST,
ST,
T, WE
W ALSO
ALS
ALLSO
SOO LOOK
LOOKK TOO THE
LOO TH
T HEE FUTURE.
FUTURE
FUTUR
FUTU
FU
F UTURE.
URE.
RE.
Wee wil
will continue
conti
co tinueue to evolve
evolv
ev ve and
a d look
loo
ok fo
for new
n w and
nd exciting
excit
ex iting
g ways
wa s too deliver
deli
deliver
er
trusted
trus ed clinical
usted cl icall research
clinic researc
rese rch and
r search a d insightful
ins
insigh ful commentary
nsightfu
ghtful commen
comm
commentar ary to the
entary th medical
m cal community.
medica
med
edical commun
comm
community
unity.
ty.

2016 Copyright Massachusetts Medical Society. NEJM Group is a division of the Massachusetts Medical Society.

Back to Table of Contents