Histologic Subtype Has Minor Importance for Overall

Survival in Patients with Adenocarcinoma of the

Uterine Cervix
A Population-Based Study of Prognostic Factors in 505 Patients with Nonsquamous
Cell Carcinomas of the Cervix
G. Cecilie Alfsen, M.D.
1
Gunnar B. Kristensen, M.D., Ph.D.
2
Eva Skovlund, Ph.D.
3
Erik O. Pettersen, Ph.D.
4
Vera M. Abeler, M.D., Ph.D.
1
1
Department of Pathology, The Norwegian Radium
Hospital, Oslo, Norway.
2
Department of Gynecologic Oncology, The Nor-
wegian Radium Hospital, Oslo, Norway.
3
Clinical Research Ofce, The Norwegian Radium
Hospital, Oslo, Norway.
4
Department of Physics, University of Oslo, Oslo,
Norway.
Presented in abstract form (as a poster) at the
eighth meeting of the International Gynecologic
Cancer Society, Buenos Aires, Argentina, October
2226, 2000.
Supported by grants from The Norwegian Cancer
Society.
The authors thank their colleagues at all depart-
ments and laboratories of pathology in Norway for
their cooperation in providing tissue samples. Col-
leagues at several departments of gynecology and
Tove Dahl at The Cancer Registry of Norway
helped in collecting clinical data. Nhudo Trinh and
smund Nyben are thanked for excellent techni-
cal assistance.
Address for reprints: G. Cecilie Alfsen, M.D., De-
partment of Pathology, The Norwegian Radium
Hospital, N-0310 Oslo, Norway; Fax: 47 22 93
54 26; E-mail: c.g.alfsen@labmed.uio.no
Received January 31, 2001; revision received July
24, 2001; accepted August 1, 2001.
BACKGROUND. The incidence of adenocarcinoma of the uterine cervix is increasing.
For better prognostic information, the authors studied all nonsquamous cell car-
cinomas (non-SCCs) in the Norwegian population over a total of 15 years.
METHODS. All non-SCCs from three 5-year periods (19661970, 19761980, and
19861990) were reviewed and classied according to the World Health Organiza-
tion classication system, and histopathologic and clinical parameters were reg-
istered. Tissue blocks were available from all patients.
RESULTS. Of 505 patients, 417 had tumors classied as adenocarcinoma, and 88
had tumors classied as other non-SCC. The mean ages were 53 years and 52 years
for patients with adenocarcinoma and non-SCC, respectively. Sixty-two percent of
the staged patients had clinical Stage I disease according to the classication
system of the International Federation of Gynecology and Obstetrics (FIGO). In
univariate analyses, histology, architectural and nuclear grade, extension to the
vagina or corpus uteri, tumor length ( 20 mm) or tumor volume ( 3000 mm
3
),
inltration depth (in thirds of the cervical wall), thickness of the remaining wall
( 3 mm), vascular invasion, lymph node metastases, treatment, and patient age
were signicant variables in patients with FIGO Stage I disease. Variables with no
signicance in patients with Stage I disease were number of mitoses, state of
resection margins, inltration to ectocervix, tumor thickness, lymphoid reaction,
earlier or concomitant cervical intraepithelial neoplasia, stump carcinoma, DNA
ploidy or DNA index, or time period. Multivariate analyses of patients with FIGO
Stage I disease identied small cell carcinoma, corpus inltration, vascular inva-
sion, and positive lymph nodes as independent prognostic factors.
CONCLUSIONS. Small cell carcinoma was the only histologic subgroup of indepen-
dent importance for prognosis in patients with non-SCC of the uterine cervix. No
signicant difference between major subtypes of adenocarcinoma favored a sim-
plied classication. Extension to the corpus in patients with early-stage disease
was of independent signicance and should be acknowledged in planning treat-
ment. Cancer 2001;92:247183. 2001 American Cancer Society.
KEYWORDS: cervical carcinoma, adenocarcinoma, small cell carcinoma, prognosis,
prognostic factors.
A
n increase in the incidence of nonsquamous cell carcinoma (non-
SCC) of the uterine cervix has been reported from industrialized
countries over the past 30 years.
18
Reports of rising incidence rates
for cervical adenocarcinoma in young women
6
have given rise to
concern and have accelerated the search for prognostic variables.
2471
2001 American Cancer Society
Non-SCCs still are rare compared with squamous cell
carcinomas (SCCs), and only a few studies of prognos-
tic factors in patients with adenocarcinoma have in-
cluded 200 patients.
911
Not all of those studies were
performed with histopathologic review,
10
and, be-
cause all studies were center-based, bias of patient
groups may have inuenced the results.
With the objective of including enough patients
for more a precise statistical analyses of prognostic
factors, we analyzed all patients with non-SCC in the
Norwegian population over three 5-year periods
(19661970, 19761980, and 19861990). To our
knowledge, this is the rst population-based study of
patients with non-SCC of the uterine cervix in which
all patients were reviewed histopathologically.
MATERIALS AND METHODS
Patients were identied through the les of the Cancer
Registry of Norway (NCR), which has received man-
datory information on all cancer patients in Norway
since 1953. Details on the reporting system, the col-
lection of samples, and the staining of sections have
been published previously.
2
The majority of patients
had been diagnosed and treated at The Norwegian
Radium Hospital in Oslo (90% in 19661970 and
19761980 and 80% in 19861990), with tissue sam-
ples available at the Department of Pathology. The
remaining patients were treated at one of the other
three oncology centers in Norway. The handling of
surgical specimens was uniform in all pathology de-
partments that provided samples, with radial section-
ing of the wall in the uterine cervix.
Histopathology
Tumors were classied as adenocarcinomas and other
carcinomas with subgroups according to the World
Health Organization (WHO) classication system for
cervical carcinoma.
12
The distinction between endo-
cervical (EC) and endometrioid (EM) carcinoma was
based on morphology and the pattern of intracellular
mucin staining. A few, single mucin positive cells were
tolerated in otherwise typical EM tumors. If distinct
areas in an otherwise EM tumor contained mucin, the
carcinoma was classied as mixed. Tumors other than
small cell carcinoma and adenosquamous carcinoma
were classied as mixed when 25% of a second
component was present. The diagnosis of villoglandu-
lar carcinoma was made only on cone biopsies or
hysterectomy specimens. A tumor was classied as
adenosquamous only if atypical glandular structures
were found in addition to invasive squamous areas.
Tumors with single, mucin positive cells without glan-
dular structures were classied as SCC and excluded
from the study.
The diagnosis of a small cell carcinoma was con-
rmed by immunohistochemic stains for neuroendo-
crine differentiation. An additional presence of glan-
dular and/or squamous tumor differentiation was
registered but did not alter the classication as small
cell carcinoma.
All samples that were included in the current
study were reviewed by two of the authors (G.C.A. and
V.M.A.) without access to the clinical data. If discrep-
ancies in interpretation occurred, then a joint review
was undertaken, and the differences were resolved.
Other histopathologic data recorded were grade,
growth patterns, number of mitoses per 10 high-
power elds (only counted in patients from 1986
1990), tumor length in the cervical canal (cephalocau-
dal length), maximum tumor thickness, depth of
invasion (estimated in thirds of the cervical wall), and
thickness of the remaining wall rim. Measurements
were made on specimens that had not been subjected
to radiation or chemotherapy. Because radial section-
ing of the cervix had been performed, tumor volume
was estimated using the maximal tumor thickness and
length after a modied formula from Burghardt
13
:
(length) (thickness) (thickness) 1.5.
Relation to resection margins, tumor extension to
the ectocervix, the vaginal cuff, or the corpus, vessel
invasion, pelvic lymph node metastases; inammatory
reaction (Grade 13), the presence of histologic char-
acteristics consistent with human papillomavirus in-
fection, and cervical intraepithelial neoplasia (CIN) or
carcinoma in adjacent squamous epithelium also were
registered. Extension to the ectocervix and vagina was
registered from clinical records and/or histologic
slides. Inltration to the isthmus/corpus was regis-
tered when reevaluating the histologic slides (dilata-
tion and curettage material and hysterectomy speci-
mens). No distinction was made between supercial
inltration (in isthmus or corpus mucosa only) or
deep inltration in the muscular wall. When inltra-
tion in or beyond the isthmus uteri was found, special
concern was given to the possibility of a primary origin
in the corpus. All specimens with a possible primary
origin in the endometrium were excluded from the
study (for further details, see Alfsen et al.
2
).
Grading
Tumors were graded according to the degree of tubu-
lar formation in poorly differentiated, moderately dif-
ferentiated, and well differentiated tumors and ac-
cording to nuclear atypia (nuclear Grade 13).
12
Nuclear grade was evaluated in the most atypical area.
Grading was not performed on material that had been
subjected to radiation or chemotherapy. Separate ar-
chitectural and nuclear grades were determined for all
2472 CANCER November 1, 2001 / Volume 92 / Number 9
material with the exception of clear cell carcinomas,
adenosquamous carcinomas, and small cell car-
cinomas.
DNA Ploidy
The nuclear DNA content was measured by ow cy-
tometry using formalin xed tumor tissue from paraf-
n blocks. Cell suspension preparation and analysis
was performed as described by Hedley et al.
14
with
minor modications.
15
Subsequent to cutting 100-
m-thick sections for analysis, a 5-m-thick section
was stained with hematoxylin and eosin and evalu-
ated. Classication of the DNA histograms was per-
formed without knowledge of the nal outcome or
other clinical information. DNA ploidy analysis was
performed only on material from 19761980 and
19861990.
Clinical Features
The patients had been staged clinically according to
the staging system of the International Federation of
Gynecology and Obstetrics (FIGO)
16
: FIGO Stage I,
disease conned to the cervix or corpus uteri (equiv-
alent to International Union Against Cancer [UICC]
classication
17
T1N0M0); FIGO Stage II, inltration to
the vagina but not reaching the lower one-third or
inltration into the ligaments and not reaching the
pelvic walls (equivalent to UICC T2N0M0); FIGO Stage
III, inltration of the lower one-third of the vagina or
into the pelvic wall (equivalent to UICC T3aN0M0 or
T1T3aN1M0 or T3bAnyNM0); FIGO Stage IV, inltra-
tion into the urinary bladder or rectum or distant
metastases (equivalent to UICC T4AnyNM0 or
AnyTAnyNM1). FIGO Stage I disease often was sub-
staged into Stage IA or IB (macroscopically invisible or
visible tumors, respectively; equivalent to UICC T1a or
T1b). FIGO Stage IA tumors with horizontal spread 7
mm were subdivided further into Stage IA1 (tumors
3 mm in depth) and Stage IA2 (tumor depth be-
tween 3 mm and 5 mm; equivalent to UICC T1a1 and
T1a2). Larger tumors or clinically visible tumors were
staged as IB (T1b). The registration of clinical stage at
the time of diagnosis was based on information from
hospital records, histopathologic forms, or from the
NCR.
No patient was staged retrospectively by the au-
thors. Clinical stage had not been recorded for 16
patients.
A history of CIN or SCC was registered only if it
was conrmed histopathologically. Twelve patients
TABLE 1
Histologic Distribution and Clinical Characteristics in 505 Patients with Nonsquamous Cell Carcinoma of the Cervix Uteri
Histology type No.
Mean
age (yrs)
Age range
(yrs)
FIGO stage (n 489 patients)
a
I II III IV
Total (all patients) 505 53 1691 302 103 58 26
Total adenocarcinomas 417 53 1691 253 83 44 22
Adenocarcinoma
NOS 45 59 3091 20 12 7 3
Mucinous NOS 34 57 2979 15 9 8 2
Endocervical 120 50 2584 81 21 11 4
Intestinal 7 49 3375 5 2
b

b
Signet ring 4 59 3175 3 1
b

b
Endometrioid 104 51 2384 71 19 8 4
Clear cell 37 59 1685 23 3 3 5
Villoglandular 8 33 3335 7
b

b
Serous 18 59 3477 6 6 3 3
Mesonephric 3 54 4572 2 1
b

b
Mixed 37 53 2782 20 9 4 1
Total other carcinomas 88 52 2289 49 20 14 4
Adenosquamous 36 53 2480 20 12 3 1
Glassy cell 2 43 2957 2
b

b
Adenoid cystic/basal 4 72 6385 4
b

b
Carcinoid 1 75
b
1
b

b
Small cell 29 47 2289 18 5 4 2
Undifferentiated 16 54 2884 5 2 7 1
FIGO: International Federation of Gynecology and Obstetrics; NOS: not otherwise specied.
a
Sixteen patients without conrmed clinical stage are not listed.
b
No patients were found.
Adenocarcinoma of the Cervix/Alfsen et al. 2473
TABLE 2
Histopathologic Features and Overall Survival: Univariate Analysis
a
Characteristic
FIGO Stage I FIGO Stage IIIV
No.
5-yr survival
(%)
b
Log-rank
P value No.
5-yr survival
(%)
b
Log-rank
P value
Histology
Endocervical, intestinal and signet ring cell type 89 80 0.001 39 31 0.001
Adenocarcinoma and mucinous carcinoma, NOS 35 71 41 27
Endometrioid 71 93 31 45
Clear cell 23 70 11 9
Serous 6 83 12 8
Adenosquamous 20 55 16 44
Glassy cell undifferentiated 7 29 10 30
Small cell 18 33 11 0
Mixed 20 75 14 50
Villoglandular 7 100 0
All other types 6 67 2 100
Tubular differentiation
c
Well 101 93 0.001 29 41 0.305
Moderate 87 80 69 33
Poor 53 59 51 29
Cellular grade
c
Mild 10 90 0.001 1 0 0.607
Moderate 159 88 81 38
Severe 72 64 67 28
Inltration to ectocervix
d
No 62 84 0.450
Yes 181 73
Not evaluable 59
Inltration to vagina
d
No 228 84 0.001
Yes 13 31
Not evaluable 61
Inltration to corpus
d
No 177 86 0.001
Yes 38 55
Not evaluable 87
Tumor thickness (mm)
d
1.03.0 26 85 0.125
3.15.0 29 93
5.0 103 69
Not evaluable 145
Tumor length (mm)
d
1.07.0 23 83 0.013
7.120.0 61 89
20.0 27 63
Not evaluable 191
Depth in cervical wall
d
Inner two-thirds 125 87 0.001
Outer one-third 46 61
Through the wall 17 53
Not evaluable 114
Thickness of remaining wall (mm)
d
3 88 82 0.002
13 42 62
Not evaluable 172
Volume (mm
3
)
d
13000 78 87 0.001
3000 26 58
Not evaluable 198
(continued)
2474 CANCER November 1, 2001 / Volume 92 / Number 9
with a history of partial hysterectomy were included
after a review of clinical records and histopathologic
specimens ruled out spread from a primary process
originating in the corpus uteri.
For a better comparison with earlier Norwegian
studies,
2,5
patients were divided into three age groups:
age 35 years, ages 3554 years, and age 55 years.
Information on follow-up and cause of death was re-
trieved from hospital records and from survival data
provided by the NCR. Strict criteria were used for the
registration of death as disease related. When in
doubt, the cause of death was registered as unknown.
Follow-up ended in February 1999.
Statistical Methods
Overall survival was used as the endpoint in survival
analyses because the cause of death was missing in
37% of patients. Overall survival was calculated in
months from the date of diagnosis to the date of death
from any cause or February, 1999. Patients who were
still alive in February of 1999 were censored. The re-
lation between overall survival and possible prognos-
tic factors was analyzed with KaplanMeier plots and
log-rank tests. P values 0.05 were regarded as sta-
tistically signicant. All factors signicant in univari-
ate analyses (Tables 2 and 3) were entered in a Cox
proportional hazards model with time periods as a
stratication factor. This analysis was restricted to pa-
tients with FIGO Stage I disease. The Cox model was
reduced thereafter by backward elimination to include
signicant variables only. The proportionality as-
sumption was checked by visual inspection of cumu-
lative hazard plots. All variables were categorized.
Patients who were not evaluated were included in
the Cox analysis as a separate category for each of the
variables in question. Because these categories con-
sisted of patients with unknown status for the charac-
teristics in question, estimates were meaningless clin-
ically and, thus, are not shown: The P values presented
refer to the comparison of categories shown and do
not include patients who were not evaluated. The
drawback to this approach was a crude adjustment for
covariates. To conrm the results, analyses of the sub-
sets of patients with no missing values were per-
TABLE 2
(continued)
Characteristic
FIGO Stage I FIGO Stage IIIV
No.
5-yr survival
(%)
b
Log-rank
P value No.
5-yr survival
(%)
b
Log-rank
P value
Vascular invasion
Negative 124 88 0.001 42 33 0.130
Positive 110 62 56 32
Not evaluable 68 75 89 29
No. of positive lymph nodes
None 147 91 0.001 9 78 0.001
12 34 65 13 46
2 19 21 3 33
Not evaluated 102 68 162 27
Lymphoid reaction
Mild 120 73 0.072 63 32 0.160
Moderate-severe 126 80 55 33
Not evaluable 56 71 69 29
DNA ploidy
e
Diploid 82 78 0.277 26 52 0.849
Nondiploid 102 71 74 58
DNA index
e
1.3 94 78 0.327 36 59 0.443
1.3 77 69 58 57
Not evaluable 13 0
FIGO: International Federation of Gynecology and Obstetrics; NOS: not otherwise specied.
a
Total: 489 patients.
b
Overall survival.
c
Tumors of clear cell type, small cell carcinomas, or adenosquamous carcinomas were not graded.
d
Not evaluated in patients with FIGO Stage IIIV disease.
e
Only performed on patients from 19761980 and 19861990 (n 184 patients).
Adenocarcinoma of the Cervix/Alfsen et al. 2475
formed. These produced estimates of relative hazards
similar to the estimates presented, but the total num-
ber of patients was too small to yield a valuable
analysis.
To achieve histologic groups large enough for Cox
regression analyses, the histologic tumor subgroups
were recategorized into nine groups. Similarities in
histologic appearance and survival from Kaplan
Meier plots were drawn into account when dividing
the groups. Patients with EC, intestinal, and signet
ring cell carcinomas were regrouped into one group
and used as reference. Patients with carcinomas not
otherwise specied (NOS; (adenocarcinoma NOS and
mucinous carcinoma NOS) were regrouped into a sec-
ond group. Patients with EM, clear cell, serous, mixed,
adenosquamous, and small cell carcinomas all re-
mained in separate groups. The two patients with
glassy cell carcinoma were regrouped with undifferen-
tiated carcinomas. Other small subgroups of patients
with villoglandular, mesonephric, adenoid basal, ade-
noid cystic, and carcinoid tumors were excluded from
the regression analysis. Differences in proportion were
evaluated by the chi-square test. Statistical analyses
were performed using SPSS software (SPSS, Inc., Chi-
cago, IL).
RESULTS
After review, 417 tumors (83%) were subtyped as ad-
enocarcinomas, and 88 tumors (17%) were subtyped
as other carcinomas (Table 1). The minimum follow-
up was 9 years. Three hundred patients (59%) had
died before February 1999. The cause of death was
unknown in 110 patients (37%), the majority of whom
were age 54 years. Death from disease was regis-
tered in 165 patients (55%). The rate of death from
unknown cause, especially in the oldest age group,
was too large to allow for evaluation of age as a prog-
nostic factor. Therefore, age was included in the Cox
regression analyses only to adjust imbalances in age
distribution between the categories of variables.
Histopathologic and clinical parameters that had
a signicant impact on overall survival in univariate
analyses are listed in Tables 2 and 3, which include all
489 patients with known clinical stage. Variables that
had an independent signicant impact on overall sur-
vival in the Cox regression analysis of patients with
FIGO Stage I disease are summarized in Table 4. One
hundred two patients had unknown lymph node sta-
tus, and inltration to the corpus and volume were
measurable in 215 samples and 104 samples, respec-
tively.
Histologic subtyping of adenocarcinomas had no
independent impact on overall survival among pa-
tients with FIGO Stage I disease. Differences in sur-
vival between histologic subtypes of non-SCC were
found in univariate analyses (Fig. 1), but only small
cell carcinomas maintained their prognostic impact in
the Cox regression analysis. Compared with mucinous
tumors of EC, signet ring, or intestinal cell types (ref-
erence group), patients with small cell carcinomas had
a nearly four-fold risk of dying, both among patients
with FIGO Stage I disease and when patients with
disease in all stages were analyzed (results not shown).
The differentiation of glassy cell carcinomas ver-
sus undifferentiated tumors often proved difcult.
When discrepancies between the reviewing authors
occurred, the tumor was classied as undifferentiated.
Thus, it was considered correct to reclassify these two
small tumor groups together for the Cox regression
analysis. Although the histology of glassy cell/undif-
ferentiated tumors also showed a P value 0.05 in the
Cox regression analysis (Table 4), the number of pa-
tients in this group was small, and the condence
interval was considered too wide to yield safe conclu-
sions. One of the two patients with glassy cell carci-
noma who was included in this group was of fertile age
but was not pregnant.
The histologies of adenocarcinoma NOS, muci-
nous carcinoma NOS, or mixed tumors were of bor-
derline prognostic signicance in the Cox regression
analysis. However, as indicated by the nomenclature,
both of these tumor groups had heterogeneous histo-
logic patterns that also proved to have low reproduc-
ibility (unpublished results).
Patients with villoglandular carcinoma were not
included in the Cox regression analysis, but all pa-
tients with this subtype survived for at least 5 years.
No patients with minimal deviation carcinoma were
identied.
Ninety-seven percent of all patients had been
staged clinically (n 489 patients). Sixty-two percent
of staged patients had FIGO Stage I disease, 21% had
FIGO Stage II disease, 12% had FIGO Stage III disease,
and 5% had FIGO Stage IV disease. Several of the 16
patients with no clinical staging had histopathologic
characteristics consistent with FIGO stage I disease.
The prognostic impact of FIGO stage at the time of
diagnosis was strong. Seventy-six percent of patients
with Stage I disease survived for 5 years compared
with 50% of patients with clinical Stage II or greater
disease.
Of the patients with FIGO Stage I disease, 10% (31
patients) had been staged clinically with FIGO Stage IA
disease, and 87% (264 patients) had been staged clin-
ically with FIGO Stage IB disease. Clinical staging with
FIGO Stage IA and IB disease was associated weakly
with measurements of tumor size at review. Of the
patients with measurable tumor size in histologic
2476 CANCER November 1, 2001 / Volume 92 / Number 9
samples, 63% (12 of 19 patients) with FIGO Stage IA
disease had tumor thickness 5 mm and tumor
length 7 mm. The discrepancy with regard to mea-
surements in patients with clinical Stage IA disease at
review was explained only in part by differences in the
parameters measured (inltration depth vs. thick-
ness). For estimation of risk for lymph node metasta-
ses, measurements of tumor thickness or inltration
depth seemed equally useful, although the total num-
bers were small. None of the 31 patients who were
staged clinically with Stage IA disease and none of
nine patients with tumor thickness 5 mm had
lymph node metastases as long as tumor lengths mea-
sured 7 mm (Table 5).
No statistically signicant difference in survival
was found among patients who had tumors measuring
7 mm in length compared with patients who had
tumors measuring 720 mm in length (P 0.942).
Patients with FIGO Stage I small cell carcinoma were
staged clinically with Stage IB disease, and only one of
these patients had a tumor size 5 mm in thickness
and 7 mm in length. For an alternative to measure-
ments in millimeters, measurements of inltration
depth in thirds of the cervical wall proved signicantly
related to overall survival in univariate analysis in
patients with FIGO Stage I disease (Table 2). Inltra-
tion depth in greater than two-thirds of the cervical
wall implied a clearly increased risk of lymph node
metastases (Table 5).
Tumor inltration to the corpus proved a strong
TABLE 3
Clinical Features and Overall Survival: Univariate Analysis
a
Characteristic
FIGO Stage I FIGO Stages IIIV
No.
5-yr survival
(%)
b
Log-rank
P value No.
5-yr survival
(%)
b
Log-rank
P value
FIGO stage
I 302 76 0.001
II 103 46
III 58 19
IV 26 0
Missing 16
Age (yrs)
35 67 75 0.001 7 14 0.001
3554 146 83 58 48
54 89 64 122 24
Treatment
Surgery alone (Wertheim) 78 90 0.001 0 0.001
Preoperative radiation and
surgery 68 92 10 70
Preoperative radiation,
surgery, and
postoperative radiation 26 50 7 43
Postoperative radiation 55 56 14 43
Radiation only 36 53 137 28
Other treatments
c
39 82 19 16
Time period
19661970 66 79 0.381 52 31 0.188
19761980 89 69 62 24
19861980 147 78 73 37
CIN (current or previous)
Yes
d
65 89 0.099 19 37 0.128
No history known 237 72 168 30
Stump carcinoma
Yes 5 60 0.084 7 14 0.372
No 297 76 180
FIGO: International Federation of Gynecology and Obstetrics; CIN: cervical intraepitelial neoplasia.
a
Total, 489 patients.
b
Overall survival.
c
Includes patients who underwent simple hysterectomies, and adjuvant chemotherapy and 21 patients with no treatment information.
d
Includes nine patients with a history of squamous cell carcinoma and CIN.
Adenocarcinoma of the Cervix/Alfsen et al. 2477
and independent prognostic factor (Figs. 2, 3). Pa-
tients with tumor extension to the corpus uteri had a
two-fold risk of dying compared with patients who
had disease restricted to the uterine cervix (Table 4).
KaplanMeier survival plots indicated that the im-
pact of corpus inltration was strongest for patients
with adenocarcinoma. The impact of corpus inltra-
tion was distinct for every subgroup of patients with
adenocarcinoma (P 0.001), with the exception of
serous carcinomas, for which the number of patients
was too small to allow valid statistical evaluation. Cor-
pus inltration was related positively to other param-
eters of tumor volume (such as length, thickness,
depth, or volume), to older age, and to the nding of
lymph node metastases. No other parameter of size or
extension had an independent impact on overall sur-
vival in the Cox regression model.
The mode of treatment was highly dependent on
disease stage and patient age and had no impact on
overall survival in the Cox regression analysis. Status
of resection margins, number of mitoses, number of
removed lymph nodes, or time period were variables
with no signicant impact on survival (data not
shown).
DISCUSSION
The current study, to our knowledge, is unique in that
it is the rst population-based study of histopatholog-
ically reviewed non-SCC of the uterine cervix. Two
earlier population-based studies of patients with cer-
vical carcinoma have been published from Norway,
3,5
but they did not include histopathologic reviews.
Other authors have reported on prognostic factors for
patients with non-SCCs from center-based studies
outside Norway, with histopathologic reviews includ-
ed.
9,11,1836
However, the nomenclature has varied in
these studies, especially regarding the denition of
adenosquamous carcinoma and the distinction of
these tumors from adenocarcinomas. Thus, the prog-
nostic impact for patients with pure cervical adeno-
TABLE 4
Multivariate Analysis of Histologic and Clinical Prognostic Factors for Overall Survival in Patients with International Federation of Gynecology
and Obstetrics Stage I Carcinoma of the Uterine Cervix
a
Characteristic
No. of
patients
Hazard
ratio
b
95% condence
interval P value
Histology group
Endocervical, int, and sign 89 1.00 0.001
Adenocarcinoma and mucinous, NOS 35 1.80 0.983.30 0.058
Endometrioid 71 0.93 0.521.67 0.815
Clear cell 23 1.72 0.813.69 0.160
Serous 6 0.57 0.132.46 0.446
Mixed 20 2.03 0.974.25 0.062
Adenosquamous 20 1.54 0.713.34 0.273
Small cell 18 3.92 1.928.01 0.001
Glassy cell and undifferentiated 7 4.93 1.7114.2 0.003
Inltration to corpus
No 168 1 0.021
Yes 36 2.06 1.113.82
Not evaluable 85
Vascular invasion
No 113 1 0.002
Yes 109 2.25 1.353.75
Not evaluable 67
Positive lymph nodes
None 139 1 0.001
12 34 1.79 1.003.82 0.052
3 19 4.67 2.329.41 0.001
Not evaluated 97
Age (yrs)
3554 143 1 0.001
35 62 1.00 0.581.73 0.992
55 84 1.82 1.152.87 0.010
FIGO: International Federation of Gynecology and Obstetrics; int: intestinal type carcinoma; sign: signet ring cell carcinoma; NOS: not otherwise specied.
a
Total, 289 patients.
b
Reference groups are indicated by hazard ratio of 1.
2478 CANCER November 1, 2001 / Volume 92 / Number 9
carcinoma has remained unclear. Because of the strict
criteria used for the diagnosis of adenosquamous car-
cinomas in the current study, we were able to separate
the subgroups of patients with pure adenocarcinoma
and patients with adenosquamous tumors and other
non-SCCs for a more denite determination of prog-
nostic importance.
Our results conrm the impact of clinical stage
and lymph node metastasis as strong and indepen-
dent prognostic factors for patients with non-SCCs of
the uterine cervix. In addition, the results showed that
the factors small cell carcinoma histology, tumor ex-
tension into the uterine corpus, and vascular invasion
in patients with FIGO Stage I disease were signicant
risk factors in multivariate analyses.
To the authors knowledge, no population-based
studies have been presented on patients with small
cell carcinoma of the neuroendocrine type of the uter-
ine cervix. The exact incidence rate for these tumors
remains unknown, since some may be hidden among
the large numbers of squamous carcinomas. A previ-
ous center-based study from our institution
37
showed
that patients with small cell carcinoma of the uterine
cervix had a poor prognosis. Consistent with our ear-
lier ndings, we registered all tumors with any area of
small cell carcinoma differentiation as small cell car-
FIGURE 1. KaplanMeier plot for
overall survival according to histologic
subtype in patients with International
Federation of Gynecology and Obstetrics
(FIGO) Stage I disease. A) Overall sur-
vival for adenocarcinoma subgroups.
VG: villoglandular carcinoma; NOS: not
otherwise specied (adenocarcinoma or
mucinous carcinoma); EM: endometrioid
carcinoma; EC other muc.: endocer-
vical, signet ring, and intestinal type car-
cinoma. B) Overall survival for other car-
cinoma subgroups. Other: adenoid
cystic, adenoid basal, and mesonephric
carcinoma and carcinoid tumor; Glassy/
undiff.: glassy cell and undifferentiated
carcinoma.
Adenocarcinoma of the Cervix/Alfsen et al. 2479
cinomas, regardless of the proportion of other histo-
logic patterns. The current results conrm that cervi-
cal carcinomas with any areas of small cell carcinoma
are highly aggressive tumors.
Surprisingly, no histologic subtype of adenocarci-
nomas had any independent impact on overall sur-
vival. In the studies by Hurt et al.
24
and Saigo et al.,
30
it was found that patients with carcinoma of the EM
subtype had a better prognosis, whereas Raju et al.
35
found that patients with the EM subtype had a worse
prognosis compared with patients who had mucinous
tumors of the cervix. The distribution by subtype in
these studies varied considerably, with EM carcino-
mas constituting between 17% and 50% of all adeno-
carcinomas, and the criteria for diagnosis were not
dened uniformly. In the current study of a total of
240 patients with mixed or pure EC and EM carcino-
mas, no signicant differences in clinical appearance
or prognosis were identied. We recently reported a
rising incidence rate for cervical adenocarcinomas of
the EM subtype.
2
This changing pattern of adenocar-
cinoma subgroups in the cervix uteri does not seem to
have any inuence on clinical outcome.
Serous differentiation has been associated with
disease recurrence in a study of 65 patients with Stage
I and II cervical carcinoma with glandular differenti-
ation by Costa et al.
21
However, that study included
only six patients with the pure serous type (17% of all
pure adenocarcinomas). Eight additional patients in
that study were described with glandular components
of adenosquamous tumors. Serous carcinomas consti-
tuted only 4% of the adenocarcinomas in clinical Stage
I and II in the current study (12 of 336 patients). Differ-
ences in interpretation of histopathologic patterns and
grade, thus, may be of major importance with regard to
the impact and number of serous tumors.
In previous studies, patients with clear cell cervi-
cal carcinoma have been rare.
9,10
No information on
the use of diethylstilboestrol (DES) was provided in
those studies. DES use by the mother was not an
etiologic factor in young females with clear cell carci-
noma of the cervix or vagina according to an earlier
study from Norway.
38
In the current study, no patient
had a known clinical history of DES exposure, and
several patients were diagnosed before DES was intro-
duced. The natural history of clear cell carcinoma in
the cervix, thus, is unrelated to DES. Clear cell carci-
noma in our study primarily was a tumor of elderly
women. Thirty percent of these patients presented
with clinical Stage II or greater disease.
FIGURE 2. Photomicrograph of cervical adenocarcinoma inltrating in the
endometrium. The material was from preoperative dilatation and curettage
(hematoxylin and eosin stain).
TABLE 5
Frequency of Lymph Node Metastases According to Surgical Pathologic Stage or Tumor Inltration Depth in Thirds of the Cervical Wall in
Patients with International Federation of Gynecology and Obstetrics Stage I Carcinoma of the Uterine Cervix
a
Lymph node
metastases
FIGO Stage IA
b
FIGO Stage IB
b
Inner
one-third
Middle
one-third
Outer
one-third
Through the
cervical wall
No. % No. % No. % No. % No. % No. %
None 9 100 63 65 29 94 32 84 18 50 4 29
12 lymph nodes
c
19 20 2 7 5 13 10 28 3 21
2 lymph nodes
c
15 15
c
1 3 8 22 7 50
FIGO: International Federation of Gynecology and Obstetrics.
a
Total, 121 (only patients with examined lymph nodes and no preoperative radiation were included).
b
FIGO staging according to tumor thickness and length in millimeters after review.
c
No patients were found.
2480 CANCER November 1, 2001 / Volume 92 / Number 9
The current study did not nd that grade was of
independent signicance for prognosis. Other studies
have provided discrepant results.
9,11,18,19,21,22,25,27,29,35,39
This may be due to the inclusion of histologic groups of
patient with adenosquamous carcinoma
11,22,25
or clear
cell carcinoma
11
in the other studies or to the use of
different grading systems.
9,25,35
When comparing the prognosis for patients with
glandular and nonglandular carcinomas of the uterine
cervix, several authors have found inferior survival for
patients with glandular tumors.
3,5,10,23,26,34,36,39,40
Be-
cause of these differences, speculations have been
made on the relevance of a similar grading system for
patients with glandular and nonglandular carcinomas
of the uterine cervix. Glandular tumors of the uterine
cervix may have more in common with endometrial
carcinomas than with cervical SCCs, and their biologic
properties may not be reected in a staging system
based on tumors that arise from the squamous epi-
thelium in the distal part of the organ.
In the current study, the relation between clinical
staging and pathologic staging of patients with early-
stage disease after review was weak. Based on the
results from univariate analyses, overall survival was
not inuenced signicantly by tumor thickness as
long as it was 5 mm or by tumor length as long as it
was 20 mm. Except for corpus inltration, none of
the measures of tumor volume or extension reached
signicance in the multivariate analysis. However, be-
cause of the low numbers of patients evaluated in
some categories, minor differences may have been
overlooked.
Because of the anatomic condition of glands deep
in the inner one-third of the cervical wall, the mea-
surement of inltration depth in patients with adeno-
carcinomas is problematic. The signicance of depth
of inltration also may vary according to the histologic
type and pattern of growth. Thus, the application of a
staging system for patients with early-stage adenocar-
cinoma based on the depth of inltration measured in
the same manner as that used for patients with SCC
may be misleading. We therefore suggest that inltra-
tion depth is best estimated in thirds of the cervi-
cal wall.
The problems with accurate measurements of
depth and length also may have implications for com-
parative survival studies on prognostic differences be-
tween patients with adenocarcinomas and SCCs. In
the comparative study by Kleine et al.,
26
the most
important difference in prognosis was seen in patients
with FIGO Stage I and II disease who were treated by
radiotherapy but not in patients with higher stage
disease who also were treated by radiotherapy. We
agree with the conclusion of those authors that the
reported difference in outcome for patients with ade-
nocarcinomas and SCCs may be a problem of appro-
priate staging in patients with adenocarcinoma rather
than differences in biologic properties.
The thickness of the remaining cervical wall has
been reported as another useful prognostic factor in
patients with cervical carcinoma inltrating to a depth
5 mm.
34,41
Although measurements of the remain-
ing thickness of the wall proved highly signicant in
univariate analysis in our data, it was often problem-
atic to dene the remaining wall components, espe-
cially at the lateral margins, where the muscular bers
split and project fanlike into the parametrial tissue. An
accurate measure of the remaining wall thickness,
therefore, may be difcult to report.
It has been shown that DNA quantitation is of
FIGURE 3. KaplanMeier plot for overall survival according
to corpus inltration in patients with FIGO Stage I disease.
Adenocarcinoma of the Cervix/Alfsen et al. 2481
prognostic benet to patients with adenocarcinomas
of the uterine corpus
42
and epithelial tumors of the
ovary.
15,43
Studies on the impact of DNA quantitation
by ow cytometry for prognosis in patients with cer-
vical adenocarcinomas have been few and have pro-
vided divergent results.
36,44,45
A lack of inuence from
ow cytometric variables on the prognosis of patients
with SCC of the uterine cervix was reported previously
by our institution.
46
That earlier report also included
patients with adenocarcinomas and adenosquamous
carcinomas, although the numbers of patients were
small. The current study of 292 patients conrms our
earlier ndings that DNA ploidy has no independent
inuence on overall survival for patients with carcino-
mas of the uterine cervix, regardless of histologic type.
In conclusion, small cell carcinomas were the only
histologic subtype of independent prognostic impor-
tance in patients with non-SCC of the uterine cervix.
The lack of signicance for histopathologic subtyping
of adenocarcinomas indicates the need for a simpli-
ed nomenclature for these tumors of the uterine
cervix. The distinction of EC or other mucinous tu-
mors from endometrioid tumors is of interest to pa-
thologists but is of no prognostic value.
The results of the current study also indicate clin-
ically important differences in staging of patients with
cervical SCC and adenocarcinomas. In patients with
FIGO Stage I disease, tumor inltration of cervical
adenocarcinoma into the corpus should be acknowl-
edged as consistent with high-risk disease and should
have implications for further treatment.
REFERENCES
1. Anton CH, Bloss JD, Bringman D, Lee FA, DiSaia P, Manetta
A. Comparison of adenocarcinoma and squamous cell car-
cinoma of the uterine cervix: a population-based epidemi-
ologic study [see comments]. Am J Obstet Gynecol 1992;166:
150714.
2. Alfsen GC, Thoresen SO, Kristensen GB, Skovlund E, Abeler
VM. Histopathologic subtyping of cervical adenocarcinoma
reveals increasing incidence rates of endometrioid tumors
in all age groups. A population based study with review of all
nonsquamous cervical carcinomas in Norway from 1966 to
1970, 1976 to 1980, and 1986 to 1990. Cancer 2000;89:12919.
3. Bjorge T, Thoresen SO, Skare GB. Incidence, survival and
mortality in cervical cancer in Norway, 19561990. Eur J
Cancer 1993;29A:22917.
4. Devesa SS, Young-JL J, Brinton LA, Fraumeni-JF J. Recent
trends in cervix uteri cancer. Cancer 1989;64:218490.
5. Eide TJ. Cancer of the uterine cervix in Norway by histologic
type, 197084. J Natl Cancer Inst 1987;79:199205.
6. Peters RK, Chao A, Mack TM, Thomas D, Bernstein L, Hen-
derson BE. Increased frequency of adenocarcinoma of the
uterine cervix in young women in Los Angeles County. J Natl
Cancer Inst 1986;76:4238.
7. Robertson JH, Woodend B. The changing pattern of cervical
cancer in Northern Ireland 19651989. Ulster Med J 1992;61:
1923.
8. Schwartz SM, Weiss NS. Increased incidence of adenocarci-
noma of the cervix in young women in the United States.
Am J Epidemiol 1986;124:10457.
9. Chen RJ, Chang DY, Yen ML, Lee EF, Huang SC, Chow SN, et
al. Prognostic factors of primary adenocarcinoma of the
uterine cervix. Gynecol Oncol 1998;69:15764.
10. Eifel PJ, Burke TW, Morris M, Smith TL. Adenocarcinoma as
an independent risk factor for disease recurrence in patients
with Stage IB cervical carcinoma. Gynecol Oncol 1995;59:38
44.
11. Hopkins MP, Schmidt RW, Roberts JA, Morley GW. Gland
cell carcinoma (adenocarcinoma) of the cervix. Obstet Gy-
necol 1988;72:78995; 3031, 87.
12. Kurman JR, Norris HJ, Wilkinson E. Tumors of the cervix,
vagina, and vulva. In: Rosai J, Sobin LH, editors. Atlas of
tumor pathology. Washington, DC: Armed Forces Institute
of Pathology, 1992.
13. Burghardt E. Microinvasive carcinoma in gynaecological pa-
thology. Clin Obstet Gynaecol 1984;11:23957.
14. Hedley DW, Friedlander ML, Taylor IW, Rugg CA, Musgrove
EA. Method for analysis of cellular DNA content of parafn-
embedded pathological material using ow cytometry.
J Histochem Cytochem 1983;31:13335.
15. Kaern J, Trope C, Kjorstad KE, Abeler V, Pettersen EO. Cel-
lular DNA content as a new prognostic tool in patients with
borderline tumors of the ovary. Gynecol Oncol 1990;38:
4527.
16. International Federation of Gynecology and Obstetrics.
Classication and staging of malignant tumours in the fe-
male pelvis. Int J Gynaecol Obstet 1965;3:2067.
17. Sobin LH, Fleming ID. TNM classication of malignant tu-
mors, fth edition (1997). Union Internationale Contre le
Cancer and the American Joint Committee on Cancer. Can-
cer 1997;80:18034.
18. Berek JS, Hacker NF, Fu YS, Sokale JR, Leuchter RC, Lagasse
LD. Adenocarcinoma of the uterine cervix: histologic vari-
ables associated with lymph node metastasis and survival.
Obstet Gynecol 1985;65:4652.
19. Ishikawa H, Nakanishi T, Inoue T, Kuzuya K. Prognostic
factors of adenocarcinoma of the uterine cervix. Gynecol
Oncol 1999;73:426.
20. Cohn DE, Peters WA, Muntz HG, Wu R, Greer BE, Tamimi
HK, et al. Adenocarcinoma of the uterine cervix metastatic
to lymph nodes. Am J Obstet Gynecol 1998;178:11317.
21. Costa MJ, McIlnay KR, Trelford J. Cervical carcinoma with
glandular differentiation: histological evaluation predicts
disease recurrence in clinical Stage I or II patients. Hum
Pathol 1995;26:82937.
22. Goodman HM, Buttlar CA, Niloff JM, Welch WR, Marck A,
Feuer EJ, et al. Adenocarcinoma of the uterine cervix: prog-
nostic factors and patterns of recurrence. Gynecol Oncol
1989;33:2417.
23. Hopkins MP, Morley GW. A comparison of adenocarcinoma
and squamous cell carcinoma of the cervix. Obstet Gynecol
1991;77:9127.
24. Hurt WG, Silverberg SG, Frable WJ, Belgrad R, Crooks LD.
Adenocarcinoma of the cervix: histopathologic and clinical
features. Am J Obstet Gynecol 1977;129:30415.
25. Kilgore LC, Soong SJ, Gore H, Shingleton HM, Hatch KD,
Partridge EE. Analysis of prognostic features in adenocarci-
noma of the cervix. Gynecol Oncol 1988;31:13753.
26. Kleine W, Rau K, Schwoeorer D, Peiderer A. Prognosis of
the adenocarcinoma of the cervix uteri: a comparative
study. Gynecol Oncol 1989;35:1459.
2482 CANCER November 1, 2001 / Volume 92 / Number 9
27. Korhonen MO. Adenocarcinoma of the uterine cervix. Prog-
nosis and prognostic signicance of histology. Cancer 1984;
53:17603.
28. Look KY, Brunetto VL, Clarke PD, Averette HE, Major FJ,
Alvarez RD, et al. An analysis of cell type in patients with
surgically staged Stage IB carcinoma of the cervix: a Gyne-
cologic Oncology Group study. Gynecol Oncol 1996;63:304
11.
29. Matthews CM, Burke TW, Tornos C, Eifel PJ, Atkinson EN,
Stringer CA, et al. Stage I cervical adenocarcinoma: prog-
nostic evaluation of surgically treated patients. Gynecol On-
col 1993;49:1923.
30. Saigo PE, Cain JM, Kim WS, Gaynor JJ, Johnson K, Lewis J.
Prognostic factors in adenocarcinoma of the uterine cervix.
Cancer 1986;57:158493.
31. Shingleton HM, Gore H, Bradley DH, Soong SJ. Adenocar-
cinoma of the cervix. I. Clinical evaluation and pathologic
features. Am J Obstet Gynecol 1981;139:799814.
32. Tamimi HK, Figge DC. Adenocarcinoma of the uterine cer-
vix. Gynecol Oncol 1982;13:33544.
33. Stockler M, Russell P, Mcgahan S, Elliott PM, Dalrymple C,
Tattersall M. Prognosis and prognostic factors in node-neg-
ative cervix cancer. Int J Gynecol Cancer 1996;6:47782.
34. Hale RJ, Wilcox FL, Buckley CH, Tindall VR, Ryder WDJ,
Logue JP. Prognostic factors in uterine cervical carcinoma: a
clinicopathological analysis. Int J Gynecol Cancer 1991;1:19
23.
35. Raju K, Kjorstad K, Abeler V. Prognostic factors in the treat-
ment of Stage IB adenocarcinoma of the cervix. Int J Gynecol
Cancer 1991;1:6974.
36. Lai C-H, Hsueh S, Hong J-H, Chang T-C, Tseng C-J, Chou
H-H, et al. Are adenocarcinomas and adenosquamous car-
cinomas different from squamous carcinomas in Stage IB
and II cervical cancer patients undergoing primary radical
surgery? Int J Gynecol Cancer 1999;9:2836.
37. Abeler VM, Holm R, Nesland JM, Kjorstad KE. Small cell
carcinoma of the cervix. A clinicopathologic study of 26
patients. Cancer 1994;73:6727.
38. Kjorstad KE, Bergstrom J, Abeler V. Clear cell adenocarci-
noma of the cervix uteri and vagina in young women in
Norway [in Norwegian]. Tidsskr Nor Laegeforen 1989;109:
16347.
39. Chen RJ, Lin YH, Chen CA, Huang SC, Chow SN, Hsieh CY.
Inuence of histologic type and age on survival rates for
invasive cervical carcinoma in Taiwan. Gynecol Oncol 1999;
73:18490.
40. Kamura T, Tsukamoto N, Tsuruchi N, Saito T, Matsuyama T,
Akazawa K, et al. Multivariate analysis of the histopathologic
prognostic factors of cervical cancer in patients undergoing
radical hysterectomy. Cancer 1992;69:1816.
41. Kishi Y, Hashimoto Y, Sakamoto Y, Inui S. Thickness of
uninvolved bromuscular stroma and extrauterine spread
of carcinoma of the uterine cervix. Cancer 1987;60:23316.
42. Raju K, King RJB, Kaern J, Sumner D, Abeler VM, Mandalaya
S, et al. Inuence of HSP27 and steroid receptor status on
provera sensitivity, DNA-ploidy and survival of females with
endometrial cancer. Int J Cynecol Cancer 1995;5:94100.
43. Vergote IB, Kaern J, Abeler VM, Pettersen EO, De-Vos LN,
Trope CG. Analysis of prognostic factors in stage I epithelial
ovarian carcinoma: importance of degree of differentiation
and deoxyribonucleic acid ploidy in predicting relapse. Am J
Obstet Gynecol 1993;169:4052.
44. Leminen A, Paavonen J, Vesterinen E, Forss M, Wahlstrom
T, Kulomaa P, et al. Deoxyribonucleic acid ow cytometric
analysis of cervical adenocarcinoma: prognostic signi-
cance of deoxyribonucleic acid ploidy and S-phase fraction.
Am J Obstet Gynecol 1990;162:84853.
45. Kaspar HG, Dinh TV, Hannigan EV. Flow cytometry in cer-
vical adenocarcinoma. Correlating DNA content analysis
with the clinical picture. J Reprod Med 1997;42:1702.
46. Kristensen GB, Kaern J, Abeler VM, Hagmar B, Trope CG,
Pettersen EO. No prognostic impact of ow-cytometric
measured DNA ploidy and S-phase fraction in cancer of the
uterine cervix: a prospective study of 465 patients. Gynecol
Oncol 1995;57:7985.
Adenocarcinoma of the Cervix/Alfsen et al. 2483