BJD

T H E R A P E U T I CS British Journal of Dermatology

Comparison of long-term drug survival and safety of

biologic agents in patients with psoriasis vulgaris*
R. Gniadecki,1 B. Bang,1 L.E. Bryld,2 L. Iversen,3 S. Lasthein4 and L. Skov5
1
Department of Dermatology, Bispebjerg Hospital, Bispebjerg, Denmark
2
Department of Dermatology Roskilde Hospital, University of Copenhagen, Copenhagen, Denmark
3
Department of Dermatology, Marselisborg Hospital, Aarhus University, Aarhus, Denmark
4
Department of Dermatology, Odense Hospital, University of Southern Denmark, Odense, Denmark
5
Department of Dermatology Gentofte Hospital, University of Copenhagen, Copenhagen, Denmark

Summary

Correspondence
Robert Gniadecki.
E-mail: r.gniadecki@gmail.com
Accepted for publication
6 August 2014
Funding sources
No external funding.
Conflicts of interest
R.G. reports carrying out clinical trials for Abbvie,
MSD, Celgene, Novartis, Janssen and Pfizer and
has had paid consultancies and lectures from Abb-
vie, MSD, Janssen and Pfizer. B.B. has been a
member of the advisory boards and a consultant
for Abbvie, Biogen Idec, Janssen and MSD and an
investigator in clinical trials for Eli Lilly and Pfiz-
er; B.B. is currently employed by LEO Pharma.
L.E.B. has no conflicts of interests. L.I. has had
paid lectureships and consultancies for Abbvie, Al-
mirall, Janssen, LEO Pharma, MSD, Novo Nor-
disk, Otsuka, Pfizer and UCB Nordic and has
performed clinical trials for Abbvie, Amgen, Cell-
gene, MSD, Pfizer, Eli Lilly and Novartis. S.L. has
had paid consultancies from Abbvie and Galderma
Nordic. L.S. has served as investigator, advisory
board member and paid consultant for Abbvie,
Amgen, Astellas, Biogen Idec, Eli Lilly, Janssen,
LEO Pharma, MSD, Novartis and Pfizer. The
authors do not have equity in pharmaceutical
companies.
*Plain language summary available online.
DOI 10.1111/bjd.13343
Background Drug survival (time to drug discontinuation) has recently emerged as
an important parameter reflecting the long-term therapeutic performance in a
real-life setting. Biologic drug survival in psoriasis is mainly limited by a gradual
loss of efficacy over time. Previous studies have been limited by small patient
population size and short observation times and yielded discrepant survival times
for different biologics.
Objectives To calculate the long-term drug survival for adalimumab, etanercept,
infliximab and ustekinumab in a large cohort of real-life patients with psoriasis
vulgaris and to analyse the factors that influence drug survival.
Patients and methods Data were extracted from the prospective registry DERMBIO
covering all patients with psoriasis vulgaris treated with biologic agents in the
academic centres in Denmark. Drug survival was analysed using the KaplanMeier
method. The influence of different covariates on drug survival was analysed by
Cox regression.
Results Included in the analysis were 1867 treatment series (adalimumab n = 774,
etanercept n = 449, infliximab n = 253, ustekinumab n = 391) administered in
1277 patients for up to 10 years. Drug survival was significantly longer for us-
tekinumab than for anti-tumour necrosis factor (TNF)-a agents (P < 0
001). Eta-
nercept had the shortest survival time [median survival 30 months, 95%
confidence interval (CI) 25
134
9] whereas adalimumab and infliximab had
comparable survival rates (59 months, 95% CI 45
672
4; 44 months, 95% CI
3354
9, respectively). Survival was longer in men [odds ratio (OR) 1
51, 95%
CI 1
311
74 vs. women] and in patients who had not previously received any
biologic agent (OR 1
24, 95% CI 1
051
46). Loss of efficacy accounted for 67%
of all drug discontinuations.
Conclusions Ustekinumab has a significantly longer drug survival than the anti-
TNF-a agents. Switching from one biologic to another is associated with an
impairment of drug survival. Preventing loss of efficacy is a major area of medi-
cal need in the biologic therapy of psoriasis and the strategies that improve drug
survival should be further investigated.

Whats already known about this topic?

Gradual loss of efficacy of biologic agents has been observed during long-term
treatment of psoriasis.
Previous studies have had limited numbers of patients and short observation times
and showed conflicting data.

244 British Journal of Dermatology (2015) 172, pp244252 2014 British Association of Dermatologists

What does this study add?
This is the largest study to date, comprising 1277 patients followed for up to
10 years.
Long-term drug survival depends on the drug (ustekinumab > adalimumab = inf-
liximab > etanercept), sex (men > women) and previous exposure to another bio-
logic.
Almost a decade of biologic drug use in dermatology has pro-
ven their usefulness in the management of moderate and severe
psoriasis. Unlike most traditional systemic therapies the biolo-
gics do not exhibit cumulative toxicity and therefore have an
excellent safety profile in the long-term, continuous manage-
ment of patients.1 Unfortunately, there are still significant gaps
in the understanding of the long-term efficacy of biologic
drugs. Open label extension (OLE) studies have indicated that
most patients are likely to maintain the effect over years.24
Unfortunately, the results are difficult to translate to clinical
decisions, as statistics in most OLE studies rely on the last-
observation-carried-forward principle and may significantly
overestimate the true long-term efficacy. Indeed, the real-life
observational data suggest that loss of efficacy of the biologic
occurs in 1025% of patients annually.58 Another limitation
of the OLE studies is the choice of endpoint, which is usually
Psoriasis Area and Severity Index (PASI) reduction in compari-
son with the baseline PASI value before the administration of
the biologic [usually PASI50 or PASI75 (reduction in PASI score
by 50% or 75%)]. However, there is ample evidence that the
PASI reduction is not well correlated with patients perception
of treatment success and even patients who achieved PASI75
response do not consider this effect optimal.9,10
DERMBIO is a national registry of patients treated with bio-
logic drugs in Denmark, which allows for long-term monitor-
ing of treatment efficacy and safety.5 In our previous analysis
we have proposed the drug survival rate (measured as a prob-
ability to discontinue the drug) as a simple and clinically rele-
vant measure of the treatment outcome.5 We have
documented that the main reason for drug discontinuation is
loss of efficacy; this result has subsequently been confirmed
by independent studies.6,8,11,12
This paper provides a new analysis of the DERMBIO cohort,
now including 1277 patients treated with biologics (ada-
limumab, etanercept, infliximab and ustekinumab) over a 10-
year period. To our knowledge this is the largest dataset,
which shows the real-life treatment outcomes in a cohort of
patients in the whole country.
Patients and methods
Patient inclusion and exclusion
Patients who were included in this analysis are real-life
patients (i.e. excluding the patients participating in the clinical
2014 British Association of Dermatologists
Biologic survival in psoriasis, R. Gniadecki et al. 245
trials) who received biologic treatment for psoriasis vulgaris
in Denmark between March 2003 and June 2013. Us-
tekinumab has been available in Denmark since April 2009, so
the observation time for this drug has been shorter than for
the anti-tumour necrosis factor (TNF)-a agents. Eligibility for
the biologic treatment for plaque psoriasis is regulated
by the national guidelines of the Danish Dermatological
Society (http://2convert.dk/dds/wp-content/uploads/2012/
08/3bio_guides_dec_2011.pdf).5 Biologic treatments are
administered to the patients with moderate to severe psoriasis
[PASI > 10 or Dermatology Life Quality Index (DLQI) > 10
or affected body surface area > 10%] in whom treatments
previously failed or who have contraindications to topical
therapies, ultraviolet B phototherapy and methotrexate.
Patients who refuse methotrexate must have documented con-
traindication to or lack of efficacy of ciclosporin or acitretin.
Thus, according to the guidelines, patients with PASI < 10
might be included, if they reported DLQI > 10 or if they
switched from a conventional systemic therapy or another bio-
logic therapy which had had partial (yet insufficient) effect.
The criteria of the short-term therapeutic success of the ther-
apy were reduction of PASI by 50% (PASI50) or absolute PASI
value < 5 or DLQI < 10 at 3 months (2 weeks) after the
initiation of treatment, but the long-term success criteria were
at the discretion of the attending physician and the patient
and are not regulated by the guidelines. The choice of drug
was the decision of the physician, but for the patients with
concomitant psoriatic arthropathy (PsA) the guidelines sug-
gested use of a TNF-a blocker rather than ustekinumab as first
choice.
Data source and data extraction
The DERMBIO database was established in 2007 to monitor
the efficacy and side-effects of biologic drugs in Denmark.
This registry is nationwide and includes all biologic treatments
administered in academic hospital centres (88% of all treat-
ment series) and in private clinics (22% of the treatment ser-
ies). Available data for biologic treatments that had been
initiated before the launch of DERMBIO (n = 137 treatment
series) were added retrospectively to the registry. The aca-
demic hospital centres reported 100% of the administered
treatments series whereas the data for the coverage of the
treatments administered in private clinics has not been estab-
British Journal of Dermatology (2015) 172, pp244252
246 Biologic survival in psoriasis, R. Gniadecki et al.
lished but is probably higher than 80%. In order to preserve
the maximal data integrity we therefore decided to include
only data reported from the hospital centres.
Data in DERMBIO are organized as treatment sequences (ser-
ies) comprising a period of continuous treatment with a given
biologic. Thus, one patient may receive several treatment
sequences of different durations. Data from the entire database
was extracted by the administrator (Zitelab Ltd, Frederiksberg,
Denmark) to an Excel file, which for each treatment sequence
contained: (1) patients identification number, (2) sex, (3)
age, (4) weight, (5) diagnosis and duration of psoriasis, (6)
identity of the biologic agent and number of the treatment
sequence, (7) presence of PsA, (8) number of comorbidities,
(9) diagnosis of comorbidities (hypertension, ischaemic heart
disease, hypercholesterolaemia, diabetes, nonmelanoma skin
cancer, alcohol abuse, obesity), (10) date of treatment initia-
tion, (11) date of treatment discontinuation or last visit (in the
case of censored data), (12) reason for treatment discontinua-
tion (categories: adverse event, lack of efficacy, patients deci-
sion, lost to follow-up, other), (13) adverse events (categories:
infection, allergy or drug intolerance, skin rash, lupus like syn-
drome, other), (14) severity of the adverse event (minor or
major), (15) concomitant treatment with methotrexate.
In the present study we have included the treatment
sequences, which fulfilled the following criteria: valid patient
number enabling data verification, treatment with the cur-
rently approved biologics (adalimumab, etanercept, inflix-
imab, ustekinumab) and minimum treatment duration of
1 month. Interrupted series with the same biologic were
joined if the treatment break was < 1 month for adalimumab
and etanercept, 2 months for infliximab and 4 months for
ustekinumab.
Statistical analysis
Basic descriptive statistical analysis was done with Microsoft
Excel (2011 MAC version 14.3.9; Microsoft Corp, Redmond,
WA, U.S.A.) or GRAPHPAD PRISM version 5.00 for Mac
(GraphPad Software, San Diego, CA, U.S.A.). For all other analy-
ses the data were exported from the Excel file to SPSS statistical
software (version 22 for Mac, IBM, Armonk, NY, U.S.A.) envi-
ronment. Drug survival analysis was done using Cox logistic
regression with multiple covariates (described in Results).
Crude survival curves not adjusted for the covariates were com-
pared with MantelCox statistics and shown as KaplanMeier
plots. Odds ratios with 95% confidence intervals were calculated
as exp(B) values in SPSS using the proportional hazard model
for drug withdrawal. Frequency table comparisons were done
with Chi-square test. P < 0
05 was considered significant.
Results
Patient characteristics
We have identified 1277 patients who received biologic treat-
ment for plaque psoriasis in Denmark between 2003 and
British Journal of Dermatology (2015) 172, pp244252
2013. In total 1867 treatment series were administered: 774
treatment series with adalimumab, 449 with etanercept, 253
with infliximab and 391 with ustekinumab. There were 436
patients who received more than one treatment sequence with
a biologic.
Table 1 shows the basic patient characteristics as recorded
during the first treatment sequence. The groups of patients
were defined by the biologic chosen for the first treatment
sequence and they were comparable regarding age, male/
female ratio, weight, disease duration, and the baseline values
of PASI and DLQI. All groups had an overrepresentation of
male patients. The prevalence of PsA was high in the groups
treated with anti-TNF-a agents, which probably reflected the
selection of high-need patients for biologics. The frequency of
PsA was significantly lower in the ustekinumab group, reflect-
ing the priority of the anti-TNF-a agents for patients with PsA
in Danish clinical guidelines. Methotrexate was used as a con-
comitant therapy in 25
4% of cases, most commonly together
with infliximab (55
1%) and least frequently with us-
tekinumab (12
4%). A relatively high proportion of patients
had one or more comorbidities (arterial hypertension 23
4%,
obesity 23
2%, hypercholesterolaemia 15
0%, diabetes 10
1%,
ischaemic heart disease 5
9%, alcohol abuse 4
0% or other/
nonspecified 15
1%).
The short-term effect of treatment measured as the PASI
reduction after the initial 34 months of treatment was com-
parable to what could be predicted from the clinical trials with
the biologics.13 The highest proportion of PASI75 responders
(76
7%) was achieved for infliximab whereas the lowest
(53
8%) was recorded for etanercept. There was no statistical
difference between adalimumab and ustekinumab in PASI75
response rates (Table 1).
Causes of drug discontinuation
Of the total 1867 treatment series administered for 1 month
or longer, 772 treatment sequences (41
3%) were termi-
nated. The major reason for treatment termination was a loss
of efficacy, which was highest for etanercept (41
6% of all
treatment series) and lowest for ustekinumab (15
9%)
(P < 0
001, Chi-square test) (Fig. 1a). Rates of adverse
events leading to discontinuation was highest for infliximab
(14
6%) and lowest for ustekinumab (3
1%) (P < 0
001,
Chi-square test). The leading causes were minor infections
(32%, 41 treatment series), which were mostly associated
with adalimumab (46
3%; 2
5% of all administered treat-
ment sequences) and infliximab (31
7%; 5
1% of treatment
series) and were relatively infrequent for etanercept (14
6%;
1
3% of treatment series) or ustekinumab (7
3%; 0
8% of
treatment series) (P = 0
002, Chi-square test for the differ-
ence between the drugs). Other adverse events leading to
discontinuation were: skin rash or aggravation of skin symp-
toms (12%: seven patients on adalimumab, three on etaner-
cept and four on infliximab); 5
6% were reactions
considered to be intolerance or allergy (six patients on inf-
liximab, one on adalimumab) and there was one case of
2014 British Association of Dermatologists
 Biologic survival in psoriasis, R. Gniadecki et al. 247

Table 1 Patient characteristics on entry in the database (first biologic treatment)

Drug
Adalimumab Etanercept Infliximab Ustekinumab All
Number of patients (% total) 567 (44
4) 364 (28
5) 176 (13
8) 170 (13
3) 1277 (100)
Male/female (% male) 362/205 (63
8) 240/124 (65
9) 119/57 (67
6) 103/67 (60
6) 824/453 (64
5)
Agea 44
4 (13
4) 46
3 (15
0) 45
5 (13
9) 44
6 (14
7) 45
2 (14
2)
Weighta 87
4 (19
9) 88
6 (21
2) 92
0 (22
5) 89
6 (25
0) 88
6 (21
1)
Disease durationa 18
7 (12
2) 19
5 (12
8) 18
7 (13
4) 17
9 (13
4) 18
9 (12
6)
PsA (%) 216 (38
1) 144 (39
6) 77 (43
8) 24 (14
1) 461 (36
1)
DLQIa 12
6 (6
8) 11
9 (6
9) 13
9 (5
7) 11
5 (7
8) 12
5 (7
0)
Number of patients with comorbidities (%)
0 271 (47
8) 170 (46
7) 79 (44
9) 95 (55
9) 615 (48
2)
12 181 (31
9) 125 (34
3) 60 (34
1) 53 (31
2) 419 (32
8)
>2 27 (4
8) 20 (5
5) 21 (11
9) 24 (14
1) 92 (7
2)
Methotrexate (%) 124 (21
9) 82 (22
5) 97 (55
1) 21 (12
4) 324 (25
4)
PASI baselinea 12
5 (7
6) 12
6 (7
8) 15
8 (8
5) 11
4 (8
3) 12
8 (8
0)
PASI at 3 monthsa 3
0 (3
9) 4
9 (6
4) 2
9 (4
2) 2
7 (4
3) 3
5 (4
9)
PASI75 (% of subjects) 66
1 53
8b 76
7b 64
3 64
3

PsA, psoriatic arthropathy; DLQI, Dermatology Life Quality Index; PASI, Psoriasis Area and Severity Index; PASI75, at least 75% improvement
in PASI score. aMean (SD). bChi-square significant; adalimumab vs. ustekinumab NS; adalimumab vs. etanercept P < 0
01; adalimumab vs.
infliximab P < 0
05.

lupus-like syndrome (0
8%: adalimumab). Other reasons for
discontinuation such as loss to follow-up, patient decision or
other/unspecified were quite infrequent (Fig. 1). This pat-
tern did not differ between the first treatment sequence in
the biologic-naive patients and in the subsequent series of
re-treatment (Fig. 1b,c).
Predictive factors for drug survival
Previous analysis of the DERMBIO cohort suggested that
except for the type of drug, the significant negative predictors
of drug survival were the previous failure of the biologic
(naive patients having longest survival) and female sex.5 These
findings could be reproduced in the current data set (Table 2).
Both forward and backward Cox regression modelling (likeli-
hood ratio and Wald test) confirmed a shorter drug survival
in female patients and in those who were previously treated
with a biologic. However, the number of previous treatments
was not significant. Other nonsignificant covariates were: age,
weight, duration of psoriasis, presence of PsA, baseline PASI
and DLQI, concomitant methotrexate, or number of co-mor-
bidities. Comparison of the crude, unadjusted survival curves
with log-rank MantelCox test and the adjusted survival curves
(for the statistically significant covariates: sex and previous
biologic treatment) confirmed that ustekinumab was the drug
associated with the best chances of long-term survival
(Fig. 2a, Table 2). Etanercept was associated with the shortest
drug survival compared with any other drug (MantelCox test
P < 0
001, Cox regression P < 0
001) (Fig. 2a, Table 2). The
same results were reproduced when the dataset included only
the patients who discontinued the drug due to lack of efficacy
(not shown). This confirms that the loss of efficacy is the
dominant factor leading to the long-term discontinuation of
the biologic.
To further explore the differences in drug survival between
the naive patients and those who were re-treated, we com-
pared the adjusted and crude survival curves between these
groups (Fig. 2; Table 3). Although all drugs showed a trend
towards shorter survival in the non-naive patients (Table 3),
the significant difference was demonstrated only for us-
tekinumab and etanercept. Ustekinumab had lost its advantage
in patients pre-treated with another biologic and its survival
was equal to that of adalimumab or infliximab, but was still
superior to that of etanercept.
After previous exposure to a biologic there was a striking
shortening of ustekinumab survival. This was surprising
because we originally assumed that the survival of us-
tekinumab would not (or only slightly) be affected by the
previous failure of the anti-TNF-a agent as the mechanism
of action of ustekinumab differs from the TNF-a blockers.
We therefore asked whether the effect of ustekinumab was
better when given to the patients with a primary failure of
a TNF-a inhibitor. In our cohort 145 patients fulfilled
criteria of a primary lack of efficacy of a TNF-a blocker
(discontinuation of the drug within the first 6 months of
the first treatment sequence due to the lack of efficacy).
These patients were re-treated in the subsequent treatment
sequences with adalimumab (n = 61), etanercept (n = 45),
infliximab (n = 17) or ustekinumab (n = 22). In these
patients ustekinumab demonstrated similar survival times to
those of infliximab and adalimumab whereas etanercept had
a significantly shorter survival time (P < 0
01), when tested
by MantelCox test and the survival functions in the Cox
logistic regression model.

2014 British Association of Dermatologists British Journal of Dermatology (2015) 172, pp244252
248 Biologic survival in psoriasis, R. Gniadecki et al.

Table 2 Odds ratios for treatment termination. Data calculated from

(a) 80
Adverse event the Cox regression model using forward Wald method
% of treatment series Lost to follow-up
60
Lack of efficacy
95%
Other
40 Odds confidence P-
Patient decision
Covariate ratio interval value
20 Adalimumab vs. ustekinumab 1
77 1
392
26 < 0
0001
Etanercept vs. ustekinumab 2
55 1
983
29 < 0
0001
0 Infliximab vs. ustekinumab 1
99 1
502
63 < 0
0001
<
ab

ab

ab
pt

Etanercept vs. adalimumab 1
42 1
201
68 0
0001

ce
um

m
xi
er

nu
im

fli
Etanercept vs. infliximab 1
30 1
041
61 0
02
an

ki
in
al

te
et
ad

us
Naive vs. previously exposed 1
24 1
051
46 0
011
(b) Male vs. female 1
51 1
311
74 < 0
0001
100
Adverse event
% of treatment series

80 Lost to follow-up
Lack of efficacy
60 Other
Patient decision SAEs were recorded by the physicians as related or possibly
40
related to the treatment.
20

0 Discussion
ab

ab

ab
pt
ce
um

Long-term efficacy and survival of biologic drugs have previ-

xi
er

nu
im

fli
an

ki
in
al

te
et
ad

ously been reported for patients with psoriasis in a real-life

us

(c) setting by us and other investigators.58,11,12,14 The main

80
Adverse event strength of the present analysis is an unprecedentedly long
% of treatment series

Lost to follow-up
60 observation time of 10 years and a large number of patients
Lack of efficacy
Other
treated with each biologic (1277 patients vs. 119650 patients
40 in previously published studies). The DERMBIO registry is
Patient decision
prospective, nationwide and has a complete coverage of the
20
patients with psoriasis treated with the biologic drugs in the
0
hospital departments of dermatology in Denmark. This is
another important advantage of the current work, as previous
ab

ab

ab
pt
ce
um

studies originated from single centres7,8,11 or few centres,6

xi
er

nu
im

fli
an

ki
in
al

did not cover sufficient numbers of patients on each drug7,8,11

te
et
ad

us

or excluded subgroups of patients.6

Fig 1. Reasons for termination of biological treatment in all treatment
series (a), in biologically naive patients (b) and in patients previously
exposed to a biologic (c). Data are shown as proportions of the
treatment series for a given drug.
Safety
Minor adverse events were relatively common and were
recorded in 1022 treatment sequences but they were the cause
of treatment termination in only 99 series (9
7%). As could
be expected from the above-described pattern of adverse
events leading to termination, 43
8% were nonserious infec-
tions, 8
6% comprised skin rash or exacerbation of skin symp-
toms, 1
7% were allergic reactions or drug intolerance, 0
6%
lupus-like syndrome. The rest (45
3%) were not classified.
Serious adverse events (SAEs) were listed separately
(Table 4). In total 33 SAEs were recorded in the period on
the four biologic drugs (adalimumab 15, infliximab 6,
ustekinumab 2 and eternacept 10). Among these, nine were
infections, nine were different kinds of common neoplasms
and three were allergic reactions to the treatment. Twenty-two
Our initial study revealed that the discontinuation of the
biologic drugs during long-term treatment is mainly driven by
the gradual loss of efficacy. Current analyses as well as other
independent studies6,11 confirm this conclusion. On average,
the median survival for the biologics in psoriasis was
47 months and 67% of all discontinuations were attributed to
loss of efficacy (Fig. 1). Adverse events accounted for only
9
7% of all drug discontinuations. Examination of the shape
of the survival curves in Figure 2 does not reveal any obvious
plateau in the time frame of this study. This suggests that the
loss of efficacy is a stochastic phenomenon, which happens at
a relatively constant rate throughout the whole treatment per-
iod. Similar shapes of the curves were reported for the drug
discontinuation rates for patients with psoriatic arthritis and
rheumatoid arthritis. Further research is needed to determine
the cause of the loss of efficacy. Induction of anti-drug anti-
bodies, compensatory induction of other proinflammatory
cytokines, or interindividual variability in drug metabolism
can be suggested as possible causes.
The statistically significant predictors of drug discontinua-
tion were the type of biologic agent, patients sex and previ-
ous exposure to a biologic. Ustekinumab appeared to have

British Journal of Dermatology (2015) 172, pp244252 2014 British Association of Dermatologists
 Biologic survival in psoriasis, R. Gniadecki et al. 249

(a) (b)

(c) (d)

(e) (f)

Fig 2. Drug survival for the biologic agents in psoriasis. (a, b) Drug survival for all treatment sequences from the DERMBIO database (n = 1867).
Data include 774 adalimumab treatment series (n = 307 terminated sequences), 449 etanercept series (n = 257 terminated), 253 infliximab series
(n = 122 terminated) and 391 ustekinumab series (n = 86 terminated). The Cox model in panel (a) shows theoretically predicted survival curves
adjusted for sex and previous biologic treatment whereas panel (b) shows the actual survival curves (KaplanMeier analysis). Analogous analysis
for the biologically naive patients (first treatment sequence; n = 1277, 544 terminated sequences) (c, d) and for the patients pre-treated with at
least one biologic (n = 576; 221 terminations) (e, f).

2014 British Association of Dermatologists British Journal of Dermatology (2015) 172, pp244252
250 Biologic survival in psoriasis, R. Gniadecki et al.

Table 3 Comparison of naive and biologic-exposed patients

Estimated survival time (months)

Nave patients Previously exposed to a biologic All patients
Druga Median (SE) 95% CI Median (SE) 95% CI Median (SE) 95% CI
c
Adalimumab 59 (8
1) 43
174
9 50 (8
2) 34
066
0 59 (6
9) 45
672
4
Etanercept 33 (4
1)b 25
041
0 18 (2
3)b 13
622
4 30 (2
5)b 25
134
9
Infliximab 47 (4
8) 37
656
4 36 (10
0) 16
455
6 44 (5
6)c 33
054
9

SE, standard error; CI, confidence interval. a50% survival not computable for ustekinumab. bSignificant difference with adalimumab
(P < 0
001) and infliximab (P = 0
008) (logrank MantelCox test). cWorse than ustekinumab P < 0
001.

Table 4 Number of serious adverse events

Serious adverse
events Adalimumab Infliximab Eternacept Ustekinumab Total
Total 15 6 10 2 33
Infection 6 2 1 0 9
Cancer 4 2 2 1 9
Cardiovascular 2 1 3 0 6
Other 3 1 4 1 9
Related or possibly related to treatment 9 3 10 0 22

superior capacity to maintain the efficacy long-term, even after
compensation for other covariates. However, it is important to
notice that the observation time for ustekinumab was only
5 years. Etanercept was associated with the shortest drug sur-
vival. There were no statistical differences between ada-
limumab and infliximab.
These differences are of potential clinical importance but
should be interpreted with caution. The relative capacity of
the biologic to retain the efficacy long term depends not only
on the inherent property of the drug but also on the treatment
schedule, dosing and the reasons that govern the decisions to
switch to another therapy. Dose adjustment of biologics is a
common clinical practice in Denmark. We have previously
shown that therapeutic effect of infliximab can be recaptured
in a clinical setting by the increase of the dose and shortening
intervals between injections.15 Similarly, doubling of the dose
of ustekinumab helps to maintain optimal therapeutic effect in
a number of patients.16 Unfortunately, the precise dosing
schedule is very difficult to analyse, as many dose adjustments
are transient and have not been adequately reported to our
registry. Secondly, it is important to emphasize that in the
DERMBIO registry the decision of drug discontinuation is
based on a clinical judgement of its efficacy rather than objec-
tive measurement of disease activity. It is well established that
PASI correlates only poorly with the physicians and patients
perception of the severity of disease.9,10 The perception of
treatment success correlates very poorly with PASI reduction
or absolute PASI values but most patients will consider psoria-
sis PASI values higher than 45 as not satisfactorily treated.16
Thus, our data may not be directly comparable with the regis-
tries in which the objective measurements of psoriatic activity
dictate the choice of treatment.
These methodological differences may help to explain the
discrepancies between the findings in this study and the
results from other registries. Esposito et al.6 published their
findings from the Italian cohort documenting a superior drug
survival of etanercept, which is the opposite of our conclu-
sions. However, unlike our registry where all biologically trea-
ted patients with psoriasis are followed, the study of Esposito
et al. excluded the patients in whom the biologics were pre-
scribed outside the protocol and the patients who had previ-
ously been exposed to TNF-a inhibitors or who received
combination therapies. Only patients with standard dose regi-
mens were included. It is therefore conceivable that there
were important differences in the characteristics of patient
populations between our cohort and the study of Esposito
et al. Additionally, this study employed pre-defined objective
criteria of treatment response (PASI50 for the primary
response and loss of PASI50 for the secondary inefficacy) and
the majority (59
8%) received etanercept (in contrast to our
cohort where only 28
5% was treated with this drug). How-
ever, a significant proportion of patients do not consider
PASI50 as optimal and will require further treatment optimiza-
tion.9,10 Moreover, the median survival times for etanercept
reported by Esposito et al. (approximately 2000 days,
66 months) does not only disagree with our findings but also
with the longitudinal studies of van den Reek et al.8,12 who
reported median survival of etanercept in Dutch cohorts of
approximately 3640 months (after exclusion of the clinical
trial patients), a number that is within our confidence interval
of 2541 months.
The median drug survival of infliximab in the study of Es-
posito et al.6 was approximately 42 months and was in agree-
ment with our data (44 months). However, the median

British Journal of Dermatology (2015) 172, pp244252 2014 British Association of Dermatologists

survival of adalimumab was very short (23 months), which
was less than half of our value of 59 months and the
56 months reported from a Spanish cohort by L opez-Ferrer
et al.7 It can therefore be concluded that our data are sup-
ported by the independent studies from the Dutch and Spanish
cohorts and argue against the generalizability of the findings
of Esposito et al. on the patients in a real-life setting.
The novel aspect of our current study is the inclusion of us-
tekinumab, which is characterized by long treatment intervals
of 12 weeks and a different mechanism of action than ada-
limumab, etanercept or infliximab. In accordance with our
previous pilot study17 and a recent small observational study
by van den Reek et al.,12 ustekinumab showed a clear superi-
ority, surpassing any other anti-TNF-a agent in terms of long-
term efficacy. This finding was not entirely unexpected as the
5-year follow-up of 3117 patients in the open-label extension
study has been characterized by a very low drop-out rate of
6% per year and the 75% patient retention at 4 years.4,16 Our
analysis showed that 81
9% of patients who received us-
tekinumab as their first biologic remained on therapy 4 years
later. For all patients (biologically naive and retreated) the
retention rate was 70% at 4 years.
The reason for this excellent long-term efficacy has not
been elucidated, but cannot be explained by the short-term
efficacy, which was comparable with that of infliximab or
adalimumab. One of the explanations might be that us-
tekinumab is more often dose adjusted than adalimumab
(dose increase from 45 to 90 mg) due to flat pricing of us-
tekinumab in Denmark. It has been shown that such dose
adjustment will recapture efficacy in approximately 45% of
patients who would otherwise have lost the clinical
response.18
Patients sex and previous exposure to a biologic were other
significant predictors of drug survival. A striking observation
that male patients demonstrate better long-term drug survival
has been reported by our group and confirmed by others.6,8
Shorter drug survival in women is true for all biologic agents
including ustekinumab and has also been reported in the rheu-
matological literature for psoriatic arthritis and rheumatoid
arthritis.19,20 To our knowledge there is no satisfactory expla-
nation for this finding but because this phenomenon is
observed across different indications and biologic agents it is
likely to have a biological rather that psychological explanation.
The negative effect of previous biologic therapy is an inter-
esting and clinically important phenomenon. All biologics
tended to have a shorter drug survival in patients who previ-
ously received and discontinued another biologic, but the
effect was most pronounced for ustekinumab and etanercept.
As with the biologically naive patients the main reason was
loss of efficacy. The data are supported by Esposito et al.,6
who reported that the switch from another anti-TNF-a agent
to etanercept was associated with a dramatic reduction of drug
survival to 17
9 months, a figure that is almost identical to
our results (18
0 months). Lopez-Ferrer et al.7 found a signifi-
cant reduction of adalimumab survival in patients with prior
exposure to another anti-TNF-a agent. In psoriatic arthritis,
2014 British Association of Dermatologists
Biologic survival in psoriasis, R. Gniadecki et al. 251
the patients who switched the biologic agent had a median
drug survival of 1
3 years vs. 2
2 years for the patients who
remained on the same therapy.20 Interestingly, the number of
the previously used biologics was not a significant factor for
drug survival in our cohort.
There are various possible explanations for the shortening
of drug survival after switching from one biologic to another.
According to the immunization theory, the exposure to a bio-
logic causes production of anti-drug antibodies that would
impair the efficacy of the next drug. These anti-drug antibod-
ies are predominantly anti-idiotypic, which implies that they
will not react against a therapeutic antibody binding to an
unrelated target.21,22 However, we have clearly demonstrated
that the survival of ustekinumab [anti-p40 interleukin (IL)-
12/23] is impaired following previous exposure to anti-TNF-
a agents targeting a structurally unrelated TNF-a. Another
putative explanation is different rates of drug catabolism influ-
encing drug levels. Studies on infliximab showed a good cor-
relation between serum drug levels and efficacy. It is
conceivable that some patients have a higher rate of antibody
catabolism, which is independent of its specificity. However,
the analysis of the survival curves do not show a clear plateau,
which suggests that the loss of efficacy appears in a stochastic
manner during long-term treatment. A third possible explana-
tion is an immunological reorchestration. It is conceivable that
a long-term suppression of a single cytokine will cause an
induction of other proinflammatory cytokines with the redun-
dant function. This theory would explain the loss of efficacy
of ustekinumab in patients previously exposed to anti-TNF-a
agents provided that TNF-a and IL-12/23 proinflammatory
signalling pathways are interconnected. Our observation that
the patients who do not primarily respond to anti-TNF-a
agents show a diminished response to ustekinumab is also
compatible with this theory.
In summary, our data clearly show that the long-term treat-
ment success is mainly limited by the progressive loss of drug
efficacy, which in turn seems to be dissociated in its short-
term effect. Long-term efficacy seems to be further impaired
in patients who switched from one biologic to another. There
is a need for better understanding of the factors affecting
long-term drug survival to address long-term treatment opti-
mization in psoriasis.
References
1 Rustin MHA. Long-term safety of biologics in the treatment of
moderate-to-severe plaque psoriasis: review of current data. Br J
Dermatol 2012; 167 (Suppl.):311.
2 Gordon K, Papp K, Poulin Y et al. Long-term efficacy and safety of
adalimumab in patients with moderate to severe psoriasis treated
continuously over 3 years: results from an open-label extension
study for patients from REVEAL. J Am Acad Dermatol 2012; 66:241
51.
3 Papp KA, Poulin Y, Bissonnette R et al. Assessment of the long-
term safety and effectiveness of etanercept for the treatment of
psoriasis in an adult population. J Am Acad Dermatol 2010; 66:
e3345.
British Journal of Dermatology (2015) 172, pp244252
252 Biologic survival in psoriasis, R. Gniadecki et al.
4 Kimball AB, Gordon KB, Fakharzadeh S et al. Long-term efficacy of
ustekinumab in patients with moderate-to-severe psoriasis: results
from the PHOENIX 1 trial through up to 3 years. Br J Dermatol
2012; 166:86172.
5 Gniadecki R, Kragballe K, Dam TN, Skov L. Comparison of drug
survival rates for adalimumab, etanercept and infliximab in
patients with psoriasis vulgaris. Br J Dermatol 2011; 164:10916.
6 Esposito M, Gisondi P, Cassano N et al. Survival rate of anti-
TNF alpha treatments for psoriasis in routine dermatological
practice: a multicenter observational study. Br J Dermatol 2013;
169:66672.
7 Lopez-Ferrer A, Vilarrasa E, Gich IJ, Puig L. Adalimumab for the
treatment of psoriasis in real life: a retrospective cohort of 119
patients at a single Spanish centre. Br J Dermatol 2013; 169:11417.
8 Van den Reek JMPA, van L umig PPM, Driessen RJB et al. Determi-
nants of drug survival for etanercept in a long-term daily practice
cohort of patients with psoriasis. Br J Dermatol 2014; 170:41524.
9 Mattei PL, Corey KC, Kimball AB. Psoriasis Area Severity Index
(PASI) and the Dermatology Life Quality Index (DLQI): the corre-
lation between disease severity and psychological burden in
patients treated with biological therapies. J Eur Acad Dermatol Venereol
2014; 28:3337.
10 Mehren CR, Clemmensen A, Dall AB-H et al. Essential factors influ-
encing health-related-quality of life in psoriasis. J Drugs Dermatol
2014; 13:24650.
11 Menting SP, Sitaram AS, Bonnerjee-van der Stok HM et al. Drug
survival not significantly different between biologics in patients
with psoriasis vulgaris: a single-centre database analysis. Br J Derma-
tol 2014; 171:87583.
12 van den Reek JM, Zweegers J, Kievit W et al. Happy drug survival
of adalimumab, etanercept and ustekinumab in psoriasis in daily
practice care results from the BioCAPTURE network. Br J Dermatol
2014; 171:118996.
13 Reich K, Burden AD, Eaton JN, Hawkins NS. Efficacy of biologics
in the treatment of moderate to severe psoriasis: a network meta-
analysis of randomized controlled trials. Br J Dermatol 2012;
166:17988.
British Journal of Dermatology (2015) 172, pp244252
14 Papoutsaki M, Talamonti M, Giunta A et al. The impact of method-
ological approaches for presenting long-term clinical data on esti-
mates of efficacy in psoriasis illustrated by three-year treatment
data on infliximab. Dermatology 2010; 221(Suppl. 1):437.
15 Mehren CR, Gniadecki R. Dose-creep of infliximab during psoriasis
treatment: an observational study. Acta Derm Venereol 2012; 92:355
7.
16 Kimball AB, Papp KA, Wasfi Y et al. Long-term efficacy of us-
tekinumab in patients with moderate-to-severe psoriasis treated for
up to 5 years in the PHOENIX 1 study. J Eur Acad Dermatol Venereol
2013; 27:153545.
17 Clemmensen A, Spon M, Skov L et al. Responses to ustekinumab in
the anti-TNF agent-nave vs. anti-TNF agent-exposed patients with
psoriasis vulgaris. J Eur Acad Dermatol Venereol 2011; 25:103740.
18 Reich K, Puig L, Paul C et al. One-year safety and efficacy of us-
tekinumab and results of dose adjustment after switching from
inadequate methotrexate treatment: the TRANSIT randomized trial
in moderate-to-severe plaque psoriasis. Br J Dermatol 2014;
170:43544.
19 Hetland ML, Christensen IJ, Tarp U et al. Direct comparison of
treatment responses, remission rates, and drug adherence in
patients with rheumatoid arthritis treated with adalimumab, eta-
nercept, or infliximab: results from eight years of surveillance of
clinical practice in the nationwide Danish DANBIO registry. Arthritis
Rheum 2010; 62:2232.
20 Glintborg B, Ostergaard M, Krogh NS et al. Clinical response, drug
survival, and predictors thereof among 548 patients with psoriatic
arthritis who switched tumor necrosis factor a inhibitor therapy:
results from the Danish Nationwide DANBIO registry. Arthritis
Rheum 2013; 65:121323.
21 Wolbink GJ, Aarden LA, Dijkmans BAC. Dealing with immunoge-
nicity of biologicals: assessment and clinical relevance. Curr Opin
Rheumatol 2009; 21:21115.
22 Van Schouwenburg PA, van de Stadt LA, de Jong RN et al. Ada-
limumab elicits a restricted anti-idiotypic antibody response in
autoimmune patients resulting in functional neutralisation. Ann
Rheum Dis 2013; 72:1049.
2014 British Association of Dermatologists