Risk of malignancies in patients with inflammatory bowel disease treated with thiopurines or anti-TNF alpha antibodies

ORIGINAL REPORT

Risk of malignancies in patients with inflammatory bowel diseasetreated with thiopurines or anti-TNF alpha antibodies‡

Florian Beigel1, Anni Steinborn1, Fabian Schnitzler1, Cornelia Tillack1, Simone Breiteneicher1,Jestinah Mahachie John2,3, Kristel Van Steen2,3, Rüdiger P. Laubender4,5,6, Burkhard Göke1, Julia Seiderer1,Stephan Brand1*,† and Thomas Ochsenkühn1†

1Department of Medicine II, University Hospital Munich-Grosshadern, Ludwig-Maximilians-University Munich, Munich, Germany2Systems and Modeling Unit, Montefiore Institute, University of Liege, Liege, Belgium3Bioinformatics and Modeling, GIGA-R, University of Liege, Liege, Belgium4Institute of Medical Informatics, Biometry and Epidemiology (IBE), Ludwig-Maximilians-University Munich, Munich, Germany5German Cancer Consortium (DKTK), Heidelberg, Germany6German Cancer Research Center (DKFZ), Heidelberg, Germany

ABSTRACTPurpose We aimed to analyse malignancy rates and predictors for the development of malignancies in a large German inflammatory boweldisease (IBD) cohort treated with thiopurines and/or anti-tumour necrosis factor (TNF) antibodies.Methods De novo malignancies in 666 thiopurine-treated and/or anti-TNF-treated IBD patients were analysed. Patients (n= 262) weretreated with thiopurines alone and never exposed to anti-TNF antibodies (TP group). In addition, patients (n= 404) were exposed to anti-TNF antibodies (TNF+ group) with no (7.4%), discontinued (80.4%) or continued (12.1%) thiopurine therapy.Results In the TP group, 20 malignancies were observed in 18 patients compared with 8 malignancies in 7 patients in the TNF+ group(hazard ratio 4.15; 95% CI 1.82–9.44; p= 0.0007; univariate Cox regression). Moreover, 18.2% of all patients in the TP group ≥50 yearsof age developed a malignancy, compared with 3.8% of all patients <50 years of age (p= 0.0008). In the TNF+ group, 6.5% of all patients≥50 years of age developed malignancies compared with 0.3% of all patients <50 years of age (p= 0.0007). In both groups combined,thiopurine treatment duration ≥4 years was associated with the risk for skin cancer (p= 0.0024) and lymphoma (p= 0.0005).Conclusions Our data demonstrate an increased risk for the development of malignancies in IBD patients treated with thiopurines incomparison with patients treated with anti-TNF antibodies with or without thiopurines. Copyright © 2014 John Wiley & Sons, Ltd.

key words—inflammatory bowel disease; infliximab; adalimumab; thiopurines; anti-TNF antibodies; malignancy; pharmacoepidemiology;lymphoma; skin cancer; azathioprine; cancer; carcinoma; melanoma; tumor

Received 28 May 2013; Revised 8 January 2014; Accepted 10 March 2014

INTRODUCTION

Immunosuppressive agents like thiopurines and anti-tumour necrosis factor (TNF) antibodies are widely used

for long-term maintenance therapy in patients with in-flammatory bowel disease (IBD). The most commonlyobserved side effects of these drugs are infections,1–3

infusion reactions/allergies,4,5 skin pathologies,6,7 andless frequently autoimmune phenomena (e.g. lupus-likesyndrome8) andmalignancies.9,10 Given that the majorityof IBD patients have an early disease onset, severe sideeffects like malignancies restrain some clinical practi-tioners from using these drugs for their patients, whichmay lead to the delayed use of immunosuppressive andimmunomodulatory drugs in appropriate patients.The safety profiles of the most commonly used med-

ication for IBD, thiopurines and anti-TNF antibodies,have been published in various studies.11,12 There isincreasing evidence that thiopurines may increase the

*Correspondence to: S. Brand, Department of Medicine II, University HospitalMunich-Grosshadern, Ludwig-Maximilians-University Munich, Marchioninistrasse15, D-81377 Munich, Germany. E-mail: [email protected]†These authors share senior authorship.‡Prior postings and presentations: Parts of this work have been presented asan oral presentation at the United European Gastroenterology Week 2011in Stockholm and published in Gut Oct 2011; 60, Suppl. 3, A35, OP147,as an oral poster presentation at the 6th Congress of European Crohn’sand Colitis Organisation 2011 in Dublin and published in JCC Feb 2011;Vol. 5, Issue 1, S134, P293 and as a poster presentation at DigestiveDisease Week 2011 in Chicago and published in Gastroenterology 2011;140, (5), Suppl. 1, S773.

Copyright © 2014 John Wiley & Sons, Ltd.

pharmacoepidemiology and drug safety (2014)Published online in Wiley Online Library (wileyonlinelibrary.com) DOI: 10.1002/pds.3621

risk for the development of malignancies, particularlylymphoproliferative disorders.13,14 However, long-term data for anti-TNF antibodies regarding the riskfor the development of malignancies are missing.Given that a number of studies revealed safety alerts

regarding malignancies in patients treated with immu-nosuppressives and/or anti-TNF antibodies,9,10,13–17

we aimed in our study to analyse the incidence of malig-nancies in IBD patients treated with thiopurines aloneand anti-TNF antibodies with or without thiopurines ina large single-centre IBD cohort. In addition, we aimedto determine risk factors for the development of malig-nancies in IBD patients treated with thiopurines and/oranti-TNF agents.

METHODS

Study cohort

All patients to whom therapy with thiopurines oranti-TNF antibodies (infliximab or adalimumab) wasrecommended according to current national treatmentguidelines at our IBD centre in the time period from2002 to 2010 were included in this study (n=793). Ofthese 793 patients, 97 had not started the therapy becauseof unwillingness or incompliance. In 30 patients, nosufficient follow-up data were available. In total, clinicaldata of 666 patients were available for analysis. Theincidence rates of de novo solid and haematologicalmalignancies and malignancy-related deaths duringthiopurine or anti-TNF antibody treatment were reportedby retrospective analysis of medical records. Each malig-nancy was verified by histopathological reports. Patientswith a history of cancer prior to IBD therapy wereexcluded. The time of onset of the malignancy, concom-itant or previous medication and outcome were recorded.Follow-up was completed at our IBD centre.

Anti-TNF therapy and thiopurine treatment

At least the first three infliximab infusions (5mg/kgbody weight intravenously over 2 hours with inductiontherapy at 0, 2 and 6weeks and maintenance therapyevery 8weeks) or at least the first three injections ofadalimumab (160mg loading dose, following 80mgand then 40mg subcutaneously every other week) wereadministered in our outpatient centre. In the case ofthiopurine treatment, azathioprine was given in a doseof 2.0–2.5mg/kg. Patients starting thiopurine treatmentwere routinely advised to use sun-protective agents.

Statistical analysis

Survival analysis techniques were used to model thetime to malignancy development. We applied the

Kaplan–Meier product-limit estimator to estimate themedian survival time of follow-up and the Cox propor-tional hazard model to model the relationship betweentime to malignancy and covariates (i.e. to examine therelationship of the survival distribution to covariates).Additionally, we modelled the incidences for thepresence of malignancies for the patients adjusted forage and sex by using the Poisson regression and byestimating relative risks. All yearly incidence ratesare reported per 1000 patient-years.Comparison of demographic and clinical character-

istics of the two study populations was performedusing the χ2-test or Fisher’s exact test for categoricalvariables and by using the Wilcoxon–Mann–Whitneytest for continuous variables. All statistical analyseswere performed using R (R.2.13.2), and hypotheses(two sided) were tested at 5% level of significance.

RESULTS

Patients’ characteristics

Data of 666 patientswere available for analysis (Table 1).Overall, 262 patients (39.3%) were treated withthiopurines alone and were never exposed to anti-TNFantibodies (TP group). Four hundred and four patients(60.7%) were exposed to the anti-TNF antibodiesinfliximab, adalimumab or both (TNF+ group). In theTNF+ group, 30 patients (7.4%) never used thiopurines,325 patients (80.4%) had discontinued use and 49 patients(12.1%) had continued use of thiopurines as combina-tion therapy at the time of the last follow-up (Figure 1).All patients in the TNF+ group received infliximab

infusions (median 7, range 1–63), and 123 patientsin this group received adalimumab injections (median8, range 1–230), mainly because of intolerance orloss of response to infliximab. The median time ofthiopurine treatment was significantly longer in theTP group (median 19months, range 0–205months)than in the TNF+ group (median 12months, range0–240months; p = 0.004).There were no significant differences between both

groups regarding age, gender, body mass index andCrohn’s disease (CD)/ulcerative colitis distribution,but there were significant differences in age at diagnosis(median 31 years of age, range 6–81 years of age inthe TP group versus median 27 years of age, range7–71 years of age in the TNF+ group; p = 0.003) anddisease duration (median 5 years, range 0–35 years inthe TP group versus median 10 years, range0–51 years in the TNF+ group; p< 0.001). The medianfollow-up for patients was significantly shorter in theTP group with median 68weeks (range 0–447weeks)

f. beigel et al.

Copyright © 2014 John Wiley & Sons, Ltd. Pharmacoepidemiology and Drug Safety, (2014)DOI: 10.1002/pds

versus median 140weeks (range 0–465weeks) in theTNF+ group (p< 0.001).

Time to malignancy in both groups

We identified 18 patients with 20 malignancies in theTP group as compared with seven patients with eightmalignancies in the TNF+ group (hazard ratio (HR)4.15; 95% CI 1.82–9.4; p = 0.0007 in univariate Coxregression analysis; Figure 2). All seven patients in

the TNF+ group were male subjects and hadpreviously been exposed to thiopurines. The medianduration of thiopurine treatment in the 18 patients withmalignancies in the TP group was 43months (range0.5–205months), compared with 38months (range1–73months) of thiopurine treatment in the sevenpatients with malignancies in TNF+ group, whichwas statistically not significant (Figure 3; p = 0.506).However, one patient in the TP group had a very shortcourse of thiopurine treatment (0.5month), suggestingthat the development of malignancy was not primarilyrelated to the thiopurine intake. In the TP group, 11patients (61%) had continued thiopurine treatment atthe date of malignancy onset, while no patient in the

Figure 1. Study design. Data of 666 patients were available for analysis(469 patients with Crohn’s disease and 197 patients with ulcerative colitis).Two hundred and sixty-two patients were treated with thiopurines aloneand were never exposed to anti-TNF antibodies (TP group). Four hundredand four patients were exposed to the anti-TNF antibodies infliximab,adalimumab or both (TNF+ group). In the TNF+ group, 30 patients(7.4%) never used thiopurines, 325 patients (80.4%) had discontinued useand 49 patients (12.1%) had continued use of thiopurines as a combinationtherapy at the last follow-up

Figure 2. Kaplan–Mayer estimate for malignancy-free time intervals in thetwo study groups over the follow-up time. During the follow-up time, 20 ma-lignancies were observed in 18 patients in the TP group as compared witheight malignancies in seven patients in the TNF+ group (logrank p=0.0002)

Table 1. Demographic and clinical characteristics of the two study populations

Thiopurines only(TP group)

Anti-TNF-alpha antibodies ±thiopurines (TNF+ group) p-value

Patients n (%) 262 (39) 404 (61) n/aPatients with malignancy n (%) 18 (7) 7 (2) 0.001**Age (years) Median [range] 35 [14–84] 38 [14–76] 0.132*Age at onset of IBD (years) Median [range] 31 [6–81] 27 [7–71] 0.003*Disease duration (years) Median [range] 5 [0–38] 10 [0–51] <0.001*BMI (kg/m2) Median [range] 22.8 [14–41] 22.7 [13–39] 0.714*Male gender n (%) 137 (52) 202 (50) 0.587**Crohn’s disease n (%) 173 (66) 296 (73) 0.067**Ulcerative colitis n (%) 89 (34) 108 (27) 0.062**Thiopurine treatment evera n (%) 262 (100) 374 (93) < 0.001*Thiopurine treatment duration (months) Median [range] 19 [0–205] 12 [0–240] 0.004*Patients treated with infliximab n (%) 0 (0) 404 (100) n/aPatients treated with adalimumab n (%) 0 (0) 123 (30) n/aNumber of infliximab infusions Median [range] 0 [0] 7 [1–63] n/aNumber of adalimumab injections Median [range] 0 [0] 8 [1–230] n/aFollow-up (weeks) Median [range] 68 [0–447] 140 [0–465] <0.001*

n= 666.aIncludes patients with previous, discontinued and ongoing thiopurine therapy.*Mann–Whitney U test.**Fisher’s exact test.

MALIGNANCY RISK IN IBD TREATMENT


TNF+ group had continued thiopurine treatment at thedate of malignancy onset.

Types of malignancies

Figure 4A and Tables 2 and 3 show the different typesof malignancies. In the TP group, the most commonlyobservedmalignancies were skin cancers and lymphomas(Figure 4A–C). Six patients developed eight skinmalignancies in the TP group, while there was no skinmalignancy in the TNF+ group. Four lymphoproliferativemalignancies were recorded in the TP group ascompared with one in the TNF+ group (Figure 4A).This patient has also been treated with thiopurinesbefore anti-TNF treatment.In addition to skin cancers and lymphomas, eight

solid tumours were found in the TP group comparedwith seven solid tumours in the TNF+ group. Interest-ingly, five out of the seven malignancies in the TNF+group were urogenital malignancies (Figure 4A). Twopatients in the TP group and one patient in the TNF+group died from malignancy-related events.

Predictors of malignancies

To identify predictors of malignancies in both patientgroups, univariate regression analyses were used(Table 4). The variables ‘TP group’ (HR 4.15; 95%CI 1.82–9.44; p = 0.0007), age (HR 1.04; 95%CI1.01–1.08; p = 0.0039) and age at onset of IBD (HR1.04; 95% CI 1.01–1.07; p= 0.0031) were statisticallysignificant predictors for malignancy, while there wasa trend for male gender as an additional predictor formalignancy (HR 0.42; 95%CI 0.17–1.01; p=0.052). Inboth groups combined, the median age of patients with-out malignancies was 36 years of age (range 14–82yearsof age) compared with 52 years of age in patients withmalignancies (range 25–68years of age), p=3.26x10�5.The median age of all patients with malignancies in theTP group was 52.5 years of age (range 27–68years of

Figure 3. Thiopurine treatment duration in patients with malignancies. In theTP group, patients who developedmalignancies were treatedwith thiopurine fora median of 43months as compared with amedian of 38months in patients whodeveloped malignancies in the TNF+ group; * not significant

Figure 4. Comparison of malignancy types in the two study groups. (A)Distribution of malignancies in the two study groups. Twenty malignancieswere observed in 18 patients. The most commonly observed malignancieswere skin malignancies and lymphomas. Eight skin malignancies were foundin the TP group but none in the TNF+ group. In the TP group, there were threemelanoma and five non-melanoma skin cancers (three basal cell carcinoma,one spinalioma and one carcinoma in situ of the forehead). Fourlymphoproliferative malignancies were recorded in the TP group (twonon-Hodgkin lymphoma, one Hodgkin’s lymphoma and one plasmocytoma)as compared with one in the TNF+ group (one hepatosplenic T-celllymphoma). Eight solid tumours were found in the TP group (one mesotheli-oma, two breast cancers, one gastric carcinoma, one testicular carcinoma, onecolorectal cancer, one thyroid carcinoma and one prostate cancer) ascompared with seven in the TNF+ group (one cancer of unknown primary,histological squamous cell carcinoma, one penis carcinoma, one renal cellcarcinoma, two prostate cancer, one colorectal cancer and one seminoma).(B) Kaplan–Mayer estimate for lymphoma-free time interval in the two studygroups over the follow-up time. During the follow-up time, four lymphomaswere observed in the TP group as compared with one lymphoma in theTNF+ group (logrank p=0.006). (C) Kaplan–Mayer estimate for skin can-cer-free time interval in the two study groups over the follow-up time. Duringthe follow-up time, eight skin malignancies were observed in the TP group ascompared with no skin malignancies in the TNF+ group (logrank p=0.002)

f. beigel et al.


age) compared with 35.0 years of age (range 14–82yearsof age) in TP group patients without malignancy(p=0.001). The median age of all patients with malig-nancies in the TNF+ group was 57.0 years of age (range30–59 years of age) compared with 38.0 years of age(range 14–76 years of age) in TNF+ group patients with-out malignancy (p= 0.009). Stratified for age groups,18.2% of all patients older than 50 years of age in theTP group developed malignancies, but only 3.8% ofall patients younger than 50 years of age in the TP groupdevelopedmalignancies (p=0.0008). In the TNF+ group,6.5% of all patients older than 50 years developedmalignancies, but only 0.3% of all patients youngerthan 50 years of age in the TNF+ group developedmalignancies (p = 0.0007; Figure 5).The yearly incidence rate per 1000 patient-years for

malignancies dramatically increased with advanced ageand was more pronounced in male patients of the TPgroup (Figure 8). Most malignancies were recorded inmale patients between 40–60years of age (14 malignan-cies). Poisson regression (age is modelled as a continuouscovariate) estimated relative risks for the development ofmalignancy of 5.0 in the TP group (95% confidence in-terval 2.13–12.50) and 2.3 for male patients (95%confidence interval 0.98–5.60).The median duration of thiopurine treatment

was significantly higher in the TP group comparedwith that in the TNF+ group (19 vs 12months,

Wilcoxon–Mann–Whitney test, p=0.004). In the TPgroup, we found a longer thiopurine treatment durationin patients who developed skin cancers comparedwith other malignancy types (Figure 6; p = 0.049).In both groups, the majority of patients who developedmalignancies had a thiopurine treatment duration of >4years. In both groups combined, a thiopurine treatmentduration of >4 years was a risk factor for the develop-ment of skin cancer (p = 0.0024) and lymphoma(p = 0.0005) but not for the development of othermalignancies (p = 0.309; Figure 7).

DISCUSSION

Thiopurines and anti-TNF agents are the main therapeu-tic options for patients with moderate to severe IBD.However, there are only limited data on their long-termsafety and on predictors of malignancies in IBD patientsthat we aimed to investigate in this study. In this detailedanalysis of 666 IBD patients treated with thiopurinesand/or anti-TNF antibodies, we found a higher de novooccurrence of malignancies in thiopurine-treated IBDpatients (TP group) as compared with IBD patientstreated with anti-TNF antibodies alone or with previousor ongoing thiopurine treatment (TNF+ group). Of note,the follow-up time of the TP group was significantlyshorter than that of the TNF+ group, potentially

Table 2. Patients with malignancies in the TP group

Patient ID Diagnosis SexAge at onsetof IBD (years)

IBD duration(years)

Follow-uptime (months)

Duration ofthiopurinetreatment(months)

Age at onsetof malignancy

(years) MalignancyCancer-related

death

1 UC M 63 5 54 50 68 Mesothelioma Yes2 CD M 25 17 101 114 38 Basal cell carcinoma No3 CD F 24 8 76 77 27 Basal cell carcinoma

and melanomaNo

4 UC F 57 7 20 15 64 Breast cancer No5 CD M 38 18 43 0.5 52 Diffuse large

B-cell lymphomaNo

6 CD M 56 7 50 2 58 Diffuse big-cellularnon-Hodgkinlymphoma

No

7 CD M 25 3 1 11 27 Testicular carcinoma No8 UC M 39 6 4 57 44 Melanoma No9 CD M 32 32 102 24 47 Melanoma and

carcinoma in situNo

10 CD F 28 25 95 12 43 Breast cancer No11 CD M 28 20 41 36 47 Colorectal cancer Yes12 UC M 40 19 85 60 58 Hodgkin’s lymphoma No13 CD F 30 25 84 86 55 Plasmocytoma No14 CD M 36 21 86 8 57 Gastric carcinoma No15 UC F 33 10 94 88 43 Spinalioma No16 CD F 30 25 71 205 53 Basal cell carcinoma No17 CD F 34 25 102 12 57 Thyroid carcinoma No18 UC M 51 6 55 51 53 Prostate cancer No



evenunderestimating the rate of malignancies in the TPgroup. Most of the patients who developedmalignancies were older than 50 years of age and hadprevious or ongoing thiopurine treatment for more than4 years. In the TP group, the majority of the observedmalignancies were skin cancers and lymphomas.Therefore, our study raises safety concerns about thelong-term use of thiopurines in older IBD patients.Other novel findings of our study are the high preva-lence of melanomas in IBD patients treated withthiopurines alone and urogenital malignancies in IBDpatients treated with combined immunosuppression(anti-TNF antibodies and thiopurines).Our results are in agreement with recent studies

demonstrating thiopurines as a risk factor for thedevelopment of malignancies in IBD and otherautoimmune disorders.10,13,14,18 Previous studiesdemonstrated a high risk for the development of lym-phomas after thiopurine treatment16,19 and also othermalignancies in patients treated with azathioprine afterkidney transplantation.20,21 The potential of azathio-prine to influence the growth of human neoplasiawas also evaluated in mice and rats using 2-year bioas-says, which demonstrated increased incidences of skinmalignancies22,23 and lymphoma.24,25 In our study, weobserved five cases of lymphomas (four in the TPgroup and one in the TNF+ group); all of these pa-tients were male subjects, and most had long-term aza-thioprine therapy (>4 years).The mechanisms of the development of malignan-

cies in IBD patients treated with immunosuppressiveagents and anti-TNF antibodies are not completelyunderstood. It is assumed that smoking, age andcombination therapies increase the risk for severe sideeffects.14,26 Also, the Epstein–Barr virus has beenassociated with the development of lymphoma inthiopurine-treated IBD patients.27 In young malepatients, combination treatment regimens withthiopurines and anti-TNF antibodies may lead to the de-velopment of haepatosplenic T-cell lymphoma (HSTCL),an aggressive lymphomawith very poor prognosis. In ourcohort, we also observed one HSTCL in a patient withT

able

3.Patientswith

malignanciesin

theTNF+group

Patient

IDDiagnosis

Sex

Age

atonset

ofIBD

(years)

IBD

duratio

n(years)

Follow-up

time(m

onths)

Durationof

thiopurine

treatm

ent

(months)

Contin

ued

thiopurine

treatm

entat

onset

ofmalignancy

Age

atonset

ofmalignancy

(years)

No.

IFX

infusions

No.

ADA

injections

Malignancy

Cancer-

relateddeath

1CD

Male

1842

2838

No

599

n/a

CUP(histologically

squamouscell

carcinom

a)

No

2CD

Male

507

8560

No

558

3Peniscarcinom

aNo

3CD

Male

3622

8973

No

571

2Kidneycancer

andprostate

cancer

No

4CD

Male

264

491

No

3011

3Colorectalcancer

No

5CD

Male

2434

660

No

582

2HST

CL

Yes

6CD

Male

3720

7424

No

5730

n/a

Prostatecancer

No

7UC

Male

527

8020

No

5810

n/a

Sem

inom

aNo

Table 4. Univariate Cox regression model of predictors of malignancies

Hazard ratio 95%CI p-value

TP group 4.15 1.82–9.44 0.0007Age 1.04 1.01–1.08 0.0047Age at onset of IBD 1.04 1.01–1.07 0.0033Disease duration 1.01 0.96–1.06 0.7203Male gender 0.42 0.17–1.01 0.0527Crohn’s disease 0.79 0.34–1.84 0.5904Ulcerative colitis 1.26 0.54–2.94 0.5974Thiopurine treatment duration 1.00 0.99–1.01 0.9876

f. beigel et al.


long-term azathioprine treatment and subsequent anti-TNF antibody therapy, who died from HSTCL.28,29

A large retrospective nested case-control study withIBD patients linked the risk for the development ofnon-melanoma skin cancer with thiopurines andmelanoma with biologicals.30 In line with this and thefindings of other IBD studies,31–34 we found eight skinmalignancies in the TP group. However, no skin malig-nancies were observed in patients treated with anti-TNFantibodies with or without thiopurines in our study. Pre-viously, we conducted a survey on serious adverseevents in our first 100 patients treated with infliximaband found no malignancies.35 In line with these data,other IBD studies36 and a recently published largecohort study37 demonstrated that infliximab is welltolerated and the rate of malignancies did not differ frompatients who were not receiving infliximab. A largesingle-centre study of IBD patients found malignanciesin 2.9% of patients receiving infliximab and 4.5% ofpatients in the control group, which was statisticallynot significant.37 Other placebo-controlled trials withIBD patients treated with infliximab and adalimumab

revealed no increased incidences of cancer;38–40

however, follow-up periods were short.According to treatment guidelines, the majority of

anti-TNF-treated IBD patients have been treatedpreviously with thiopurines. Therefore, a direct com-parison of the malignancy risk of IBD patients withthiopurine monotherapy and anti-TNF monotherapymay be only realised as a large multicentre trial. Whilethe patients in the TP group of our study had thiopurinemonotherapy, the majority of the patients in theTNF+ group (93%) have been exposed to thiopurinestoo. However, only a small fraction of the TNF+group (12.1%) had concomitant, continued thiopurinetherapy. Remarkably, none of the patients with anti-TNF therapy only and no history of thiopurine treat-ment developed malignancies.A limitation of our and other IBD safety studies is the

lack of comparison groups not receiving immunosup-pressives or anti-TNF antibodies. However, comparisonof such patient groups would be biased because ofdifferences in disease type and severity. Moreover, ourtertiary centre study population characterised by a highpercentage of patients with severe IBD requiring immu-nosuppressive and/or immunomodulatory treatmentmay not reflect the general IBD population. Therefore,malignancy rates may be overestimated when ourresults are extrapolated to the general IBD population.Second, the influence of an immortal time bias couldnot be excluded, given that the follow-up starts at thedate of the first prescription of an anti-TNF drug inthe TNF+ group. However, the time period betweenthe start of thiopurine treatment and the start ofanti-TNF therapy is very short in our cohort (median13months), and on the other side, the time to malig-nancy is much longer in the majority of cases in theTP group. Hence, the immortal period, during whichthe patients in the anti-TNF group cannot incur themalignancy, is very short.

Figure 6. Thiopurine treatment duration in the TP group stratified for thetype of malignancy. In the TP group, the median of thiopurine treatment inpatients who developed skin cancers was 82.5months as compared with13.5months in patients who developed other malignancies; *p= 0.049

Figure 5. Percentage of patients with malignancies matched for age groups. Matched for age groups, 18.2% of all patients older than 50 years of age in the TPgroup developed malignancies compared with 3.8% of all patients younger than 50 years of age; *p= 0.0008. In the TNF+ group, 6.5% of all patients older50 years of age developed malignancies compared with 0.3% of all patients younger 50 years of age; **p= 0.0007



The American TREAT registry (prospective, multi-center, long-term infliximab registry in North Americafor CD patients) and other studies did not identify theuse of infliximab as an independent factor for the de-velopment of malignancies,41–43 while a recentanalysis of the FDA Adverse Event Reporting Systemon the incidence of T-cell non-Hodgkin lymphomarevealed an increased risk of lymphoma when usinganti-TNF antibodies in combination with thiopurinesbut not anti-TNF antibodies alone.44 Another study,which analysed pooled primary safety data from 10IBD landmark trials in infliximab-treated andimmunomodulator-treated patients, demonstrated thatplacebo-treated patients receiving thiopurines had ahigher incidence of malignancy versus no thiopurinetreatment,45 supporting the main findings of our study.However, in contrast to our study, the maximumfollow-up time was only 54weeks, given that the dataanalysis was based on the serious adverse eventreports of the studies45; therefore, long-term data arelacking in this meta-analysis.In our study, the majority of patients in both groups

who developedmalignancieswere treatedwith thiopurinesfor more than 4years, suggesting that long-term use ofthiopurines increases the risk for malignancies. However,in clinical practice, longer thiopurine treatment durationis often necessary because relapse rates after withdrawalof thiopurines are high.46,47 In IBD patients older than64 years of age, the risk and incidence of severe sideeffects are elevated, independently from the immunosup-pressive drugs that are used.48,49 Similar to the Clearingand Settlement Advisory and Monitoring Expert(CESAME) study16 that demonstrated age, male gender,disease duration and continued thiopurine therapy asindependent risk factors for lymphoproliferative disor-ders in IBD patients, in our study, age and continuedthiopurine use were predictors for the development of

malignancies in a univariate Cox regression model. Themedian age in patients who developed malignancies inboth groups was >50years of age. When data werematched for age groups, 18.2% of all patients older50 years of age in the TP group with continued thiopurineuse developed malignancies, while 6.5% of all patientsolder 50 years of age in the TNF+ group, in which themajority of patients discontinued thiopurine treatment,developed malignancies. The incidence rate per 1000patient-years for malignancies in our study was 10 inthe age group of male patients younger than 40 years ofage in the TP group but increased up to 60 in malepatients with the age of 40–60years of age (Figure 8).This is a considerably higher incidence rate than whatwas previously reported in the CESAME study,16 which,however, only reported the rates for lymphoma andshowed a maximum incidence rate of lymphoma per1000 patient-years of 5.41 in patients >65years of agewith continued thiopurine therapy. In contrast to oursingle-centre cohort study with a high-risk population

Figure 8. Yearly incidence rate per 1000 patient-years for malignancieswith dependence on age stratified by sex and medication

Figure 7. Percentage of patients with malignancies matched for thiopurine treatment duration. Duration of thiopurine treatment longer than 4 years was asignificant risk factor for the development of skin malignancies and lymphoma but not for the development of other cancers; *p= 0.0024, **p= 0.0005 and***p= 0.309, respectively

f. beigel et al.


from a tertiary IBD centre, the CESAME study was across-sectional study including a wide range of IBDpatients, resulting most likely in significant differencesregarding disease activity and treatment intensity.In conclusion, we observed significantly more

malignancies in IBD patients treated with thiopurinesalone as compared with patients treated with anti-TNFantibodies with or without thiopurine treatment.Predisposing factors for the development of malignanciesin both groups combined were continued thiopurine useand older age. Patients treated with thiopurines and/oranti-TNF antibodies older than 50 years of age andpatients with thiopurine treatment duration for more than4 years had an increased risk for the development ofmalignancies, particularly skin cancers and lymphomas.Further, long-term studies are required to analyse ifmonotherapy with anti-TNF antibodies has a betterrisk–benefit ratio compared with thiopurines or combina-tion therapy regarding the development of malignancies.

CONFLICT OF INTEREST

The authors declare no conflict of interest.

KEY POINTS• IBD patients with long-term thiopurine treatmenthave an increased malignancy risk and should bemonitored closely for malignancies, particularlyfor lymphoproliferative disorders and skin cancers.

• The risk–benefit ratio of thiopurine treatmentwith or without anti-TNF therapy, particularlyin male IBD patients older than 40 years of age,should be critically evaluated.

• Anti-TNF monotherapy as a first-line therapy shouldbe taken into account for selected patient groups.

ETHICS STATEMENT

This was a retrospective observational study of medicalrecords. Statistical data analysis was anonymised, andall patients were treated according current nationalguidelines. Therefore, the ethics committee of the Uni-versity of Munich was consulted, and a formal writtenwaiver for the need of ethics approval was obtained.

ACKNOWLEDGEMENTS

F. Beigel is supported by grants from the DeutscheForschungsgemeinschaft (DFG, BE 4490/2-1). S.Brand is supported by grants from the DeutscheForschungsgemeinschaft (DFG, BR 1912/6-1) and

the Else Kröner-Fresenius-Stiftung (Else KrönerExzellenzstipendium 2010; 2010_EKES.32). J.MahachieJohn and K. Van Steen acknowledge research opportuni-ties offered by the Belgian Network DYSCO (DynamicalSystems, Control, and Optimization), funded by theInteruniversity Attraction Poles Programme, initiated bythe Belgian State, Science Policy Office. Their workwas also supported in part by the IST Programme of theEuropean Community, under the PASCAL2 Network ofExcellence (Pattern Analysis, Statistical Modelling andComputational Learning), IST-2007-216886.This work contains parts of the unpublished doctoral

degree thesis of A. Steinborn.

REFERENCES

1. Rahier JF, Ben-Horin S, Chowers Y, et al. European evidence-based Consensuson the prevention, diagnosis and management of opportunistic infections in in-flammatory bowel disease. J Crohns Colitis 2009; 3: 47–91.

2. Schneeweiss S, Korzenik J, Solomon DH, Canning C, Lee J, Bressler B.Infliximab and other immunomodulating drugs in patients with inflammatorybowel disease and the risk of serious bacterial infections. Aliment PharmacolTher 2009; 30: 253–264.

3. Toruner M, Loftus EV, Jr, Harmsen WS, et al. Risk factors for opportunisticinfections in patients with inflammatory bowel disease. Gastroenterology 2008;134: 929–936.

4. Steenholdt C, Svenson M, Bendtzen K, Thomsen OO, Brynskov J, AinsworthMA. Acute and delayed hypersensitivity reactions to infliximab and adalimumabin a patient with Crohn’s disease. J Crohns Colitis 2012; 6: 108–111.

5. Sandborn WJ. A review of immune modifier therapy for inflammatory boweldisease: azathioprine, 6-mercaptopurine, cyclosporine, and methotrexate. Am JGastroenterol 1996; 91: 423–433.

6. Collamer AN, Guerrero KT, Henning JS, Battafarano DF. Psoriatic skin lesionsinduced by tumor necrosis factor antagonist therapy: a literature review andpotential mechanisms of action. Arthritis Rheum 2008; 59: 996–1001.

7. O’Donovan P, Perrett CM, Zhang X, et al. Azathioprine and UVA light generatemutagenic oxidative DNA damage. Science 2005; 309: 1871–1874.

8. Beigel F, Schnitzler F, Paul Laubender R, et al. Formation of antinuclear anddouble-strand DNA antibodies and frequency of lupus-like syndrome inanti-TNF-alpha antibody-treated patients with inflammatory bowel disease.Inflamm Bowel Dis 2011; 17: 91–98.

9. Biancone L, Calabrese E, Petruzziello C, Pallone F. Treatment with biologictherapies and the risk of cancer in patients with IBD. Nat Clin PractGastroenterol Hepatol 2007; 4: 78–91.

10. Smith MA, Irving PM, Marinaki AM, Sanderson JD. Review article: malignancyon thiopurine treatment with special reference to inflammatory bowel disease.Aliment Pharmacol Ther 2010; 32: 119–130.

11. Zabana Y, Domenech E, Manosa M, et al. Infliximab safety profile andlong-term applicability in inflammatory bowel disease: 9-year experience inclinical practice. Aliment Pharmacol Ther 2010; 31: 553–560.

12. Huang ML, Ran ZH, Shen J, Li XB, Xu XT, Xiao SD. Efficacy and safety ofadalimumab in Crohn’s disease: Meta-analysis of placebo-controlled trials. JDig Dis 2011; 12: 165–172.

13. Kandiel A, Fraser AG, Korelitz BI, Brensinger C, Lewis JD. Increased risk oflymphoma among inflammatory bowel disease patients treated with azathioprineand 6-mercaptopurine. Gut 2005; 54: 1121–1125.

14. Armstrong RG, West J, Card TR. Risk of cancer in inflammatory bowel diseasetreated with azathioprine: a UK population-based case-control study. Am JGastroenterol 2010; 105: 1604–1609.

15. Wolfe F, Michaud K. Biologic treatment of rheumatoid arthritis and the risk ofmalignancy: analyses from a large US observational study. Arthritis Rheum2007; 56: 2886–2895.

16. Beaugerie L, Brousse N, Bouvier AM, et al. Lymphoproliferative disorders inpatients receiving thiopurines for inflammatory bowel disease: a prospectiveobservational cohort study. Lancet 2009; 374: 1617–1625.

17. Krathen MS, Gottlieb AB, Mease PJ. Pharmacologic immunomodulation andcutaneous malignancy in rheumatoid arthritis, psoriasis, and psoriatic arthritis.J Rheumatol 2010; 37: 2205–2215.

18. Vos AC, Bakkal N, Minnee RC, et al. Risk of malignant lymphoma in patientswith inflammatory bowel diseases: a Dutch nationwide study. Inflamm BowelDis 2011; 17: 1837–1845.



19. Khan N, Abbas AM, Lichtenstein GR, Loftus EV, Jr., Bazzano LA. Risk of lym-phoma in patients with ulcerative colitis treated with thiopurines: a nationwideretrospective cohort study. Gastroenterology 2013; 145: 1007–1015 e1003.

20. Kinlen LJ, Sheil AG, Peto J, Doll R. Collaborative United Kingdom-Australasianstudy of cancer in patients treated with immunosuppressive drugs. Br Med J1979; 2: 1461–1466.

21. Watorek E, Boratynska M, Smolska D, Patrzalek D, Klinger M. Malignancy afterrenal transplantation in the new era of immunosuppression. Ann Transplant2011; 16: 14–18.

22. Reeve VE, Greenoak GE, Gallagher CH, Canfield PJ, Wilkinson FJ. Effect ofimmunosuppressive agents and sunscreens on UV carcinogenesis in the hairlessmouse. Aust J Exp Biol Med Sci 1985; 63(Pt 6): 655–665.

23. Kelly GE, Meikle W, Sheil AG. Effects of immunosuppressive therapy on the in-duction of skin tumors by ultraviolet irradiation in hairless mice. Transplantation1987; 44: 429–434.

24. Frankel HH, Yamamoto RS, Weisburger EK, Weisburger JH. Chronic toxicity ofazathioprine and the effect of this immunosuppressant on liver tumor inductionby the carcinogen N-hydroxy-N-2-fluorenylacetamide. Toxicol Appl Pharmacol1970; 17: 462–480.

25. Cohen SM, Erturk E, Skibba JL, Bryan GT. Azathioprine induction of lympho-mas and squamous cell carcinomas in rats. Cancer Res 1983; 43: 2768–2772.

26. Pedersen N, Duricova D, Elkjaer M, Gamborg M, Munkholm P, Jess T. Risk ofextra-intestinal cancer in inflammatory bowel disease: meta-analysis ofpopulation-based cohort studies. Am J Gastroenterol 2010; 105: 1480–1487.

27. Dayharsh GA, Loftus EV, Jr, Sandborn WJ, et al. Epstein-Barr virus-positivelymphoma in patients with inflammatory bowel disease treated with azathioprineor 6-mercaptopurine. Gastroenterology 2002; 122: 72–77.

28. Rosh JR, Gross T, Mamula P, Griffiths A, Hyams J. Hepatosplenic T-cell lym-phoma in adolescents and young adults with Crohn’s disease: a cautionary tale?Inflamm Bowel Dis 2007; 13: 1024–1030.

29. Beigel F, Jurgens M, Tillack C, et al. Hepatosplenic T-cell lymphoma in a patientwith Crohn’s disease. Nat Rev Gastroenterol Hepatol 2009; 6: 433–436.

30. Long MD, Martin CF, Pipkin CA, Herfarth HH, Sandler RS, Kappelman MD.Risk of melanoma and nonmelanoma skin cancer among patients with inflamma-tory bowel disease. Gastroenterology 2012; 143: 390–399.e391.

31. Peyrin-Biroulet L, Khosrotehrani K, Carrat F, et al. Increased Risk For Non-Melanoma Skin Cancers In Patients Who Receive Thiopurines For InflammatoryBowel Disease. Gastroenterology 2011; 141(5): 1621–28.e1-5.

32. Maddox JS, Soltani K. Risk of nonmelanoma skin cancer with azathioprine use.Inflamm Bowel Dis 2008; 14: 1425–1431.

33. Long MD, Kappelman MD, Pipkin CA. Nonmelanoma skin cancer in inflamma-tory bowel disease: a review. Inflamm Bowel Dis 17: 1423–1427.

34. Singh H, Nugent Z, Demers AA, Bernstein CN. Increased risk of nonmelanomaskin cancers among individuals with inflammatory bowel disease. Gastroenterol-ogy 2011; 141: 1612–1620.

35. Seiderer J, Goke B, Ochsenkuhn T. Safety aspects of infliximab in inflammatorybowel disease patients. A retrospective cohort study in 100 patients of a GermanUniversity Hospital. Digestion 2004; 70: 3–9.

36. Zabana Y, Domenech E, Manosa M, et al. Infliximab safety profile and long-term applicability in inflammatory bowel disease: 9-year experience in clinicalpractice. Aliment Pharmacol Ther 31: 553–560.

37. Fidder H, Schnitzler F, Ferrante M, et al. Long-term safety of infliximab for thetreatment of inflammatory bowel disease: a single-centre cohort study. Gut 2009;58: 501–508.

38. Hanauer SB, Feagan BG, Lichtenstein GR, et al. Maintenance infliximab forCrohn’s disease: the ACCENT I randomised trial. Lancet 2002; 359: 1541–1549.

39. Rutgeerts P, Sandborn WJ, Feagan BG, et al. Infliximab for induction and main-tenance therapy for ulcerative colitis. N Engl J Med 2005; 353: 2462–2476.

40. Colombel JF, Sandborn WJ, Panaccione R, et al. Adalimumab safety in globalclinical trials of patients with Crohn’s disease. Inflamm Bowel Dis 2009; 15:1308–1319.

41. Lichtenstein GR, Feagan BG, Cohen RD, et al. Serious infections and mortalityin association with therapies for Crohn’s disease: TREAT registry. ClinGastroenterol Hepatol 2006; 4: 621–630.

42. Caspersen S, Elkjaer M, Riis L, et al. Infliximab for inflammatory bowel diseasein Denmark 1999-2005: clinical outcome and follow-up evaluation ofmalignancy and mortality. Clin Gastroenterol Hepatol 2008; 6: 1212–1217;quiz 1176.

43. de Vries HS, van Oijen MG, de Jong DJ. Serious events with infliximab in pa-tients with inflammatory bowel disease: a 9-year cohort study in the Netherlands.Drug Saf 2008; 31: 1135–1144.

44. Deepak P, Sifuentes H, Sherid M, Stobaugh D, Sadozai Y, Ehrenpreis ED. T-CellNon-Hodgkin’s Lymphomas Reported to the FDA AERS With Tumor NecrosisFactor-Alpha (TNF-alpha) Inhibitors: Results of the REFURBISH Study. Am JGastroenterol 2013; 108(1): 99–105.

45. Lichtenstein GR, Rutgeerts P, Sandborn WJ, et al. A Pooled Analysis of Infec-tions, Malignancy, and Mortality in Infliximab- and Immunomodulator-TreatedAdult Patients With Inflammatory Bowel Disease. Am J Gastroenterol 2012;107(7): 1051–63.

46. Treton X, Bouhnik Y, Mary JY, et al. Azathioprine withdrawal in patients withCrohn’s disease maintained on prolonged remission: a high risk of relapse. ClinGastroenterol Hepatol 2009; 7: 80–85.

47. French H, Mark Dalzell A, Srinivasan R, El-Matary W. Relapse rate followingazathioprine withdrawal in maintaining remission for Crohn’s disease: a meta-analysis. Dig Dis Sci 2011; 56: 1929–1936.

48. Hamzaoglu H, Cooper J, Alsahli M, Falchuk KR, Peppercorn MA, Farrell RJ.Safety of infliximab in Crohn’s disease: a large single-center experience. InflammBowel Dis 2010; 16: 2109–2116.

49. Hemminki K, Li X, Sundquist J, Sundquist K. Cancer risks in Crohn disease pa-tients. Ann Oncol 2009; 20: 574–580.

f. beigel et al.


Documents

Risk of malignancies in patients with inflammatory bowel disease treated with thiopurines or anti-TNF alpha antibodies