- Email: [email protected]
Risk of Lymphoma in Patients With Ulcerative Colitis Treated With Thiopurines: A Nationwide Retrospective Cohort Study
Accepted Manuscript Risk of Lymphoma in Patients with Ulcerative Colitis Treated with Thiopurines—a Nationwide Retrospective Cohort Study Nabeel Khan, MD Ali M. Abbas, MD, MPH Gary R. Lichtenstein, MD Edward V. Loftus, Lydia A. Bazzano, MD, PhD PII: DOI: Reference:
S0016-5085(13)01086-X 10.1053/j.gastro.2013.07.035 YGAST 58576
To appear in: Gastroenterology Accepted Date: 17 July 2013 Please cite this article as: Khan N, Abbas AM, Lichtenstein GR, Loftus Jr. EV, Bazzano LA, Risk of Lymphoma in Patients with Ulcerative Colitis Treated with Thiopurines—a Nationwide Retrospective Cohort Study, Gastroenterology (2013), doi: 10.1053/j.gastro.2013.07.035. This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain. All studies published in Gastroenterology are embargoed until 3PM ET of the day they are published as corrected proofs on-line. Studies cannot be publicized as accepted manuscripts or uncorrected proofs.
ACCEPTED MANUSCRIPT Lymphoma And Thiopurine In Ulcerative Colitis
Risk of Lymphoma in Patients with Ulcerative Colitis Treated with Thiopurines—a Nationwide Retrospective Cohort Study
RI PT
Nabeel Khan, MD1, 2, Ali M. Abbas, MD, MPH1, 2,3, Gary R. Lichtenstein, MD4, Edward V. Loftus Jr., MD5 , Lydia A. Bazzano, MD, PhD1, 6
(1) Section of Gastroenterology, Department of internal medicine at Southeast Louisiana
SC
Veterans Health Care System, New Orleans, LA, USA (2) Section of Gastroenterology and Hepatology, Department of internal medicine at Tulane University Health Sciences Center, New
M AN U
Orleans, LA, USA (3) Department of internal medicine at University of Florida, Gainesville, FL, USA (4) Division of Gastroenterology, Department of Medicine, University of Pennsylvania School of Medicine , Philadelphia, PA, USA (5) Division of Gastroenterology and, Hepatology, Department of Internal Medicine, Mayo Clinic, Rochester, MN, USA (6) Department of
TE D
Epidemiology and of Medicine at Tulane University Health Sciences , Center for Health Research at Ochsner Health System, New Orleans, LA, USA Grant Support:
EP
This study was funded in full by the Department of Veterans Affairs, Veterans Health Administration – Office of Research & Development Health Services. Grant number (VA
AC C
Project No. 425). Abbreviations:
5-ASA: 5-aminosalicylic acid 6-MP: 6-mercaptopurine Anti-TNF: Anti-tumor necrosis factor AZA: Azathioprine 1
ACCEPTED MANUSCRIPT Lymphoma And Thiopurine In Ulcerative Colitis
CDW: Corporate Data Warehouse EBV: Epstein-Barr virus
IBD: Inflammatory Bowel Disease IRR: Incidence rate ratio
IQR: Interquartile range PBM: Pharmacy Benefits Management
SIR: Standardized incidence ratios
M AN U
SEER: Surveillance Epidemiology and End Results
SC
ICD-9: International Classification of Diseases, Ninth Revision
RI PT
HR: Hazard ratios
SLVHCS: Southeast Louisiana Veterans Health Care System TMPT: Thiopurine methyltransferase
TE D
UC: Ulcerative colitis VA: Veterans Affairs healthcare system
Corresponding Author: to whom requests for reprints should be addressed:
EP
Nabeel Khan, M.D.
Assistant Professor of Medicine
AC C
Director, Inflammatory Bowel Disease Tulane University School of Medicine 1430 Tulane Avenue, SL35 New Orleans, LA 70112 Tel: 504-988-7505 Fax: 504-988-2188
2
ACCEPTED MANUSCRIPT Lymphoma And Thiopurine In Ulcerative Colitis
Email: [email protected], [email protected] Disclosure Statement: All authors have no potential conflicts (financial, professional, or personal) that are relevant to
RI PT
this manuscript. The study was conducted and the manuscript was written and reviewed solely by the authors
SC
Writing Assistance: Authors listed above wrote, edit and proof read the current report.
M AN U
Disclaimer:
Study sponsor had no role in study design, analysis and interpretation of the data and in the writing of the report. The contents of this report do not represent the views of the Department of Veterans Affairs or the United States Government. We would like to acknowledge the kind
TE D
support of Joel and Nancy Barnett Foundation. Specific author contributions:
Both Nabeel Khan and Ali Abbas contributed equally to the study. They contributed to study
EP
concept and design; acquisition of data; statistical analysis; medical chart review; interpretation
AC C
of the results; drafting of the manuscript; critical revision of the manuscript for important intellectual content. Gary R. Lichtenstein contributed to interpretation of the results; critical revision of the manuscript. Edward V. Loftus Jr. contributed to interpretation of the results; critical revision of the manuscript. Lydia Bazzano contributed to study design; statistical analysis supervision; interpretation of the results; critical revision of the manuscript.
3
ACCEPTED MANUSCRIPT Lymphoma And Thiopurine In Ulcerative Colitis
ABSTRACT Background & Aims: There is controversy over whether treatment of patients ulcerative
RI PT
colitis (UC) with thiopurines increases their risk of lymphoma. We evaluated the risk of lymphoma (ongoing, residual, and per year of therapy) among thiopurine-treated patients with
SC
UC.
Methods: We obtained nationwide data from the Veterans Affairs (VA) healthcare system, from
M AN U
2001 through 2011. We performed a retrospective cohort study, analyzing data from 36,891 patients with UC from their date of diagnosis at the VA until a diagnosis of lymphoma or October, 1st, 2011 (subjects followed for a median of 6.7 y). Thiopurine exposure was assessed from the VA pharmacy database. Patients who developed lymphoma were identified based on
TE D
ICD-9 codes and confirmed by manual chart review.
Results: Of the patients with UC, 4734 (13%) used thiopurines for a median of 1 y. Lymphomas
EP
developed in 119 patients who had not taken thiopurines, 18 who were taking the drugs, and 5 who had stopped taking them. The incidence rates of lymphoma were 0.60 per 1000 person-y
AC C
among patients who had not taken thiopurines, 2.31 among patients taking the drugs, and 0.28 among patients who had stopped taking them. The incidence rates of lymphoma during the 1st y, 2nd y, 3rd y, 4th y, and >4 y of thiopurine therapy were 0.9, 1.6, 1.6, 5, and 8.9 per 1000 personyears, respectively. The age-, sex- and race-adjusted hazard ratios of developing lymphoma were 4.2 (95% confidence interval, 2.5–6.8; P<.0001) while on thiopurines and 0.5 (95% confidence interval, 0.2–1.3; P=.17) after stopping them, compared to patients who had not taken thiopurines. 4
ACCEPTED MANUSCRIPT Lymphoma And Thiopurine In Ulcerative Colitis
Conclusions: Based on a retrospective, nationwide cohort study, patients with UC have a 4-fold increase in risk of lymphoma while taking thiopurines, compared with patients who have not
therapy reduces risk of lymphoma.
M AN U
INTRODUCTION
SC
Keywords: IBD, inflammatory bowel disease, cancer risk, medication
RI PT
taken these drugs. The risk increased gradually for successive ys of therapy. Stopping thiopurine
Ulcerative colitis (UC) is a chronic inflammatory disorder of unknown etiology. Based on the largest study of prevalence to date, we estimate that more than 700,000 Americans are currently living with UC 1. Corticosteroids are commonly used for induction of remission in UC . However, they are associated with a number of adverse effects, including weight gain, mood
TE D
2
disorders, and osteoporosis, and often lack long-term efficacy 3-6. These two issues have encouraged a number of steroid-sparing therapies to be employed in UC. Such therapies include
EP
thiopurines [azathioprine (AZA), and 6-mercaptopurine (6-MP)], anti-tumor necrosis factor (anti-TNF) agents such as infliximab and adalimumab. Although these medications spare the
AC C
patients the potential adverse effects of long-term corticosteroid use, they have their own spectrum of side effects. Lymphoma is one of the most concerning and potentially lethal adverse outcomes of immunosuppressive therapy and has been linked to thiopurine use 7. Immunosuppressive therapies may contribute to the development of lymphomas in a variety of ways. Decreased immune surveillance of Epstein-Barr virus (EBV) infected B-cells is thought to contribute to lymphoma development 8, and thiopurine use, in particular, has been
5
ACCEPTED MANUSCRIPT Lymphoma And Thiopurine In Ulcerative Colitis
implicated in the development of EBV-positive lymphomas 9-13. Moreover, thiopurine may cause direct damage to DNA by incorporating thiopurine nucleotides during replication, which renders DNA unstable and interferes with further replication cycles and repair mechanisms,
RI PT
resulting in mutations 14-17.
Thiopurine use is widely accepted as a risk factor for lymphoproliferative disorders including lymphoma in post-organ transplant settings. Some reports suggests that they increase
SC
the risk of lymphoproliferative diseases by up to 20- and 200-fold for renal and heart transplant recipients, respectively 7. However, there is no such consensus about the extent of this risk for
M AN U
patients with inflammatory bowel diseases who are treated with thiopurine. Some studies reported no increase in the risk 18, 19, while another study reported an increase of 31 times 20. To date, there are four population-based cohorts published about this issue in the United States and Europe, and they showed conflicting results 21-24. Similarly, the two published meta-analyses
TE D
failed to reach a consistency regarding this issue 25, 26.
The Veteran Affairs (VA) administration has the largest integrated healthcare system in the United States, serving approximately 8.3 million veterans each year 27. We took advantage
EP
of the VA population-based database to (1) estimate the risk of lymphoma among patients who were treated with thiopurine and compare that with UC patients who have not taken this class of
AC C
medication, (2) investigate the potential residual risk of lymphoma after discontinuation of thiopurine, and (3) evaluate the effect of the cumulative duration of thiopurine therapy on the incidence rate of lymphoma. METHODS
Study Population
6
ACCEPTED MANUSCRIPT Lymphoma And Thiopurine In Ulcerative Colitis
Nationwide data were obtained from the VA Pharmacy Benefits Management (PBM) and Corporate Data Warehouse (CDW) databases. Veterans who were seen and followed in the VA health care system from October 1st, 2001 to October 1st, 2011 were identified using the
RI PT
International Classification of Diseases, Ninth Revision (ICD-9) codes for UC (556.xx). The study was approved by the Institutional Review Boards of the Southeast Louisiana Veterans Health Care System (SLVHCS).
SC
Study Design
We conducted a retrospective cohort study in which UC patients were followed-up from
M AN U
the index date (first ICD-9 code for UC entered in their record in the VA system during the observation period) or the first time they filled a prescription for 5-aminosalicylic acid (5-ASA) compounds from the VA pharmacy if that preceded the entry of the first ICD-9 code (start date). The follow-up was concluded at the time the patients developed lymphoma (our outcome of
TE D
interest) or October 1st, 2011 (end of the observation period). To be included in the analysis, the patients had to have been followed in the VA system for at least one month (i.e. a minimum follow-up duration between the start and the end day of one month).
EP
Assessment of Lymphoma Outcome
The medical records of the identified UC patients were queried for the presence of
AC C
lymphoma diagnosis using ICD-9 codes (200-202.28, 202.7-202.98). Medical charts of the identified cases were manually reviewed to confirm the diagnosis of lymphoma according to preset criteria that considers biopsy results or a note from a hematologist as adequate proves for correct diagnosis. The date, type, and location of lymphoma were abstracted from the charts. We excluded prevalent cases of lymphoma (i.e. patients diagnosed with lymphoma prior to October 1st, 2001 or UC diagnosis). In order to identify the current and the residual effects that
7
ACCEPTED MANUSCRIPT Lymphoma And Thiopurine In Ulcerative Colitis
thiopurine therapy might have on the risk of lymphoma, we classified the identified cases into those occurring during or up to six months after thiopurine treatment (co-incidental) as compared to lymphomas diagnosed after discontinuation of thiopurine therapy for six months or more
RI PT
(non-coincidental). Assessment of Exposure and Other Predictors
Automated data abstraction was used to identify the periods of exposure to thiopurine
SC
during the follow-up period from the VA pharmacy database. Person-years of exposure to thiopurine were calculated in a time-dependent manner, accounting for the initiation and
M AN U
discontinuation of therapy, with a study subject potentially contributing person-years of followup to any or all the three exposure categories (1) unexposed: follow-up periods from all the patients who never filled thiopurine and the period prior to the first filling of thiopurine users, (2) during exposure: periods of exposure to thiopurine started at the filling day and ended when
TE D
the medication day supply was exhausted (3) after stopping: follow-up periods after the last fill of thiopurine was exhausted and the gaps between exposure periods if present 21, 23. Other predictors such as age at the start of follow-up (<40, 40-64, >65 years old), sex and race were
EP
also collected.
Infliximab exposure during the follow-up period was identified as present versus absent
AC C
manner due to its low prevalence in the VA system (1%). We calculated the incidence rates of lymphoma among infliximab users as a monotherapy or in combination with thiopurines and we performed a sensitivity analysis after excluding infliximab users to identify its effect on the observed lymphomas. Statistical Analysis
8
ACCEPTED MANUSCRIPT Lymphoma And Thiopurine In Ulcerative Colitis
To study the effect of ongoing and past thiopurine therapy on the risk of lymphoma, all UC patients were included, and the incidence rate of lymphoma (per 1000 person-years) was calculated for unexposed patients, those currently being treated with thiopurine, and for those
RI PT
who stopped thiopurine therapy. Thiopurine exposure was included in a multivariate Cox
regression model as a time-dependent variable. The age-, sex- and race-adjusted hazard ratios (HR) of lymphoma while on thiopurine treatment and after stopping it as compared to unexposed
SC
patients were calculated.
In order to improve the generalizability of our findings, we age-adjusted the
M AN U
aforementioned incidence rates using weights that represent the age group proportions of the 2000 United States standard population. Standardized incidence ratios (SIR) of non-Hodgkin lymphoma were calculated by comparing these rates with the age adjusted rate of non-Hodgkin lymphoma as reported by Surveillance Epidemiology and End Results (SEER) database28.
TE D
We investigated the effect of cumulative duration of thiopurine therapy on the incidence of lymphoma by performing a life table analysis with the follow-up time classified into five intervals at the end of each year (1st year, 2nd year, 3rd year, 4th year, and beyond 4 years) of exposure or follow-up
EP
(Table 4). For each interval, we identified the total population at the start of the interval, person year contribution and number of lymphoma cases that were diagnosed during interval. The latter two
AC C
parameters were used to calculate the per-year incidence rate of lymphoma for those using thiopurines and those who were not. We then compared the rate of lymphoma occurrence between the exposed and unexposed participants within each interval using a multivariate Cox regression analysis adjusted for age, sex, and race. The “number needed to harm” per each year of treatment with thiopurines was calculated from the incidence rates of lymphoma among exposed and non-exposed. To identify whether different exposure durations had affected the observed residual risk of lymphoma after discontinuation, we examined the incidence rate of lymphoma after discontinuation of
9
ACCEPTED MANUSCRIPT Lymphoma And Thiopurine In Ulcerative Colitis thiopurines in the 1st, 2nd, 3rd, 4th, and after 4 years of use, by dividing the number of events by the person-years of follow-up after stopping the thiopurine. An evaluation of the effect of different dosage utilized was performed by comparing those who
RI PT
used thiopurines at below median doses against the rest of thiopurine users. No statistically significant difference was observed based on different dosing groups. However, our ability to examine this issue was limited due to the absence of weight information needed for calculating dosing parameters, the
SC
differences in the formulations and dosages of thiopurines used at the VA hospitals, dose fluctuation over treatment period, and the limited power to perform stratified analysis.
M AN U
All data extraction and analysis procedures were conducted using SAS version 9.2 for Windows (SAS Institute Inc., Cary, NC, USA). Sensitivity and Reliability of Case Analyses
Although a previous study has established the sensitivity and the specificity of using ICD-9 codes for UC diagnosis in the VA system to be 84% and 99%, respectively 29, we
reliability of our results.
TE D
performed a comprehensive sensitivity analysis to establish the validity of our methods and
In the primary analysis we included all the identified patients with a minimum of 1 ICD-9
EP
code for UC in their records. In sensitivity analyses, we included patients who had ≥2 UC codes,
AC C
then ≥3, then ≥4. We further added the number of times a 5-ASA prescription (oral or rectal mesalamine, sulfasalazine, balsalazide, and olsalazine) was filled in the VA pharmacy database during the follow-up period as a criterion (once, twice, three or more times). Based on the combination of these 2 criteria, we repeated our analysis to evaluate the robustness of our results. We then randomly selected 100 patients from those who had ≥4 ICD-9 codes for UC recorded in their medical record and ≥3 fills of 5-ASA from the pharmacy database and manually reviewed
10
ACCEPTED MANUSCRIPT Lymphoma And Thiopurine In Ulcerative Colitis
their medical records to calculate the positive predictive value of this method of identifying UC patients. To identify whether short-term use of thiopurines affected our results, we performed a sensitivity
RI PT
analysis by rerunning the multivariate analysis after excluding patients with less than six months, less than one year, and less than one and a half year of exposure. The observed hazard ratios of lymphoma while taking thiopurines when requiring a minimum of six months, one year, and one and a half year of
SC
exposure were consistent with the one obtained by including the entire cohort (Appendix 1, Table 1). To investigate whether the observed reduction in the risk of lymphoma upon discontinuation was
M AN U
associated with receiving colectomy, we performed a sensitivity analysis after excluding patients who had colectomy over the follow-up period. The observed reduction in the risk of lymphoma remained consistent (Appendix 1, Table 2).
RESULTS
We identified 37,191 UC patients followed in the VA system during the observational
TE D
period. A total of 325 patients had diagnostic codes for lymphoma in their records. Manual chart review confirmed 253 cases of lymphoma (among them, 111 prevalent cases of lymphoma were excluded). We included only incident lymphoma cases (happened during the follow-up
EP
time) confirmed by manual chart review in our analysis. Among the remainder of the patients
AC C
with a diagnostic code for lymphoma, we found other hematological conditions in 18 patients, suspected lymphoma in 11 patients, and no evidence of lymphoma in 43 patients. All these groups were reclassified as non-lymphoma patients. Patients with less than one month of followup in the VA system were also excluded (n=189). Of the 142 confirmed lymphoma cases, 119 occurred among those who never used a thiopurine, while 23 cases occurred among thiopurine users: 18 of these occurred during thiopurine use or within 6 months of discontinuation (co-
11
ACCEPTED MANUSCRIPT Lymphoma And Thiopurine In Ulcerative Colitis
incidental) and 5 occurred more than one year after discontinuation of thiopurine use (noncoincidental) (Table 1). We included 36,891 patients with 224,773 total person-years of follow-up in our
RI PT
analyses. Median follow-up was 6.7 years [interquartile range (IQR), 3.2-9.2]. Patients were predominantly white (76%) males (93%) with a median age at the start of the follow-up period of 60 years (IQR, 50-69). Thiopurines were used by 4,734 patients (13%) and 5-ASA compounds
SC
were used by 22,664 (61%) throughout the follow-up period (Table 2).
Median duration of follow-up among the thiopurine-unexposed was 5.72 years (IQR,
M AN U
2.25-8.87) with total person-years for this period available for analysis of 199,046. The median duration of thiopurine exposure was 0.97 years (IQR, 0.28-2.5) with total person-years for this period available for analysis of 7,778. Median duration of follow-up after stopping thiopurine was 3.52 years (IQR, 1.63-5.77) with total person-years for this period available for analysis of
TE D
17,950.
The overall incidence rate of lymphoma for the entire cohort was 0.63 cases per 1000 person-years. The incidence rate of lymphoma was 2.31 and 0.28 per 1000 person-years while
EP
on thiopurine and after stopping it, respectively, compared to 0.60 per 1000 person-years for unexposed patients. The age- and thiopurine-stratified incidence rate of lymphoma is shown in
AC C
(Figure 1). Using Cox regression analysis, the age-, sex- and race-adjusted HRs of developing lymphoma while on thiopurine and after stopping it were 4.2 (95% CI, 2.5-6.8, p<0.0001) and 0.5 (95% CI, 0.2-1.3, p=0.17), respectively, as compared to unexposed patients. Other predictors of lymphoma were age >65 years (HR, 2.6; 95% CI, 1.2-5.7; p=0.01) and male gender (HR, 2.5; 95% CI, 0.8-7.9; p=0.12), (Table 3).
12
ACCEPTED MANUSCRIPT Lymphoma And Thiopurine In Ulcerative Colitis
The age-adjusted incidence rate of lymphoma was 1.46 and 0.19 per 1000 person-years while on thiopurine and after stopping it, respectively, compared to 0.32 per 1000 person-years for unexposed patients. The SIR of non-Hodgkin lymphoma while on thiopurine was 7.4 (95%
RI PT
CI, 4.6-11.9; p<0.0001) as compared to the age standardized SEER population (Appendix 2). The results of the thiopurine cumulative effect analysis are presented in Table 4. The incidence rate of lymphoma during the 1st, 2nd, 3rd, 4th and after 4 years of thiopurine use was 0.9, 1.6, 1.6, 5, and
SC
8.9 cases per 1000 person-years, respectively. Using the incidence rate among unexposed patients during the successive follow-up intervals as a reference, the age- sex- race- adjusted HR of lymphoma was 1.2,
M AN U
3.8, 2.9, 16 (p<0.0001), and 14.4 (p<0.0001) during the 1st, 2nd, 3rd, 4th and >4 years of thiopurine use, respectively. The risk of lymphoma was statistically significantly elevated in the fourth and subsequent years of exposure to thiopurine. The number needed to harm was 17,344 patients for the first year compared to 122 patients for those with more than four years of therapy.
The incidence rates of lymphoma after thiopurine discontinuation in the 1st, 2nd, 3rd, 4th
TE D
and after 4 years of use was 0, 0.66, 0.53, 0.71 and 0.46 per 1000 person-years respectively. These rates were within the observed range among the unexposed group (overall, 0.62; range 0.35-0.87 per 1000 person-years). The number of patients, number of lymphomas following
AC C
Infliximab use
EP
thiopurine discontinuation, and the total person-years analyzed are presented in Table 5.
We had a very low rate of infliximab use in this cohort. Among the included UC patients, only 251 (1%) were treated with infliximab during the follow-up period. Of these, 177 (71% of infliximab users) patients used infliximab concomitantly with thiopurines. Among the latter group only one lymphoma was observed. Among infliximab monotherapy users, none developed lymphoma. The incidence rate among infliximab users is presented in (Appendix 3, Table 1). Although the incidence rate among patients exposed to thiopurine and infliximab was
13
ACCEPTED MANUSCRIPT Lymphoma And Thiopurine In Ulcerative Colitis
higher compared to those exposed only to thiopurine (3.84 compared to 2.26 per 1000 personyears respectively), our study is underpowered to detect the difference between these rates because of the limitation of low sample size of infliximab users. To identify the extent of
RI PT
infliximab effect on the observed association between thiopurine use and lymphoma risk, we performed a sensitivity analysis by excluding all infliximab users (Appendix 3, Table 2). The observed hazard ratio of lymphoma did not change after excluding infliximab users, indicating
SC
that the observed association is associated predominantly with thiopurine use. Results of Sensitivity Analyses
M AN U
Using the different combinations of the number of ICD-9 codes indicating UC diagnoses and the number of 5-ASA compounds filled in the nation-wide pharmacy database as inclusion criteria, the HR of lymphoma while on thiopurine showed consistency ranging from 3.9-5.6 with overlapping confidence intervals. After reviewing the medical charts of a randomly-selected 100
TE D
patients with a minimum of 4 ICD-9 codes indicating a diagnosis of UC and at least 3 fills of 5ASA compounds, we found that 92 patients had a confirmed diagnosis of UC, 4 had indeterminate colitis, 2 had no medical charts in the VA CAPRI system, and 2 had Crohn’s
EP
disease. Thus, the positive predictive value of this definition in identifying UC patients was
AC C
92%. Full results of sensitivity analyses are presented in (Appendix 1, Table 3). DISCUSSION
In this nationwide cohort of UC patients from the VA healthcare system, treatment with a thiopurine was associated with an approximate four-fold and a seven-fold increase in the risk of lymphoma relative to UC patients who did not use a thiopurine and to SEER general population respectively. This finding is very consistent with the previous population-based prospective cohort study reported by Beaugerie et al. 2009 21. Additionally, we have shown that the
14
ACCEPTED MANUSCRIPT Lymphoma And Thiopurine In Ulcerative Colitis
magnitude of this risk is significantly associated with the duration of treatment, with therapy duration of four years as a threshold for achieving statistical significance. Furthermore, we have shown that stopping the thiopurine reduced the risk of lymphoma during a median follow-up
RI PT
time of 3.5 years.
Several autoimmune and chronic inflammatory disorders are associated with increased risk of lymphoma 30-34. It is reasonable to think that Inflammatory Bowel Disease (IBD) patients
SC
would have a baseline increased lymphoma risk due to the presence of a chronic inflammatory process. However, many well-designed, population-based studies have failed to show any
M AN U
increased risk 22, 35-37. Unfortunately, there is less consistency in the literature regarding the effect of thiopurine on the risk of lymphoma in the setting of IBD.
Whether the increase in risk of lymphoma associated with thiopurine is constant (risk starts at the first intake and stops upon cessation) or cumulative (affected by the duration and/or
TE D
the dose of thiopurine treatment) is still controversial. Although many IBD and post-transplant studies have concluded that the risk is constant 7, 21, 22, several other studies have shown a cumulative dose/duration effect among rheumatoid arthritis patients 38, 39. A recent study by
EP
Nguyen et al. in 2009 showed a significant increase in the frequency of mutations with both cumulative dose and duration of thiopurine treatment for IBD patients after controlling for
AC C
disease duration 40. Our study is the first to show the cumulative effect of thiopurine treatment on the risk of lymphoma among UC patients. Although the risk of mutation described by Nguyen et al. lacked a treatment duration threshold (a continuous increase along with therapy duration), the risk in our cohort showed an interesting pattern. Lymphoma risk increased to 1.23.8 times for the first three years of therapy, but mostly it lacked the statistical significance.
15
ACCEPTED MANUSCRIPT Lymphoma And Thiopurine In Ulcerative Colitis
However, after three years of thiopurine therapy, the risk increased significantly to 14-16 times as compared to those not on thiopurines (Table 4, Figure 2). Previous studies reported a significant increase in the risk of lymphoma among IBD
RI PT
patients on thiopurines, with an effect size of 2-31 times 9, 20, 21, 41, 42, but none investigated the effect of treatment duration on that risk. The inconsistency in the reported magnitude of the observed risk and the lack of data about the effect of long-term therapy have made weighing the
SC
benefits versus the risks of initiating or continuing thiopurine therapy a challenging task for both patients and clinicians. Lewis et al. in 2000 assessed the gain in quality-adjusted life expectancy
M AN U
resulting from AZA treatment in Crohn’s disease against the potential for serious side effects including the risk of lymphoma 43, and concluded that alternative therapy is the preferred treatment strategy if the increased risk of lymphoma exceeded 9.8 folds compared to non-users. Using the results from Lewis et al. analysis (in the absence of comparable study that dealt with
TE D
UC patients), our study shows that the benefits of using thiopurine outweigh the risk of lymphoma in the first three years of therapy. For the fourth year of therapy, the magnitude of the risk started to approach (IRR=8.8) and after the fourth year it exceeded it (IRR=15.5) the
EP
proposed threshold of 9.8 by Lewis et al 43. Thus considering a withdrawal of therapy trial within three years of therapy onset is a reasonable option.
AC C
It has been reported that men had a higher risk of lymphoma than women 44, 45. In our cohort, the overall incidence rate was 0.66 compared to 0.21 cases per 1000 person-years for men and women, respectively. After adjusting for age, race and thiopurine use, men had a nonsignificant trend towards developing more lymphoma compared to women (HR= 2.5, 0.8-7.8, p=0.1). Failure to achieve statistical significance is attributed to the very small number of women in the military veteran population (2467 patients, 7%).
16
ACCEPTED MANUSCRIPT Lymphoma And Thiopurine In Ulcerative Colitis
In the absence of comparable treatment with a lower risk of malignancy, thiopurines remain a mainstay in treating steroid-dependent IBD patients. However, there are several measures that can be undertaken that may reduce the potential risks of therapy. First of all,
RI PT
patients for thiopurine therapy should be selected with caution, especially among older males. Secondly, assessment with thiopurine methyltransferase (TMPT) genotype or enzyme activity with subsequent weight-based dose adjustment and monitoring will result in less prolonged
SC
leukopenia. Finally, one could consider a trial of withdrawal of thiopurine therapy after three years, especially in cases of long-term remission, as this group has been demonstrated to have a
M AN U
lower recurrence rate 46, 47.
Our study is the largest to examine the risk of lymphoma associated with thiopurine therapy in UC patients to date. Individual level patient data was available and referral bias is unlikely to exist in our analysis due to the fact that the database was derived from a nationwide
TE D
source in which all the states are represented. The availability of pharmacy data allowed us to objectively assess the extent of exposure to thiopurine over extended periods of time and to identify the exact date of therapy withdrawal rather than depending on the self-reported or health
EP
provider-reported information about the medication exposure 21. Furthermore, we performed a stratified analysis to identify the residual risk based after therapy withdrawal according to the
AC C
cumulative duration of prior exposure. Additionally, our results were consistent with different study designs 48, and with extensive sensitivity analyses. Finally, the VA database has been previously validated and used frequently for research regarding inflammatory bowel disease 29, 49-51
.
The investigators acknowledge some limitations of the present work. Firstly, although this study is based on the largest integrated nationwide healthcare system in the United States 27,
17
ACCEPTED MANUSCRIPT Lymphoma And Thiopurine In Ulcerative Colitis
it is not a population-based cohort in the full sense due to the different demographics of the VA compared to the general United States population, specifically the age, race, and sex distribution. The VA population is predominantly comprised of middle- to old-aged Caucasian-males, which
RI PT
can limit the external validity of the study to patients within these demographics.
Due to the fact that we are investigating the rate of lymphoma among a high risk
population (old, white, males), our observed incidence rate might be higher than what might be
SC
expected from the general population databases. Nonetheless, we believe that our multivariate hazard ratio of thiopurine exposure is not overestimated, as we are comparing within the same
M AN U
population (VA thiopurine-exposed patients with UC compared to VA thiopurine-unexposed patients with UC). We tried to limit this effect by adjusting our incidence rates according to the age distribution of the US census in 2000, the same reference population used in SEER database. This increased the comparability of these rates with those derived from the general population.
TE D
However, male gender and white race are other risk factors for lymphoma that we were not able to control for simultaneously with the age to calculate the standardized rates28. Thus our standardized incidence rates and the ratios driven from them might be still overestimated.
EP
Secondly, the study is susceptible to limitations inherent to the retrospective design. The retrospective design is the most suitable way to address this question due to the low incidence
AC C
rate of lymphoma and the ability to obtain a long follow-up period. Thirdly, there is a possibility that thiopurine exposure as assessed from the pharmacy data does not completely reflect the extent of exposure due to the presence of dual health insurance eligibility and additional sources of medication other than the VA pharmacy. The investigators believe that this will have a very limited effect on the results of our analysis, as previous reports showed a very high rate of VA pharmacy use by veterans irrespective of their dual eligibility status due its convenience and
18
ACCEPTED MANUSCRIPT Lymphoma And Thiopurine In Ulcerative Colitis
financial plausibility 52-54. Lastly, EBV status was unknown for almost all of the patients, as it is rarely recorded as a diagnostic code and the chart review did not reveal further information related to EBV status.
RI PT
In summary, in this nationwide cohort of UC patients followed in the VA system, patients were four times more likely to develop lymphoma while receiving a thiopurine compared to
Stopping thiopurine therapy reduced the risk of lymphoma. Figures Legends:
SC
unexposed patients. The risk of lymphoma increased gradually for successive years of therapy.
M AN U
Figure 1: Incidence rate of lymphoma stratified by age and thiopurine use Figure 2: Age, Sex, Race adjusted HR of lymphoma while on thiopurine compared to unexposed
AC C
EP
TE D
during each follow-up interval (Cox regression analysis)
19
ACCEPTED MANUSCRIPT Lymphoma And Thiopurine In Ulcerative Colitis
REFERENCES 1.
Kappelman MD, Rifas-Shiman SL, Kleinman K, et al. The prevalence and geographic
RI PT
distribution of Crohn's disease and ulcerative colitis in the United States. Clin Gastroenterol Hepatol 2007;5:1424-9. 2.
Ford AC, Bernstein CN, Khan KJ, et al. Glucocorticosteroid therapy in inflammatory
SC
bowel disease: systematic review and meta-analysis. Am J Gastroenterol 2011;106:590-9; quiz 600.
Gurwitz JH, Bohn RL, Glynn RJ, et al. Glucocorticoids and the risk for initiation of
M AN U
3.
hypoglycemic therapy. Arch Intern Med 1994;154:97-101. 4.
Stuck AE, Minder CE, Frey FJ. Risk of infectious complications in patients taking glucocorticosteroids. Rev Infect Dis 1989;11:954-63.
Lukert BP, Raisz LG. Glucocorticoid-induced osteoporosis. Rheum Dis Clin North Am 1994;20:629-50.
6.
Mankin HJ. Nontraumatic necrosis of bone (osteonecrosis). N Engl J Med
EP
1992;326:1473-9. 7.
TE D
5.
Opelz G, Döhler B. Lymphomas after solid organ transplantation: a collaborative
8.
AC C
transplant study report. Am J Transplant 2004;4:222-30. Pietersma F, Piriou E, van Baarle D. Immune surveillance of EBV-infected B cells and the development of non-Hodgkin lymphomas in immunocompromised patients. Leuk Lymphoma 2008;49:1028-41.
20
ACCEPTED MANUSCRIPT Lymphoma And Thiopurine In Ulcerative Colitis
9.
Dayharsh GA, Loftus EV, Sandborn WJ, et al. Epstein-Barr virus-positive lymphoma in patients with inflammatory bowel disease treated with azathioprine or 6-mercaptopurine. Gastroenterology 2002;122:72-7. Larvol L, Soule JC, Le Tourneau A. Reversible lymphoma in the setting of azathioprine therapy for Crohn's disease. N Engl J Med 1994;331:883-4.
11.
RI PT
10.
Munkholm P, Langholz E, Davidsen M, et al. Intestinal cancer risk and mortality in
12.
SC
patients with Crohn's disease. Gastroenterology 1993;105:1716-23.
Wakefield AJ, Fox JD, Sawyerr AM, et al. Detection of herpesvirus DNA in the large
M AN U
intestine of patients with ulcerative colitis and Crohn's disease using the nested polymerase chain reaction. J Med Virol 1992;38:183-90. 13.
Wong NA, Harrison DJ. Colorectal neoplasia in ulcerative colitis-recent advances. Histopathology 2001;39:221-34.
Krynetskaia NF, Krynetski EY, Evans WE. Human RNase H-mediated RNA cleavage
TE D
14.
from DNA-RNA duplexes is inhibited by 6-deoxythioguanosine incorporation into DNA. Mol Pharmacol 1999;56:841-8.
Krynetskaia NF, Cai X, Nitiss JL, et al. Thioguanine substitution alters DNA cleavage
EP
15.
mediated by topoisomerase II. FASEB J 2000;14:2339-44. Ling YH, Chan JY, Beattie KL, et al. Consequences of 6-thioguanine incorporation into
AC C
16.
DNA on polymerase, ligase, and endonuclease reactions. Mol Pharmacol 1992;42:802-7. 17.
Somerville L, Krynetski EY, Krynetskaia NF, et al. Structure and dynamics of thioguanine-modified duplex DNA. J Biol Chem 2003;278:1005-11.
18.
Connell WR, Kamm MA, Dickson M, et al. Long-term neoplasia risk after azathioprine treatment in inflammatory bowel disease. Lancet 1994;343:1249-52.
21
ACCEPTED MANUSCRIPT Lymphoma And Thiopurine In Ulcerative Colitis
19.
Warman JI, Korelitz BI, Fleisher MR, et al. Cumulative experience with short- and longterm toxicity to 6-mercaptopurine in the treatment of Crohn's disease and ulcerative colitis. J Clin Gastroenterol 2003;37:220-5. Farrell RJ, Ang Y, Kileen P, et al. Increased incidence of non-Hodgkin's lymphoma in
RI PT
20.
inflammatory bowel disease patients on immunosuppressive therapy but overall risk is low. Gut 2000;47:514-9.
Beaugerie L, Brousse N, Bouvier AM, et al. Lymphoproliferative disorders in patients
SC
21.
study. Lancet 2009;374:1617-25. 22.
M AN U
receiving thiopurines for inflammatory bowel disease: a prospective observational cohort
Lewis JD, Bilker WB, Brensinger C, et al. Inflammatory bowel disease is not associated with an increased risk of lymphoma. Gastroenterology 2001;121:1080-7.
23.
Herrinton LJ, Liu L, Weng X, et al. Role of thiopurine and anti-TNF therapy in
24.
TE D
lymphoma in inflammatory bowel disease. Am J Gastroenterol 2011;106:2146-53. Lakatos PL, Lovasz BD, David G, et al. The risk of lymphoma and immunomodulators in patients with inflammatory bowel diseases: Results from a population-based cohort in
25.
EP
Eastern Europe. J Crohns Colitis 2012.
Kandiel A, Fraser AG, Korelitz BI, et al. Increased risk of lymphoma among
AC C
inflammatory bowel disease patients treated with azathioprine and 6-mercaptopurine. Gut 2005;54:1121-5. 26.
Masunaga Y, Ohno K, Ogawa R, et al. Meta-analysis of risk of malignancy with immunosuppressive drugs in inflammatory bowel disease. Ann Pharmacother 2007;41:21-8.
22
ACCEPTED MANUSCRIPT Lymphoma And Thiopurine In Ulcerative Colitis
27.
The National Center for Veterans Analysis and Statistics (NCVAS), Trends in the Utilization of VA Programs and Services. Washington, DC.: Available from: http://www.va.gov/vetdata/docs/quickfacts/Utilization-slideshow.pdf, 2012. National Cancer Institute - Surveillance Epidemiology and End Results (SEER) Stat Fact
RI PT
28.
Sheets: Non-Hodgkin Lymphoma. 2012, Accessed 5-21-2013, Available in: http://seer.cancer.gov/statfacts/html/nhl.html, 2012.
Thirumurthi S, Chowdhury R, Richardson P, et al. Validation of ICD-9-CM diagnostic
SC
29.
codes for inflammatory bowel disease among veterans. Dig Dis Sci 2010;55:2592-8. Smedby KE, Askling J, Mariette X, et al. Autoimmune and inflammatory disorders and
M AN U
30.
risk of malignant lymphomas--an update. J Intern Med 2008;264:514-27. 31.
Mellemkjaer L, Pfeiffer RM, Engels EA, et al. Autoimmune disease in individuals and close family members and susceptibility to non-Hodgkin's lymphoma. Arthritis Rheum
32.
TE D
2008;58:657-66.
Zintzaras E, Voulgarelis M, Moutsopoulos HM. The risk of lymphoma development in autoimmune diseases: a meta-analysis. Arch Intern Med 2005;165:2337-44. Baecklund E, Ekbom A, Sparén P, et al. Disease activity and risk of lymphoma in
EP
33.
patients with rheumatoid arthritis: nested case-control study. BMJ 1998;317:180-1. Baecklund E, Iliadou A, Askling J, et al. Association of chronic inflammation, not its
AC C
34.
treatment, with increased lymphoma risk in rheumatoid arthritis. Arthritis Rheum 2006;54:692-701. 35.
Askling J, Brandt L, Lapidus A, et al. Risk of haematopoietic cancer in patients with inflammatory bowel disease. Gut 2005;54:617-22.
23
ACCEPTED MANUSCRIPT Lymphoma And Thiopurine In Ulcerative Colitis
36.
Loftus EV, Tremaine WJ, Habermann TM, et al. Risk of lymphoma in inflammatory bowel disease. Am J Gastroenterol 2000;95:2308-12.
37.
Karlén P, Löfberg R, Broström O, et al. Increased risk of cancer in ulcerative colitis: a
38.
RI PT
population-based cohort study. Am J Gastroenterol 1999;94:1047-52.
Asten P, Barrett J, Symmons D. Risk of developing certain malignancies is related to duration of immunosuppressive drug exposure in patients with rheumatic diseases. J
39.
SC
Rheumatol 1999;26:1705-14.
Silman AJ, Petrie J, Hazleman B, et al. Lymphoproliferative cancer and other malignancy
Ann Rheum Dis 1988;47:988-92. 40.
M AN U
in patients with rheumatoid arthritis treated with azathioprine: a 20 year follow up study.
Nguyen T, Vacek PM, O'Neill P, et al. Mutagenicity and potential carcinogenicity of thiopurine treatment in patients with inflammatory bowel disease. Cancer Res
41.
TE D
2009;69:7004-12.
Kinlen LJ. Incidence of cancer in rheumatoid arthritis and other disorders after immunosuppressive treatment. Am J Med 1985;78:44-9. Armstrong RG, West J, Card TR. Risk of cancer in inflammatory bowel disease treated
EP
42.
with azathioprine: a UK population-based case-control study. Am J Gastroenterol
43.
AC C
2010;105:1604-9.
Lewis JD, Schwartz JS, Lichtenstein GR. Azathioprine for maintenance of remission in Crohn's disease: benefits outweigh the risk of lymphoma. Gastroenterology 2000;118:1018-24.
44.
Results NCI-SEaE. SEER Stat Fact Sheets: Non-Hodgkin Lymphoma, 2012.
24
ACCEPTED MANUSCRIPT Lymphoma And Thiopurine In Ulcerative Colitis
45.
Bosetti C, Levi F, Ferlay J, et al. Incidence and mortality from non-Hodgkin lymphoma in Europe: the end of an epidemic? Int J Cancer 2008;123:1917-23.
46.
Smith MA, Irving PM, Marinaki AM, et al. Review article: malignancy on thiopurine
RI PT
treatment with special reference to inflammatory bowel disease. Aliment Pharmacol Ther 2010;32:119-30. 47.
Bitton A, Peppercorn MA, Antonioli DA, et al. Clinical, biological, and histologic
SC
parameters as predictors of relapse in ulcerative colitis. Gastroenterology 2001;120:1320.
Abbas A, Koleva Y, Khan N. Lymphoproliferative Disorders in Patients Receiving
M AN U
48.
Thiopurines for Ulcerative Colitis: A Nationwide 10-Year Retrospective Cohort from the Veterans Affairs Healthcare System. Am J Gastroenterol 2012;107:S693. 49.
Khan N, Abbas AM, Bazzano LA, et al. Long-term oral mesalazine adherence and the
TE D
risk of disease flare in ulcerative colitis: nationwide 10-year retrospective cohort from the veterans affairs healthcare system. Aliment Pharmacol Ther 2012;36:755-64. 50.
Hou JK, Kramer JR, Richardson P, et al. Risk of colorectal cancer among Caucasian and
51.
EP
African American veterans with ulcerative colitis. Inflamm Bowel Dis 2012;18:1011-7. Sonnenberg A, Richardson PA, Abraham NS. Hospitalizations for inflammatory bowel
52.
AC C
disease among US military veterans 1975-2006. Dig Dis Sci 2009;54:1740-5. Veterans Health Care: Use of VA Services by Medicare-Eligible Veterans, U.S. Government Accountability Office Report, 1994. 53.
Morgan RO, Petersen LA, Hasche JC, et al. VHA pharmacy use in veterans with Medicare drug coverage. Am J Manag Care 2009;15:e1-8.
25
ACCEPTED MANUSCRIPT Lymphoma And Thiopurine In Ulcerative Colitis
Wolinsky FD, Miller TR, An H, et al. Dual use of Medicare and the Veterans Health
EP
TE D
M AN U
SC
RI PT
Administration: are there adverse health outcomes? BMC Health Serv Res 2006;6:131.
AC C
54.
26
ACCEPTED MANUSCRIPT
Appendix 1, results of sensitivity analyses
entire cohort excluding < 6 months excluding < 1 year excluding < 1.5 years
36891 35172 34462 33986
4734 3015 2305 1829
7778 7376 6850 6260
18 17 15 15
2.3 2.3 2.2 2.4
HR while on treatment p
UCI LCI
4.2 4.0 3.8 4.1
2.5 2.4 2.2 2.4
RI PT
Table 1: Sensitivity analysis, excluding short term thiopurine users P-Y No. of No. of patients exposed exposure included patients analyzed events IR
<0.0001 <0.0001 <0.0001 <0.0001
M AN U
SC
Notes: P-Y: person-years, IR: Incidence rate per 1000 person-years, HR: Hazard Ratio, UCI and LCI: Upper and Lower limits of the 95% confidence interval respectively
EP
TE D
Table 2: Sensitivity analysis excluding patients who required colectomy No. No. total events IR Thiopurine P-Y 36,297 199,046 119 0.6 All Cohort Unexposed 4,734 7,778 18 2.31 During Exposure 4,662 17,950 5 0.28 After Stopping 33,308 177,540 103 0.58 Unexposed Patients who did not require During Exposure 4,158 7,145 17 2.38 colectomy 4158 14,956 4 0.27 After Stopping
HR p 1 (reference) 4.2 <0.0001 0.5 0.17 1 (reference) 4.4 <0.0001 0.5 0.22
95%CI 2.5-6.8 0.2-1.3 2.6-7.3 0.19-1.5
AC C
Notes: P-Y: person years contributions, events: lymphoma cases, IR: Incidence rate per 1000 person-years, HR: Hazard Ratio, 95%CI: the 95% confidence interval.
6.8 6.8 6.5 7.1
ACCEPTED MANUSCRIPT
One time One time One time One time
22,664 19,285 17,257 15,775
84 70 65 57
5.4 5.2 5.6 4.8
3.2 2.9 3.1 2.5
9.2 9.2 10.0 9.1
<0.0001 <0.0001 <0.0001 <0.0001
One time Two times Three times Four times
Two times Two times Two times Two times
21,117 18,262 16,505 15,187
78 65 61 54
4.9 4.6 4.9 4.0
2.8 2.5 2.6 2.0
8.6 8.5 9.2 8.1
<0.0001 <0.0001 <0.0001 <0.0001
One time Two times Three times Four times
Three times Three times Three times Three times
19,849 17,336 15,791 14,608
78 65 61 54
4.8 4.5 4.8 3.9
2.7 2.4 2.6 2.0
8.4 8.3 9.0 7.9
<0.0001 <0.0001 <0.0001 <0.0001
M AN U
SC
One time Two times Three times Four times
RI PT
Table 3: Adjusted hazard ratios of lymphoma while using thiopurine based on different ulcerative colitis definition criteria, sensitivity analysis results. Included Events UC ICD9 no. 5-ASA fills no. patients no. no. HR UCI LCI p 36,891 142 4.2 2.5 6.8 <0.0001 One time 27,068 105 4.2 2.4 7.3 <0.0001 Two times 23,035 93 4.6 2.6 8.0 <0.0001 Three times 20,401 80 4.0 2.1 7.4 <0.0001 Four times
AC C
EP
TE D
Notes: HR: Hazard Ratio, UCI and LCI: Upper and Lower limits of the 95% confidence interval respectively
ACCEPTED MANUSCRIPT
Appendix 2: Age standardization results Table: Incidence rate of NHL among VA population compared to SEER population SEER VA Unexposed VA During Exposure VA After Stopping
SIR reference group 1.66 7.47 0.98
95% CI 0.95-2.90 4.67-11.98 0.52-1.84
RI PT
*IR 0.20 0.33 1.47 0.19
AC C
EP
TE D
M AN U
SC
IR: Incidence rate, SIR: Standardized incidence ratio, 95%CI: 95% confidence interval limits. *Rates are per 1000 person-year and are age-adjusted to the 2000 US Standard Population
ACCEPTED MANUSCRIPT
Appendix 3: Infliximab use analysis results
IRR 95%CI 1 3.6 2.2- 6.0 NA 6.2 0.8- 44.2
RI PT
Table 1: Incidence rate of lymphoma by thiopurine and infliximab exposure No. of P-Y P-Y patients Follow-up Exposure events IR 32083 190944 . 119 0.62 Unexposed 4557 . 7517 17 2.26 Thiopurine exposed 74 482 . 0 0.00 Infliximab exposed . 260 1 3.84 Thiopurine and Infliximab 177
Notes: P-Y: person-years of follow-up, IR: Incidence Rate, Rate is per 1000 person-years, IRR: Incidence Rate Ratio, 95%CI: the 95% confidence interval limit of the IRR.
M AN U
SC
Table 2: Sensitivity analysis excluding patients used infliximab No. of thiopurine P-Y of exposed exposure events IR 4734 7778 18 2.3 All cohort 7517 17 2.3 IFX users excluded 4557
HR CI 4.2 2.5-6.8 4.0 2.4-6.7
p <0.0001 <0.0001
AC C
EP
TE D
Notes: Rate is per 1000 person-years, HR: Hazard Ratio of lymphoma while on thiopurine compared to unexposed, CI: the 95% confidence interval.
ACCEPTED MANUSCRIPT
Colon Small bowel Skin Rectum Disseminated Liver, Spleen Small bowel Disseminated LN right cervical Disseminated LN right axilla LN mediastinum LN left cervical Colon Brain Disseminated LN unspecified LN unspecified
AZA+IFX MP MP MP AZA MP AZA AZA AZA MP AZA AZA MP MP AZA MP MP MP
0.7 6.1 4.0 3.8 9.1 1.6 0.7 0.1 6.9 2.8 3.7 5.8 7.6 1.5 2.0 3.7 1.7 6.3
LN left axillary Rectum Bone marrow Bone marrow LN right axilla
MP+MTX MP MP MP AZA
1.5 3.0 2.7 1.4 5.2
EP
AC C
M AN U
SC
RI PT
Lymphoma Site
Therapy Duration, Thiopurines years
TE D
Table 1: Lymphoma cases among thiopurine users Colitis Duration, Age, years Lymphoma Type ID years Sex Race Coincidental Lymphomas 1 70 M White 2.5 Follicular B cell 2 61 M White 17.3 Enteropathic T cell 3 54 M White 4.5 T-cell 4 77 M Unknown 2.2 DLBCL 5 50 M White 20.8 DLBCL 6 63 M White 21.1 DLBCL 7 38 M White 2.1 Follicular B cell 8 76 M White 0.7 Mantle cell 9* 49 M White 24.4 NHL unspecified 10 81 M White 5.3 DLBCL 11 73 M White 8.5 Mantle cell 12 70 M White 25.2 DLBCL 13 41 M White 15.3 Follicular B cell 14 71 M White 38.5 DLBCL 15 42 M White 3.8 DLBCL 16 66 M White 20.5 DLBCL 17* 76 M White 13.4 NHL unspecified 18* 72 M Unknown 6.8 NHL unspecified Non-Coincidental Lymphomas 19(1) 54 M White 10.2 Mantle cell 20(2) 77 M Unknown 21.7 DLBCL 21(3) 49 M White 22.9 Marginal Zone 22(4) 67 M White 5.7 Follicular B cell 23(5) 44 M Non-white 8.5 Mantle cell
Notes: DLBCL, diffuse large B-cell lymphoma; NHL, non-Hodgkin lymphoma; LN, lymph node; IFX, infliximab; AZA, azathioprine; MP, 6-mercaptopurine; MTX, methotrexate. * Insufficient records found in the VA system, diagnosis and/or care was provided by non-VA provider (1) Patient had MP for 2 years then switched to MTX for 2 years then was diagnosed with lymphoma (2) Lymphoma was diagnosed 1.2 year after stopping MP (3) Lymphoma was diagnosed 6.4 years after stopping MP (4) Lymphoma was diagnosed 3.4 years after stopping MP
ACCEPTED MANUSCRIPT
AC C
EP
TE D
M AN U
SC
RI PT
(5) Lymphoma was diagnosed 2.6 years after stopping AZA
ACCEPTED MANUSCRIPT
SC
12% 52% 36% 7% 93% 76% 11% 12%
TE D
EP AC C
RI PT
%
13% 8% 5% 1% 2%
M AN U
Table 2: Demographic characteristics and medication use among the entire ulcerative colitis cohort. Number Demographics 4348 Age Group at induction <40 19276 40-65 13267 >65 2467 Sex Female 34424 Male 28170 Race White 4188 Non-white 4533 Unknown Immune Suppression Use 4734 Thiopurines Any 3126 Azathioprine 1963 6-mercaptopurine 251 Infliximab 671 Methotrexate 5-Aminosalicylic Acid Use 22664 Any 16484 Oral Mesalamine 7973 Rectal Mesalamine 6071 Sulfasalazine 2774 Balsalazide 300 Olsalazine
61% 45% 22% 16% 8% 1%
ACCEPTED MANUSCRIPT
M AN U
SC
RI PT
Table 3: Cox proportional hazards regression analysis with time to lymphoma as the dependent variable. No. No. % Person Rate HR p UCI LCI total events Years 4,348 7 0.16 24,879 0.28 1 (reference) Age <40 19,276 70 0.36 116,225 0.60 2.0 0.08 0.9 4.4 40-65 13,267 65 0.49 83,667 0.78 2.6 0.02 1.2 5.7 >65 2,467 3 0.12 14,114 0.21 1 (reference) Sex Female 34,424 139 0.40 210,656 0.66 2.5 0.12 0.8 7.9 Male 28,170 117 0.42 172,120 0.68 1 (reference) Race Caucasians 4,188 13 0.31 24,499 0.53 0.9 0.71 0.5 1.6 Non Caucasians 4,533 12 0.26 28,152 0.43 0.6 0.11 0.3 1.1 Unknown 36,297* 119 0.33 199,046 0.60 1 (reference) Thiopurine Unexposed 18 0.38 7,778 2.31 4.2 <0.0001 2.5 6.8 During Exposure 4,734 4,662** 5 0.11 17,950 0.28 0.5 0.17 0.2 1.3 After Stopping
AC C
EP
TE D
Notes: Rate is per 1000 person-years, HR: Hazard Ratio, UCI and LCI: Upper and Lower limits of the 95% confidence interval respectively. * Total number of patients who contributed person-years to the Unexposed period was 32,157 patients who never used thiopurine and 4,140 patients who later on used thiopurine during the follow-up period. ** Among thiopurine users (4,734 patients), 4,662 patients had follow-up after stopping it.
ACCEPTED MANUSCRIPT
Thiopurine users
year of thiopurine use 1st (0-1) 2nd (1-2) 3rd (2-3) 4th (3-4) ≥ 4 years OVERALL
*No.
Thiopurine non-users
P-Y
E
IR
year of follow-up 1st (0-1) 2nd (1-2) 3rd (2-3) 4th (3-4) ≥ 4 years OVERALL
SC
Total No. included in the analysis
RI PT
Table 4: Life table analysis comparing the exposure and the follow-up periods
*No.
P-Y
E
IR
Age, Sex, Race adjusted HR on thiopurine users compared to non-users during each follow-up interval (Cox regression analysis)
HR
LCI
UCI
p
NNH
AC C
EP
TE D
M AN U
36,891 4,734 3,233 3 0.93 32,157 30,951 27 0.87 1.2 0.4 4.0 0.77 17,344 31,834 2,305 1,856 3 1.62 29,529 28,095 13 0.46 3.8 1.1 13.5 0.04 864 28,116 1,478 1,212 2 1.65 26,638 25,324 14 0.55 2.9 0.7 13.0 0.16 911 25,029 991 799 4 5.01 24,038 22,896 8 0.35 16.0 4.8 53.8 <0.0001 215 22,400 605 678 6 8.85 21,795 84,160 57 0.68 14.4 5.9 35.0 <0.0001 122 36,891 4,734 7,778 18 2.31 32,157 191426 119 0.62 4.2 2.5 6.8 <0.0001 591 Notes: IR: Incidence Rate, Rate is per 1000 person-years, *Total number of patients who contributed person-years to the follow-up or exposure interval, P-Y: person-years of exposure among thiopurine users and follow-up among thiopurine non-users, E: (Events) lymphoma cases observed during each interval, HR: Hazard Ratio, NNH: number needed to harm.
ACCEPTED MANUSCRIPT
Table 5: Incidence rate of lymphoma after different cumulative exposure rates.
E
After Stopping Thiopurine Year of Median FU stopping *No. of after stopping thiopurine stoppers (years) 1st (0-1) 2,429 3.6 2nd (1-2) 827 3.1 3rd (2-3) 487 3.6 4th (3-4) 386 3.3 ≥ 4 years 605 3.7 OVERALL 4,734 3.5
RI PT
While Using Thiopurine Year of thiopurine use *No. P-Y 1st (0-1) 4,734 3,233 2nd (1-2) 2,305 1,856 3rd (2-3) 1,478 1,212 4th (3-4) 991 799 ≥ 4 years 605 678 OVERALL 4,734 7,778
IR
P-Y of FU after stopping E
IR
AC C
EP
TE D
M AN U
SC
3 0.93 9,472 0 0 3 1.62 3,021 2 0.66 2 1.65 1,881 1 0.53 4 5.01 1,403 1 0.71 6 8.85 2,173 1 0.46 18 2.31 17,950 5 0.28 Notes: IR: Incidence Rate, Rate is per 1000 person-years, *Total number of patients who contributed person-years to the exposure or follow-up interval, P-Y: person-years of exposure while on thiopurine and follow-up after stopping it, E: (Events) lymphoma cases observed during each interval.
S0016-5085(13)01086-X 10.1053/j.gastro.2013.07.035 YGAST 58576
To appear in: Gastroenterology Accepted Date: 17 July 2013 Please cite this article as: Khan N, Abbas AM, Lichtenstein GR, Loftus Jr. EV, Bazzano LA, Risk of Lymphoma in Patients with Ulcerative Colitis Treated with Thiopurines—a Nationwide Retrospective Cohort Study, Gastroenterology (2013), doi: 10.1053/j.gastro.2013.07.035. This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain. All studies published in Gastroenterology are embargoed until 3PM ET of the day they are published as corrected proofs on-line. Studies cannot be publicized as accepted manuscripts or uncorrected proofs.
ACCEPTED MANUSCRIPT Lymphoma And Thiopurine In Ulcerative Colitis
Risk of Lymphoma in Patients with Ulcerative Colitis Treated with Thiopurines—a Nationwide Retrospective Cohort Study
RI PT
Nabeel Khan, MD1, 2, Ali M. Abbas, MD, MPH1, 2,3, Gary R. Lichtenstein, MD4, Edward V. Loftus Jr., MD5 , Lydia A. Bazzano, MD, PhD1, 6
(1) Section of Gastroenterology, Department of internal medicine at Southeast Louisiana
SC
Veterans Health Care System, New Orleans, LA, USA (2) Section of Gastroenterology and Hepatology, Department of internal medicine at Tulane University Health Sciences Center, New
M AN U
Orleans, LA, USA (3) Department of internal medicine at University of Florida, Gainesville, FL, USA (4) Division of Gastroenterology, Department of Medicine, University of Pennsylvania School of Medicine , Philadelphia, PA, USA (5) Division of Gastroenterology and, Hepatology, Department of Internal Medicine, Mayo Clinic, Rochester, MN, USA (6) Department of
TE D
Epidemiology and of Medicine at Tulane University Health Sciences , Center for Health Research at Ochsner Health System, New Orleans, LA, USA Grant Support:
EP
This study was funded in full by the Department of Veterans Affairs, Veterans Health Administration – Office of Research & Development Health Services. Grant number (VA
AC C
Project No. 425). Abbreviations:
5-ASA: 5-aminosalicylic acid 6-MP: 6-mercaptopurine Anti-TNF: Anti-tumor necrosis factor AZA: Azathioprine 1
ACCEPTED MANUSCRIPT Lymphoma And Thiopurine In Ulcerative Colitis
CDW: Corporate Data Warehouse EBV: Epstein-Barr virus
IBD: Inflammatory Bowel Disease IRR: Incidence rate ratio
IQR: Interquartile range PBM: Pharmacy Benefits Management
SIR: Standardized incidence ratios
M AN U
SEER: Surveillance Epidemiology and End Results
SC
ICD-9: International Classification of Diseases, Ninth Revision
RI PT
HR: Hazard ratios
SLVHCS: Southeast Louisiana Veterans Health Care System TMPT: Thiopurine methyltransferase
TE D
UC: Ulcerative colitis VA: Veterans Affairs healthcare system
Corresponding Author: to whom requests for reprints should be addressed:
EP
Nabeel Khan, M.D.
Assistant Professor of Medicine
AC C
Director, Inflammatory Bowel Disease Tulane University School of Medicine 1430 Tulane Avenue, SL35 New Orleans, LA 70112 Tel: 504-988-7505 Fax: 504-988-2188
2
ACCEPTED MANUSCRIPT Lymphoma And Thiopurine In Ulcerative Colitis
Email: [email protected], [email protected] Disclosure Statement: All authors have no potential conflicts (financial, professional, or personal) that are relevant to
RI PT
this manuscript. The study was conducted and the manuscript was written and reviewed solely by the authors
SC
Writing Assistance: Authors listed above wrote, edit and proof read the current report.
M AN U
Disclaimer:
Study sponsor had no role in study design, analysis and interpretation of the data and in the writing of the report. The contents of this report do not represent the views of the Department of Veterans Affairs or the United States Government. We would like to acknowledge the kind
TE D
support of Joel and Nancy Barnett Foundation. Specific author contributions:
Both Nabeel Khan and Ali Abbas contributed equally to the study. They contributed to study
EP
concept and design; acquisition of data; statistical analysis; medical chart review; interpretation
AC C
of the results; drafting of the manuscript; critical revision of the manuscript for important intellectual content. Gary R. Lichtenstein contributed to interpretation of the results; critical revision of the manuscript. Edward V. Loftus Jr. contributed to interpretation of the results; critical revision of the manuscript. Lydia Bazzano contributed to study design; statistical analysis supervision; interpretation of the results; critical revision of the manuscript.
3
ACCEPTED MANUSCRIPT Lymphoma And Thiopurine In Ulcerative Colitis
ABSTRACT Background & Aims: There is controversy over whether treatment of patients ulcerative
RI PT
colitis (UC) with thiopurines increases their risk of lymphoma. We evaluated the risk of lymphoma (ongoing, residual, and per year of therapy) among thiopurine-treated patients with
SC
UC.
Methods: We obtained nationwide data from the Veterans Affairs (VA) healthcare system, from
M AN U
2001 through 2011. We performed a retrospective cohort study, analyzing data from 36,891 patients with UC from their date of diagnosis at the VA until a diagnosis of lymphoma or October, 1st, 2011 (subjects followed for a median of 6.7 y). Thiopurine exposure was assessed from the VA pharmacy database. Patients who developed lymphoma were identified based on
TE D
ICD-9 codes and confirmed by manual chart review.
Results: Of the patients with UC, 4734 (13%) used thiopurines for a median of 1 y. Lymphomas
EP
developed in 119 patients who had not taken thiopurines, 18 who were taking the drugs, and 5 who had stopped taking them. The incidence rates of lymphoma were 0.60 per 1000 person-y
AC C
among patients who had not taken thiopurines, 2.31 among patients taking the drugs, and 0.28 among patients who had stopped taking them. The incidence rates of lymphoma during the 1st y, 2nd y, 3rd y, 4th y, and >4 y of thiopurine therapy were 0.9, 1.6, 1.6, 5, and 8.9 per 1000 personyears, respectively. The age-, sex- and race-adjusted hazard ratios of developing lymphoma were 4.2 (95% confidence interval, 2.5–6.8; P<.0001) while on thiopurines and 0.5 (95% confidence interval, 0.2–1.3; P=.17) after stopping them, compared to patients who had not taken thiopurines. 4
ACCEPTED MANUSCRIPT Lymphoma And Thiopurine In Ulcerative Colitis
Conclusions: Based on a retrospective, nationwide cohort study, patients with UC have a 4-fold increase in risk of lymphoma while taking thiopurines, compared with patients who have not
therapy reduces risk of lymphoma.
M AN U
INTRODUCTION
SC
Keywords: IBD, inflammatory bowel disease, cancer risk, medication
RI PT
taken these drugs. The risk increased gradually for successive ys of therapy. Stopping thiopurine
Ulcerative colitis (UC) is a chronic inflammatory disorder of unknown etiology. Based on the largest study of prevalence to date, we estimate that more than 700,000 Americans are currently living with UC 1. Corticosteroids are commonly used for induction of remission in UC . However, they are associated with a number of adverse effects, including weight gain, mood
TE D
2
disorders, and osteoporosis, and often lack long-term efficacy 3-6. These two issues have encouraged a number of steroid-sparing therapies to be employed in UC. Such therapies include
EP
thiopurines [azathioprine (AZA), and 6-mercaptopurine (6-MP)], anti-tumor necrosis factor (anti-TNF) agents such as infliximab and adalimumab. Although these medications spare the
AC C
patients the potential adverse effects of long-term corticosteroid use, they have their own spectrum of side effects. Lymphoma is one of the most concerning and potentially lethal adverse outcomes of immunosuppressive therapy and has been linked to thiopurine use 7. Immunosuppressive therapies may contribute to the development of lymphomas in a variety of ways. Decreased immune surveillance of Epstein-Barr virus (EBV) infected B-cells is thought to contribute to lymphoma development 8, and thiopurine use, in particular, has been
5
ACCEPTED MANUSCRIPT Lymphoma And Thiopurine In Ulcerative Colitis
implicated in the development of EBV-positive lymphomas 9-13. Moreover, thiopurine may cause direct damage to DNA by incorporating thiopurine nucleotides during replication, which renders DNA unstable and interferes with further replication cycles and repair mechanisms,
RI PT
resulting in mutations 14-17.
Thiopurine use is widely accepted as a risk factor for lymphoproliferative disorders including lymphoma in post-organ transplant settings. Some reports suggests that they increase
SC
the risk of lymphoproliferative diseases by up to 20- and 200-fold for renal and heart transplant recipients, respectively 7. However, there is no such consensus about the extent of this risk for
M AN U
patients with inflammatory bowel diseases who are treated with thiopurine. Some studies reported no increase in the risk 18, 19, while another study reported an increase of 31 times 20. To date, there are four population-based cohorts published about this issue in the United States and Europe, and they showed conflicting results 21-24. Similarly, the two published meta-analyses
TE D
failed to reach a consistency regarding this issue 25, 26.
The Veteran Affairs (VA) administration has the largest integrated healthcare system in the United States, serving approximately 8.3 million veterans each year 27. We took advantage
EP
of the VA population-based database to (1) estimate the risk of lymphoma among patients who were treated with thiopurine and compare that with UC patients who have not taken this class of
AC C
medication, (2) investigate the potential residual risk of lymphoma after discontinuation of thiopurine, and (3) evaluate the effect of the cumulative duration of thiopurine therapy on the incidence rate of lymphoma. METHODS
Study Population
6
ACCEPTED MANUSCRIPT Lymphoma And Thiopurine In Ulcerative Colitis
Nationwide data were obtained from the VA Pharmacy Benefits Management (PBM) and Corporate Data Warehouse (CDW) databases. Veterans who were seen and followed in the VA health care system from October 1st, 2001 to October 1st, 2011 were identified using the
RI PT
International Classification of Diseases, Ninth Revision (ICD-9) codes for UC (556.xx). The study was approved by the Institutional Review Boards of the Southeast Louisiana Veterans Health Care System (SLVHCS).
SC
Study Design
We conducted a retrospective cohort study in which UC patients were followed-up from
M AN U
the index date (first ICD-9 code for UC entered in their record in the VA system during the observation period) or the first time they filled a prescription for 5-aminosalicylic acid (5-ASA) compounds from the VA pharmacy if that preceded the entry of the first ICD-9 code (start date). The follow-up was concluded at the time the patients developed lymphoma (our outcome of
TE D
interest) or October 1st, 2011 (end of the observation period). To be included in the analysis, the patients had to have been followed in the VA system for at least one month (i.e. a minimum follow-up duration between the start and the end day of one month).
EP
Assessment of Lymphoma Outcome
The medical records of the identified UC patients were queried for the presence of
AC C
lymphoma diagnosis using ICD-9 codes (200-202.28, 202.7-202.98). Medical charts of the identified cases were manually reviewed to confirm the diagnosis of lymphoma according to preset criteria that considers biopsy results or a note from a hematologist as adequate proves for correct diagnosis. The date, type, and location of lymphoma were abstracted from the charts. We excluded prevalent cases of lymphoma (i.e. patients diagnosed with lymphoma prior to October 1st, 2001 or UC diagnosis). In order to identify the current and the residual effects that
7
ACCEPTED MANUSCRIPT Lymphoma And Thiopurine In Ulcerative Colitis
thiopurine therapy might have on the risk of lymphoma, we classified the identified cases into those occurring during or up to six months after thiopurine treatment (co-incidental) as compared to lymphomas diagnosed after discontinuation of thiopurine therapy for six months or more
RI PT
(non-coincidental). Assessment of Exposure and Other Predictors
Automated data abstraction was used to identify the periods of exposure to thiopurine
SC
during the follow-up period from the VA pharmacy database. Person-years of exposure to thiopurine were calculated in a time-dependent manner, accounting for the initiation and
M AN U
discontinuation of therapy, with a study subject potentially contributing person-years of followup to any or all the three exposure categories (1) unexposed: follow-up periods from all the patients who never filled thiopurine and the period prior to the first filling of thiopurine users, (2) during exposure: periods of exposure to thiopurine started at the filling day and ended when
TE D
the medication day supply was exhausted (3) after stopping: follow-up periods after the last fill of thiopurine was exhausted and the gaps between exposure periods if present 21, 23. Other predictors such as age at the start of follow-up (<40, 40-64, >65 years old), sex and race were
EP
also collected.
Infliximab exposure during the follow-up period was identified as present versus absent
AC C
manner due to its low prevalence in the VA system (1%). We calculated the incidence rates of lymphoma among infliximab users as a monotherapy or in combination with thiopurines and we performed a sensitivity analysis after excluding infliximab users to identify its effect on the observed lymphomas. Statistical Analysis
8
ACCEPTED MANUSCRIPT Lymphoma And Thiopurine In Ulcerative Colitis
To study the effect of ongoing and past thiopurine therapy on the risk of lymphoma, all UC patients were included, and the incidence rate of lymphoma (per 1000 person-years) was calculated for unexposed patients, those currently being treated with thiopurine, and for those
RI PT
who stopped thiopurine therapy. Thiopurine exposure was included in a multivariate Cox
regression model as a time-dependent variable. The age-, sex- and race-adjusted hazard ratios (HR) of lymphoma while on thiopurine treatment and after stopping it as compared to unexposed
SC
patients were calculated.
In order to improve the generalizability of our findings, we age-adjusted the
M AN U
aforementioned incidence rates using weights that represent the age group proportions of the 2000 United States standard population. Standardized incidence ratios (SIR) of non-Hodgkin lymphoma were calculated by comparing these rates with the age adjusted rate of non-Hodgkin lymphoma as reported by Surveillance Epidemiology and End Results (SEER) database28.
TE D
We investigated the effect of cumulative duration of thiopurine therapy on the incidence of lymphoma by performing a life table analysis with the follow-up time classified into five intervals at the end of each year (1st year, 2nd year, 3rd year, 4th year, and beyond 4 years) of exposure or follow-up
EP
(Table 4). For each interval, we identified the total population at the start of the interval, person year contribution and number of lymphoma cases that were diagnosed during interval. The latter two
AC C
parameters were used to calculate the per-year incidence rate of lymphoma for those using thiopurines and those who were not. We then compared the rate of lymphoma occurrence between the exposed and unexposed participants within each interval using a multivariate Cox regression analysis adjusted for age, sex, and race. The “number needed to harm” per each year of treatment with thiopurines was calculated from the incidence rates of lymphoma among exposed and non-exposed. To identify whether different exposure durations had affected the observed residual risk of lymphoma after discontinuation, we examined the incidence rate of lymphoma after discontinuation of
9
ACCEPTED MANUSCRIPT Lymphoma And Thiopurine In Ulcerative Colitis thiopurines in the 1st, 2nd, 3rd, 4th, and after 4 years of use, by dividing the number of events by the person-years of follow-up after stopping the thiopurine. An evaluation of the effect of different dosage utilized was performed by comparing those who
RI PT
used thiopurines at below median doses against the rest of thiopurine users. No statistically significant difference was observed based on different dosing groups. However, our ability to examine this issue was limited due to the absence of weight information needed for calculating dosing parameters, the
SC
differences in the formulations and dosages of thiopurines used at the VA hospitals, dose fluctuation over treatment period, and the limited power to perform stratified analysis.
M AN U
All data extraction and analysis procedures were conducted using SAS version 9.2 for Windows (SAS Institute Inc., Cary, NC, USA). Sensitivity and Reliability of Case Analyses
Although a previous study has established the sensitivity and the specificity of using ICD-9 codes for UC diagnosis in the VA system to be 84% and 99%, respectively 29, we
reliability of our results.
TE D
performed a comprehensive sensitivity analysis to establish the validity of our methods and
In the primary analysis we included all the identified patients with a minimum of 1 ICD-9
EP
code for UC in their records. In sensitivity analyses, we included patients who had ≥2 UC codes,
AC C
then ≥3, then ≥4. We further added the number of times a 5-ASA prescription (oral or rectal mesalamine, sulfasalazine, balsalazide, and olsalazine) was filled in the VA pharmacy database during the follow-up period as a criterion (once, twice, three or more times). Based on the combination of these 2 criteria, we repeated our analysis to evaluate the robustness of our results. We then randomly selected 100 patients from those who had ≥4 ICD-9 codes for UC recorded in their medical record and ≥3 fills of 5-ASA from the pharmacy database and manually reviewed
10
ACCEPTED MANUSCRIPT Lymphoma And Thiopurine In Ulcerative Colitis
their medical records to calculate the positive predictive value of this method of identifying UC patients. To identify whether short-term use of thiopurines affected our results, we performed a sensitivity
RI PT
analysis by rerunning the multivariate analysis after excluding patients with less than six months, less than one year, and less than one and a half year of exposure. The observed hazard ratios of lymphoma while taking thiopurines when requiring a minimum of six months, one year, and one and a half year of
SC
exposure were consistent with the one obtained by including the entire cohort (Appendix 1, Table 1). To investigate whether the observed reduction in the risk of lymphoma upon discontinuation was
M AN U
associated with receiving colectomy, we performed a sensitivity analysis after excluding patients who had colectomy over the follow-up period. The observed reduction in the risk of lymphoma remained consistent (Appendix 1, Table 2).
RESULTS
We identified 37,191 UC patients followed in the VA system during the observational
TE D
period. A total of 325 patients had diagnostic codes for lymphoma in their records. Manual chart review confirmed 253 cases of lymphoma (among them, 111 prevalent cases of lymphoma were excluded). We included only incident lymphoma cases (happened during the follow-up
EP
time) confirmed by manual chart review in our analysis. Among the remainder of the patients
AC C
with a diagnostic code for lymphoma, we found other hematological conditions in 18 patients, suspected lymphoma in 11 patients, and no evidence of lymphoma in 43 patients. All these groups were reclassified as non-lymphoma patients. Patients with less than one month of followup in the VA system were also excluded (n=189). Of the 142 confirmed lymphoma cases, 119 occurred among those who never used a thiopurine, while 23 cases occurred among thiopurine users: 18 of these occurred during thiopurine use or within 6 months of discontinuation (co-
11
ACCEPTED MANUSCRIPT Lymphoma And Thiopurine In Ulcerative Colitis
incidental) and 5 occurred more than one year after discontinuation of thiopurine use (noncoincidental) (Table 1). We included 36,891 patients with 224,773 total person-years of follow-up in our
RI PT
analyses. Median follow-up was 6.7 years [interquartile range (IQR), 3.2-9.2]. Patients were predominantly white (76%) males (93%) with a median age at the start of the follow-up period of 60 years (IQR, 50-69). Thiopurines were used by 4,734 patients (13%) and 5-ASA compounds
SC
were used by 22,664 (61%) throughout the follow-up period (Table 2).
Median duration of follow-up among the thiopurine-unexposed was 5.72 years (IQR,
M AN U
2.25-8.87) with total person-years for this period available for analysis of 199,046. The median duration of thiopurine exposure was 0.97 years (IQR, 0.28-2.5) with total person-years for this period available for analysis of 7,778. Median duration of follow-up after stopping thiopurine was 3.52 years (IQR, 1.63-5.77) with total person-years for this period available for analysis of
TE D
17,950.
The overall incidence rate of lymphoma for the entire cohort was 0.63 cases per 1000 person-years. The incidence rate of lymphoma was 2.31 and 0.28 per 1000 person-years while
EP
on thiopurine and after stopping it, respectively, compared to 0.60 per 1000 person-years for unexposed patients. The age- and thiopurine-stratified incidence rate of lymphoma is shown in
AC C
(Figure 1). Using Cox regression analysis, the age-, sex- and race-adjusted HRs of developing lymphoma while on thiopurine and after stopping it were 4.2 (95% CI, 2.5-6.8, p<0.0001) and 0.5 (95% CI, 0.2-1.3, p=0.17), respectively, as compared to unexposed patients. Other predictors of lymphoma were age >65 years (HR, 2.6; 95% CI, 1.2-5.7; p=0.01) and male gender (HR, 2.5; 95% CI, 0.8-7.9; p=0.12), (Table 3).
12
ACCEPTED MANUSCRIPT Lymphoma And Thiopurine In Ulcerative Colitis
The age-adjusted incidence rate of lymphoma was 1.46 and 0.19 per 1000 person-years while on thiopurine and after stopping it, respectively, compared to 0.32 per 1000 person-years for unexposed patients. The SIR of non-Hodgkin lymphoma while on thiopurine was 7.4 (95%
RI PT
CI, 4.6-11.9; p<0.0001) as compared to the age standardized SEER population (Appendix 2). The results of the thiopurine cumulative effect analysis are presented in Table 4. The incidence rate of lymphoma during the 1st, 2nd, 3rd, 4th and after 4 years of thiopurine use was 0.9, 1.6, 1.6, 5, and
SC
8.9 cases per 1000 person-years, respectively. Using the incidence rate among unexposed patients during the successive follow-up intervals as a reference, the age- sex- race- adjusted HR of lymphoma was 1.2,
M AN U
3.8, 2.9, 16 (p<0.0001), and 14.4 (p<0.0001) during the 1st, 2nd, 3rd, 4th and >4 years of thiopurine use, respectively. The risk of lymphoma was statistically significantly elevated in the fourth and subsequent years of exposure to thiopurine. The number needed to harm was 17,344 patients for the first year compared to 122 patients for those with more than four years of therapy.
The incidence rates of lymphoma after thiopurine discontinuation in the 1st, 2nd, 3rd, 4th
TE D
and after 4 years of use was 0, 0.66, 0.53, 0.71 and 0.46 per 1000 person-years respectively. These rates were within the observed range among the unexposed group (overall, 0.62; range 0.35-0.87 per 1000 person-years). The number of patients, number of lymphomas following
AC C
Infliximab use
EP
thiopurine discontinuation, and the total person-years analyzed are presented in Table 5.
We had a very low rate of infliximab use in this cohort. Among the included UC patients, only 251 (1%) were treated with infliximab during the follow-up period. Of these, 177 (71% of infliximab users) patients used infliximab concomitantly with thiopurines. Among the latter group only one lymphoma was observed. Among infliximab monotherapy users, none developed lymphoma. The incidence rate among infliximab users is presented in (Appendix 3, Table 1). Although the incidence rate among patients exposed to thiopurine and infliximab was
13
ACCEPTED MANUSCRIPT Lymphoma And Thiopurine In Ulcerative Colitis
higher compared to those exposed only to thiopurine (3.84 compared to 2.26 per 1000 personyears respectively), our study is underpowered to detect the difference between these rates because of the limitation of low sample size of infliximab users. To identify the extent of
RI PT
infliximab effect on the observed association between thiopurine use and lymphoma risk, we performed a sensitivity analysis by excluding all infliximab users (Appendix 3, Table 2). The observed hazard ratio of lymphoma did not change after excluding infliximab users, indicating
SC
that the observed association is associated predominantly with thiopurine use. Results of Sensitivity Analyses
M AN U
Using the different combinations of the number of ICD-9 codes indicating UC diagnoses and the number of 5-ASA compounds filled in the nation-wide pharmacy database as inclusion criteria, the HR of lymphoma while on thiopurine showed consistency ranging from 3.9-5.6 with overlapping confidence intervals. After reviewing the medical charts of a randomly-selected 100
TE D
patients with a minimum of 4 ICD-9 codes indicating a diagnosis of UC and at least 3 fills of 5ASA compounds, we found that 92 patients had a confirmed diagnosis of UC, 4 had indeterminate colitis, 2 had no medical charts in the VA CAPRI system, and 2 had Crohn’s
EP
disease. Thus, the positive predictive value of this definition in identifying UC patients was
AC C
92%. Full results of sensitivity analyses are presented in (Appendix 1, Table 3). DISCUSSION
In this nationwide cohort of UC patients from the VA healthcare system, treatment with a thiopurine was associated with an approximate four-fold and a seven-fold increase in the risk of lymphoma relative to UC patients who did not use a thiopurine and to SEER general population respectively. This finding is very consistent with the previous population-based prospective cohort study reported by Beaugerie et al. 2009 21. Additionally, we have shown that the
14
ACCEPTED MANUSCRIPT Lymphoma And Thiopurine In Ulcerative Colitis
magnitude of this risk is significantly associated with the duration of treatment, with therapy duration of four years as a threshold for achieving statistical significance. Furthermore, we have shown that stopping the thiopurine reduced the risk of lymphoma during a median follow-up
RI PT
time of 3.5 years.
Several autoimmune and chronic inflammatory disorders are associated with increased risk of lymphoma 30-34. It is reasonable to think that Inflammatory Bowel Disease (IBD) patients
SC
would have a baseline increased lymphoma risk due to the presence of a chronic inflammatory process. However, many well-designed, population-based studies have failed to show any
M AN U
increased risk 22, 35-37. Unfortunately, there is less consistency in the literature regarding the effect of thiopurine on the risk of lymphoma in the setting of IBD.
Whether the increase in risk of lymphoma associated with thiopurine is constant (risk starts at the first intake and stops upon cessation) or cumulative (affected by the duration and/or
TE D
the dose of thiopurine treatment) is still controversial. Although many IBD and post-transplant studies have concluded that the risk is constant 7, 21, 22, several other studies have shown a cumulative dose/duration effect among rheumatoid arthritis patients 38, 39. A recent study by
EP
Nguyen et al. in 2009 showed a significant increase in the frequency of mutations with both cumulative dose and duration of thiopurine treatment for IBD patients after controlling for
AC C
disease duration 40. Our study is the first to show the cumulative effect of thiopurine treatment on the risk of lymphoma among UC patients. Although the risk of mutation described by Nguyen et al. lacked a treatment duration threshold (a continuous increase along with therapy duration), the risk in our cohort showed an interesting pattern. Lymphoma risk increased to 1.23.8 times for the first three years of therapy, but mostly it lacked the statistical significance.
15
ACCEPTED MANUSCRIPT Lymphoma And Thiopurine In Ulcerative Colitis
However, after three years of thiopurine therapy, the risk increased significantly to 14-16 times as compared to those not on thiopurines (Table 4, Figure 2). Previous studies reported a significant increase in the risk of lymphoma among IBD
RI PT
patients on thiopurines, with an effect size of 2-31 times 9, 20, 21, 41, 42, but none investigated the effect of treatment duration on that risk. The inconsistency in the reported magnitude of the observed risk and the lack of data about the effect of long-term therapy have made weighing the
SC
benefits versus the risks of initiating or continuing thiopurine therapy a challenging task for both patients and clinicians. Lewis et al. in 2000 assessed the gain in quality-adjusted life expectancy
M AN U
resulting from AZA treatment in Crohn’s disease against the potential for serious side effects including the risk of lymphoma 43, and concluded that alternative therapy is the preferred treatment strategy if the increased risk of lymphoma exceeded 9.8 folds compared to non-users. Using the results from Lewis et al. analysis (in the absence of comparable study that dealt with
TE D
UC patients), our study shows that the benefits of using thiopurine outweigh the risk of lymphoma in the first three years of therapy. For the fourth year of therapy, the magnitude of the risk started to approach (IRR=8.8) and after the fourth year it exceeded it (IRR=15.5) the
EP
proposed threshold of 9.8 by Lewis et al 43. Thus considering a withdrawal of therapy trial within three years of therapy onset is a reasonable option.
AC C
It has been reported that men had a higher risk of lymphoma than women 44, 45. In our cohort, the overall incidence rate was 0.66 compared to 0.21 cases per 1000 person-years for men and women, respectively. After adjusting for age, race and thiopurine use, men had a nonsignificant trend towards developing more lymphoma compared to women (HR= 2.5, 0.8-7.8, p=0.1). Failure to achieve statistical significance is attributed to the very small number of women in the military veteran population (2467 patients, 7%).
16
ACCEPTED MANUSCRIPT Lymphoma And Thiopurine In Ulcerative Colitis
In the absence of comparable treatment with a lower risk of malignancy, thiopurines remain a mainstay in treating steroid-dependent IBD patients. However, there are several measures that can be undertaken that may reduce the potential risks of therapy. First of all,
RI PT
patients for thiopurine therapy should be selected with caution, especially among older males. Secondly, assessment with thiopurine methyltransferase (TMPT) genotype or enzyme activity with subsequent weight-based dose adjustment and monitoring will result in less prolonged
SC
leukopenia. Finally, one could consider a trial of withdrawal of thiopurine therapy after three years, especially in cases of long-term remission, as this group has been demonstrated to have a
M AN U
lower recurrence rate 46, 47.
Our study is the largest to examine the risk of lymphoma associated with thiopurine therapy in UC patients to date. Individual level patient data was available and referral bias is unlikely to exist in our analysis due to the fact that the database was derived from a nationwide
TE D
source in which all the states are represented. The availability of pharmacy data allowed us to objectively assess the extent of exposure to thiopurine over extended periods of time and to identify the exact date of therapy withdrawal rather than depending on the self-reported or health
EP
provider-reported information about the medication exposure 21. Furthermore, we performed a stratified analysis to identify the residual risk based after therapy withdrawal according to the
AC C
cumulative duration of prior exposure. Additionally, our results were consistent with different study designs 48, and with extensive sensitivity analyses. Finally, the VA database has been previously validated and used frequently for research regarding inflammatory bowel disease 29, 49-51
.
The investigators acknowledge some limitations of the present work. Firstly, although this study is based on the largest integrated nationwide healthcare system in the United States 27,
17
ACCEPTED MANUSCRIPT Lymphoma And Thiopurine In Ulcerative Colitis
it is not a population-based cohort in the full sense due to the different demographics of the VA compared to the general United States population, specifically the age, race, and sex distribution. The VA population is predominantly comprised of middle- to old-aged Caucasian-males, which
RI PT
can limit the external validity of the study to patients within these demographics.
Due to the fact that we are investigating the rate of lymphoma among a high risk
population (old, white, males), our observed incidence rate might be higher than what might be
SC
expected from the general population databases. Nonetheless, we believe that our multivariate hazard ratio of thiopurine exposure is not overestimated, as we are comparing within the same
M AN U
population (VA thiopurine-exposed patients with UC compared to VA thiopurine-unexposed patients with UC). We tried to limit this effect by adjusting our incidence rates according to the age distribution of the US census in 2000, the same reference population used in SEER database. This increased the comparability of these rates with those derived from the general population.
TE D
However, male gender and white race are other risk factors for lymphoma that we were not able to control for simultaneously with the age to calculate the standardized rates28. Thus our standardized incidence rates and the ratios driven from them might be still overestimated.
EP
Secondly, the study is susceptible to limitations inherent to the retrospective design. The retrospective design is the most suitable way to address this question due to the low incidence
AC C
rate of lymphoma and the ability to obtain a long follow-up period. Thirdly, there is a possibility that thiopurine exposure as assessed from the pharmacy data does not completely reflect the extent of exposure due to the presence of dual health insurance eligibility and additional sources of medication other than the VA pharmacy. The investigators believe that this will have a very limited effect on the results of our analysis, as previous reports showed a very high rate of VA pharmacy use by veterans irrespective of their dual eligibility status due its convenience and
18
ACCEPTED MANUSCRIPT Lymphoma And Thiopurine In Ulcerative Colitis
financial plausibility 52-54. Lastly, EBV status was unknown for almost all of the patients, as it is rarely recorded as a diagnostic code and the chart review did not reveal further information related to EBV status.
RI PT
In summary, in this nationwide cohort of UC patients followed in the VA system, patients were four times more likely to develop lymphoma while receiving a thiopurine compared to
Stopping thiopurine therapy reduced the risk of lymphoma. Figures Legends:
SC
unexposed patients. The risk of lymphoma increased gradually for successive years of therapy.
M AN U
Figure 1: Incidence rate of lymphoma stratified by age and thiopurine use Figure 2: Age, Sex, Race adjusted HR of lymphoma while on thiopurine compared to unexposed
AC C
EP
TE D
during each follow-up interval (Cox regression analysis)
19
ACCEPTED MANUSCRIPT Lymphoma And Thiopurine In Ulcerative Colitis
REFERENCES 1.
Kappelman MD, Rifas-Shiman SL, Kleinman K, et al. The prevalence and geographic
RI PT
distribution of Crohn's disease and ulcerative colitis in the United States. Clin Gastroenterol Hepatol 2007;5:1424-9. 2.
Ford AC, Bernstein CN, Khan KJ, et al. Glucocorticosteroid therapy in inflammatory
SC
bowel disease: systematic review and meta-analysis. Am J Gastroenterol 2011;106:590-9; quiz 600.
Gurwitz JH, Bohn RL, Glynn RJ, et al. Glucocorticoids and the risk for initiation of
M AN U
3.
hypoglycemic therapy. Arch Intern Med 1994;154:97-101. 4.
Stuck AE, Minder CE, Frey FJ. Risk of infectious complications in patients taking glucocorticosteroids. Rev Infect Dis 1989;11:954-63.
Lukert BP, Raisz LG. Glucocorticoid-induced osteoporosis. Rheum Dis Clin North Am 1994;20:629-50.
6.
Mankin HJ. Nontraumatic necrosis of bone (osteonecrosis). N Engl J Med
EP
1992;326:1473-9. 7.
TE D
5.
Opelz G, Döhler B. Lymphomas after solid organ transplantation: a collaborative
8.
AC C
transplant study report. Am J Transplant 2004;4:222-30. Pietersma F, Piriou E, van Baarle D. Immune surveillance of EBV-infected B cells and the development of non-Hodgkin lymphomas in immunocompromised patients. Leuk Lymphoma 2008;49:1028-41.
20
ACCEPTED MANUSCRIPT Lymphoma And Thiopurine In Ulcerative Colitis
9.
Dayharsh GA, Loftus EV, Sandborn WJ, et al. Epstein-Barr virus-positive lymphoma in patients with inflammatory bowel disease treated with azathioprine or 6-mercaptopurine. Gastroenterology 2002;122:72-7. Larvol L, Soule JC, Le Tourneau A. Reversible lymphoma in the setting of azathioprine therapy for Crohn's disease. N Engl J Med 1994;331:883-4.
11.
RI PT
10.
Munkholm P, Langholz E, Davidsen M, et al. Intestinal cancer risk and mortality in
12.
SC
patients with Crohn's disease. Gastroenterology 1993;105:1716-23.
Wakefield AJ, Fox JD, Sawyerr AM, et al. Detection of herpesvirus DNA in the large
M AN U
intestine of patients with ulcerative colitis and Crohn's disease using the nested polymerase chain reaction. J Med Virol 1992;38:183-90. 13.
Wong NA, Harrison DJ. Colorectal neoplasia in ulcerative colitis-recent advances. Histopathology 2001;39:221-34.
Krynetskaia NF, Krynetski EY, Evans WE. Human RNase H-mediated RNA cleavage
TE D
14.
from DNA-RNA duplexes is inhibited by 6-deoxythioguanosine incorporation into DNA. Mol Pharmacol 1999;56:841-8.
Krynetskaia NF, Cai X, Nitiss JL, et al. Thioguanine substitution alters DNA cleavage
EP
15.
mediated by topoisomerase II. FASEB J 2000;14:2339-44. Ling YH, Chan JY, Beattie KL, et al. Consequences of 6-thioguanine incorporation into
AC C
16.
DNA on polymerase, ligase, and endonuclease reactions. Mol Pharmacol 1992;42:802-7. 17.
Somerville L, Krynetski EY, Krynetskaia NF, et al. Structure and dynamics of thioguanine-modified duplex DNA. J Biol Chem 2003;278:1005-11.
18.
Connell WR, Kamm MA, Dickson M, et al. Long-term neoplasia risk after azathioprine treatment in inflammatory bowel disease. Lancet 1994;343:1249-52.
21
ACCEPTED MANUSCRIPT Lymphoma And Thiopurine In Ulcerative Colitis
19.
Warman JI, Korelitz BI, Fleisher MR, et al. Cumulative experience with short- and longterm toxicity to 6-mercaptopurine in the treatment of Crohn's disease and ulcerative colitis. J Clin Gastroenterol 2003;37:220-5. Farrell RJ, Ang Y, Kileen P, et al. Increased incidence of non-Hodgkin's lymphoma in
RI PT
20.
inflammatory bowel disease patients on immunosuppressive therapy but overall risk is low. Gut 2000;47:514-9.
Beaugerie L, Brousse N, Bouvier AM, et al. Lymphoproliferative disorders in patients
SC
21.
study. Lancet 2009;374:1617-25. 22.
M AN U
receiving thiopurines for inflammatory bowel disease: a prospective observational cohort
Lewis JD, Bilker WB, Brensinger C, et al. Inflammatory bowel disease is not associated with an increased risk of lymphoma. Gastroenterology 2001;121:1080-7.
23.
Herrinton LJ, Liu L, Weng X, et al. Role of thiopurine and anti-TNF therapy in
24.
TE D
lymphoma in inflammatory bowel disease. Am J Gastroenterol 2011;106:2146-53. Lakatos PL, Lovasz BD, David G, et al. The risk of lymphoma and immunomodulators in patients with inflammatory bowel diseases: Results from a population-based cohort in
25.
EP
Eastern Europe. J Crohns Colitis 2012.
Kandiel A, Fraser AG, Korelitz BI, et al. Increased risk of lymphoma among
AC C
inflammatory bowel disease patients treated with azathioprine and 6-mercaptopurine. Gut 2005;54:1121-5. 26.
Masunaga Y, Ohno K, Ogawa R, et al. Meta-analysis of risk of malignancy with immunosuppressive drugs in inflammatory bowel disease. Ann Pharmacother 2007;41:21-8.
22
ACCEPTED MANUSCRIPT Lymphoma And Thiopurine In Ulcerative Colitis
27.
The National Center for Veterans Analysis and Statistics (NCVAS), Trends in the Utilization of VA Programs and Services. Washington, DC.: Available from: http://www.va.gov/vetdata/docs/quickfacts/Utilization-slideshow.pdf, 2012. National Cancer Institute - Surveillance Epidemiology and End Results (SEER) Stat Fact
RI PT
28.
Sheets: Non-Hodgkin Lymphoma. 2012, Accessed 5-21-2013, Available in: http://seer.cancer.gov/statfacts/html/nhl.html, 2012.
Thirumurthi S, Chowdhury R, Richardson P, et al. Validation of ICD-9-CM diagnostic
SC
29.
codes for inflammatory bowel disease among veterans. Dig Dis Sci 2010;55:2592-8. Smedby KE, Askling J, Mariette X, et al. Autoimmune and inflammatory disorders and
M AN U
30.
risk of malignant lymphomas--an update. J Intern Med 2008;264:514-27. 31.
Mellemkjaer L, Pfeiffer RM, Engels EA, et al. Autoimmune disease in individuals and close family members and susceptibility to non-Hodgkin's lymphoma. Arthritis Rheum
32.
TE D
2008;58:657-66.
Zintzaras E, Voulgarelis M, Moutsopoulos HM. The risk of lymphoma development in autoimmune diseases: a meta-analysis. Arch Intern Med 2005;165:2337-44. Baecklund E, Ekbom A, Sparén P, et al. Disease activity and risk of lymphoma in
EP
33.
patients with rheumatoid arthritis: nested case-control study. BMJ 1998;317:180-1. Baecklund E, Iliadou A, Askling J, et al. Association of chronic inflammation, not its
AC C
34.
treatment, with increased lymphoma risk in rheumatoid arthritis. Arthritis Rheum 2006;54:692-701. 35.
Askling J, Brandt L, Lapidus A, et al. Risk of haematopoietic cancer in patients with inflammatory bowel disease. Gut 2005;54:617-22.
23
ACCEPTED MANUSCRIPT Lymphoma And Thiopurine In Ulcerative Colitis
36.
Loftus EV, Tremaine WJ, Habermann TM, et al. Risk of lymphoma in inflammatory bowel disease. Am J Gastroenterol 2000;95:2308-12.
37.
Karlén P, Löfberg R, Broström O, et al. Increased risk of cancer in ulcerative colitis: a
38.
RI PT
population-based cohort study. Am J Gastroenterol 1999;94:1047-52.
Asten P, Barrett J, Symmons D. Risk of developing certain malignancies is related to duration of immunosuppressive drug exposure in patients with rheumatic diseases. J
39.
SC
Rheumatol 1999;26:1705-14.
Silman AJ, Petrie J, Hazleman B, et al. Lymphoproliferative cancer and other malignancy
Ann Rheum Dis 1988;47:988-92. 40.
M AN U
in patients with rheumatoid arthritis treated with azathioprine: a 20 year follow up study.
Nguyen T, Vacek PM, O'Neill P, et al. Mutagenicity and potential carcinogenicity of thiopurine treatment in patients with inflammatory bowel disease. Cancer Res
41.
TE D
2009;69:7004-12.
Kinlen LJ. Incidence of cancer in rheumatoid arthritis and other disorders after immunosuppressive treatment. Am J Med 1985;78:44-9. Armstrong RG, West J, Card TR. Risk of cancer in inflammatory bowel disease treated
EP
42.
with azathioprine: a UK population-based case-control study. Am J Gastroenterol
43.
AC C
2010;105:1604-9.
Lewis JD, Schwartz JS, Lichtenstein GR. Azathioprine for maintenance of remission in Crohn's disease: benefits outweigh the risk of lymphoma. Gastroenterology 2000;118:1018-24.
44.
Results NCI-SEaE. SEER Stat Fact Sheets: Non-Hodgkin Lymphoma, 2012.
24
ACCEPTED MANUSCRIPT Lymphoma And Thiopurine In Ulcerative Colitis
45.
Bosetti C, Levi F, Ferlay J, et al. Incidence and mortality from non-Hodgkin lymphoma in Europe: the end of an epidemic? Int J Cancer 2008;123:1917-23.
46.
Smith MA, Irving PM, Marinaki AM, et al. Review article: malignancy on thiopurine
RI PT
treatment with special reference to inflammatory bowel disease. Aliment Pharmacol Ther 2010;32:119-30. 47.
Bitton A, Peppercorn MA, Antonioli DA, et al. Clinical, biological, and histologic
SC
parameters as predictors of relapse in ulcerative colitis. Gastroenterology 2001;120:1320.
Abbas A, Koleva Y, Khan N. Lymphoproliferative Disorders in Patients Receiving
M AN U
48.
Thiopurines for Ulcerative Colitis: A Nationwide 10-Year Retrospective Cohort from the Veterans Affairs Healthcare System. Am J Gastroenterol 2012;107:S693. 49.
Khan N, Abbas AM, Bazzano LA, et al. Long-term oral mesalazine adherence and the
TE D
risk of disease flare in ulcerative colitis: nationwide 10-year retrospective cohort from the veterans affairs healthcare system. Aliment Pharmacol Ther 2012;36:755-64. 50.
Hou JK, Kramer JR, Richardson P, et al. Risk of colorectal cancer among Caucasian and
51.
EP
African American veterans with ulcerative colitis. Inflamm Bowel Dis 2012;18:1011-7. Sonnenberg A, Richardson PA, Abraham NS. Hospitalizations for inflammatory bowel
52.
AC C
disease among US military veterans 1975-2006. Dig Dis Sci 2009;54:1740-5. Veterans Health Care: Use of VA Services by Medicare-Eligible Veterans, U.S. Government Accountability Office Report, 1994. 53.
Morgan RO, Petersen LA, Hasche JC, et al. VHA pharmacy use in veterans with Medicare drug coverage. Am J Manag Care 2009;15:e1-8.
25
ACCEPTED MANUSCRIPT Lymphoma And Thiopurine In Ulcerative Colitis
Wolinsky FD, Miller TR, An H, et al. Dual use of Medicare and the Veterans Health
EP
TE D
M AN U
SC
RI PT
Administration: are there adverse health outcomes? BMC Health Serv Res 2006;6:131.
AC C
54.
26
ACCEPTED MANUSCRIPT
Appendix 1, results of sensitivity analyses
entire cohort excluding < 6 months excluding < 1 year excluding < 1.5 years
36891 35172 34462 33986
4734 3015 2305 1829
7778 7376 6850 6260
18 17 15 15
2.3 2.3 2.2 2.4
HR while on treatment p
UCI LCI
4.2 4.0 3.8 4.1
2.5 2.4 2.2 2.4
RI PT
Table 1: Sensitivity analysis, excluding short term thiopurine users P-Y No. of No. of patients exposed exposure included patients analyzed events IR
<0.0001 <0.0001 <0.0001 <0.0001
M AN U
SC
Notes: P-Y: person-years, IR: Incidence rate per 1000 person-years, HR: Hazard Ratio, UCI and LCI: Upper and Lower limits of the 95% confidence interval respectively
EP
TE D
Table 2: Sensitivity analysis excluding patients who required colectomy No. No. total events IR Thiopurine P-Y 36,297 199,046 119 0.6 All Cohort Unexposed 4,734 7,778 18 2.31 During Exposure 4,662 17,950 5 0.28 After Stopping 33,308 177,540 103 0.58 Unexposed Patients who did not require During Exposure 4,158 7,145 17 2.38 colectomy 4158 14,956 4 0.27 After Stopping
HR p 1 (reference) 4.2 <0.0001 0.5 0.17 1 (reference) 4.4 <0.0001 0.5 0.22
95%CI 2.5-6.8 0.2-1.3 2.6-7.3 0.19-1.5
AC C
Notes: P-Y: person years contributions, events: lymphoma cases, IR: Incidence rate per 1000 person-years, HR: Hazard Ratio, 95%CI: the 95% confidence interval.
6.8 6.8 6.5 7.1
ACCEPTED MANUSCRIPT
One time One time One time One time
22,664 19,285 17,257 15,775
84 70 65 57
5.4 5.2 5.6 4.8
3.2 2.9 3.1 2.5
9.2 9.2 10.0 9.1
<0.0001 <0.0001 <0.0001 <0.0001
One time Two times Three times Four times
Two times Two times Two times Two times
21,117 18,262 16,505 15,187
78 65 61 54
4.9 4.6 4.9 4.0
2.8 2.5 2.6 2.0
8.6 8.5 9.2 8.1
<0.0001 <0.0001 <0.0001 <0.0001
One time Two times Three times Four times
Three times Three times Three times Three times
19,849 17,336 15,791 14,608
78 65 61 54
4.8 4.5 4.8 3.9
2.7 2.4 2.6 2.0
8.4 8.3 9.0 7.9
<0.0001 <0.0001 <0.0001 <0.0001
M AN U
SC
One time Two times Three times Four times
RI PT
Table 3: Adjusted hazard ratios of lymphoma while using thiopurine based on different ulcerative colitis definition criteria, sensitivity analysis results. Included Events UC ICD9 no. 5-ASA fills no. patients no. no. HR UCI LCI p 36,891 142 4.2 2.5 6.8 <0.0001 One time 27,068 105 4.2 2.4 7.3 <0.0001 Two times 23,035 93 4.6 2.6 8.0 <0.0001 Three times 20,401 80 4.0 2.1 7.4 <0.0001 Four times
AC C
EP
TE D
Notes: HR: Hazard Ratio, UCI and LCI: Upper and Lower limits of the 95% confidence interval respectively
ACCEPTED MANUSCRIPT
Appendix 2: Age standardization results Table: Incidence rate of NHL among VA population compared to SEER population SEER VA Unexposed VA During Exposure VA After Stopping
SIR reference group 1.66 7.47 0.98
95% CI 0.95-2.90 4.67-11.98 0.52-1.84
RI PT
*IR 0.20 0.33 1.47 0.19
AC C
EP
TE D
M AN U
SC
IR: Incidence rate, SIR: Standardized incidence ratio, 95%CI: 95% confidence interval limits. *Rates are per 1000 person-year and are age-adjusted to the 2000 US Standard Population
ACCEPTED MANUSCRIPT
Appendix 3: Infliximab use analysis results
IRR 95%CI 1 3.6 2.2- 6.0 NA 6.2 0.8- 44.2
RI PT
Table 1: Incidence rate of lymphoma by thiopurine and infliximab exposure No. of P-Y P-Y patients Follow-up Exposure events IR 32083 190944 . 119 0.62 Unexposed 4557 . 7517 17 2.26 Thiopurine exposed 74 482 . 0 0.00 Infliximab exposed . 260 1 3.84 Thiopurine and Infliximab 177
Notes: P-Y: person-years of follow-up, IR: Incidence Rate, Rate is per 1000 person-years, IRR: Incidence Rate Ratio, 95%CI: the 95% confidence interval limit of the IRR.
M AN U
SC
Table 2: Sensitivity analysis excluding patients used infliximab No. of thiopurine P-Y of exposed exposure events IR 4734 7778 18 2.3 All cohort 7517 17 2.3 IFX users excluded 4557
HR CI 4.2 2.5-6.8 4.0 2.4-6.7
p <0.0001 <0.0001
AC C
EP
TE D
Notes: Rate is per 1000 person-years, HR: Hazard Ratio of lymphoma while on thiopurine compared to unexposed, CI: the 95% confidence interval.
ACCEPTED MANUSCRIPT
Colon Small bowel Skin Rectum Disseminated Liver, Spleen Small bowel Disseminated LN right cervical Disseminated LN right axilla LN mediastinum LN left cervical Colon Brain Disseminated LN unspecified LN unspecified
AZA+IFX MP MP MP AZA MP AZA AZA AZA MP AZA AZA MP MP AZA MP MP MP
0.7 6.1 4.0 3.8 9.1 1.6 0.7 0.1 6.9 2.8 3.7 5.8 7.6 1.5 2.0 3.7 1.7 6.3
LN left axillary Rectum Bone marrow Bone marrow LN right axilla
MP+MTX MP MP MP AZA
1.5 3.0 2.7 1.4 5.2
EP
AC C
M AN U
SC
RI PT
Lymphoma Site
Therapy Duration, Thiopurines years
TE D
Table 1: Lymphoma cases among thiopurine users Colitis Duration, Age, years Lymphoma Type ID years Sex Race Coincidental Lymphomas 1 70 M White 2.5 Follicular B cell 2 61 M White 17.3 Enteropathic T cell 3 54 M White 4.5 T-cell 4 77 M Unknown 2.2 DLBCL 5 50 M White 20.8 DLBCL 6 63 M White 21.1 DLBCL 7 38 M White 2.1 Follicular B cell 8 76 M White 0.7 Mantle cell 9* 49 M White 24.4 NHL unspecified 10 81 M White 5.3 DLBCL 11 73 M White 8.5 Mantle cell 12 70 M White 25.2 DLBCL 13 41 M White 15.3 Follicular B cell 14 71 M White 38.5 DLBCL 15 42 M White 3.8 DLBCL 16 66 M White 20.5 DLBCL 17* 76 M White 13.4 NHL unspecified 18* 72 M Unknown 6.8 NHL unspecified Non-Coincidental Lymphomas 19(1) 54 M White 10.2 Mantle cell 20(2) 77 M Unknown 21.7 DLBCL 21(3) 49 M White 22.9 Marginal Zone 22(4) 67 M White 5.7 Follicular B cell 23(5) 44 M Non-white 8.5 Mantle cell
Notes: DLBCL, diffuse large B-cell lymphoma; NHL, non-Hodgkin lymphoma; LN, lymph node; IFX, infliximab; AZA, azathioprine; MP, 6-mercaptopurine; MTX, methotrexate. * Insufficient records found in the VA system, diagnosis and/or care was provided by non-VA provider (1) Patient had MP for 2 years then switched to MTX for 2 years then was diagnosed with lymphoma (2) Lymphoma was diagnosed 1.2 year after stopping MP (3) Lymphoma was diagnosed 6.4 years after stopping MP (4) Lymphoma was diagnosed 3.4 years after stopping MP
ACCEPTED MANUSCRIPT
AC C
EP
TE D
M AN U
SC
RI PT
(5) Lymphoma was diagnosed 2.6 years after stopping AZA
ACCEPTED MANUSCRIPT
SC
12% 52% 36% 7% 93% 76% 11% 12%
TE D
EP AC C
RI PT
%
13% 8% 5% 1% 2%
M AN U
Table 2: Demographic characteristics and medication use among the entire ulcerative colitis cohort. Number Demographics 4348 Age Group at induction <40 19276 40-65 13267 >65 2467 Sex Female 34424 Male 28170 Race White 4188 Non-white 4533 Unknown Immune Suppression Use 4734 Thiopurines Any 3126 Azathioprine 1963 6-mercaptopurine 251 Infliximab 671 Methotrexate 5-Aminosalicylic Acid Use 22664 Any 16484 Oral Mesalamine 7973 Rectal Mesalamine 6071 Sulfasalazine 2774 Balsalazide 300 Olsalazine
61% 45% 22% 16% 8% 1%
ACCEPTED MANUSCRIPT
M AN U
SC
RI PT
Table 3: Cox proportional hazards regression analysis with time to lymphoma as the dependent variable. No. No. % Person Rate HR p UCI LCI total events Years 4,348 7 0.16 24,879 0.28 1 (reference) Age <40 19,276 70 0.36 116,225 0.60 2.0 0.08 0.9 4.4 40-65 13,267 65 0.49 83,667 0.78 2.6 0.02 1.2 5.7 >65 2,467 3 0.12 14,114 0.21 1 (reference) Sex Female 34,424 139 0.40 210,656 0.66 2.5 0.12 0.8 7.9 Male 28,170 117 0.42 172,120 0.68 1 (reference) Race Caucasians 4,188 13 0.31 24,499 0.53 0.9 0.71 0.5 1.6 Non Caucasians 4,533 12 0.26 28,152 0.43 0.6 0.11 0.3 1.1 Unknown 36,297* 119 0.33 199,046 0.60 1 (reference) Thiopurine Unexposed 18 0.38 7,778 2.31 4.2 <0.0001 2.5 6.8 During Exposure 4,734 4,662** 5 0.11 17,950 0.28 0.5 0.17 0.2 1.3 After Stopping
AC C
EP
TE D
Notes: Rate is per 1000 person-years, HR: Hazard Ratio, UCI and LCI: Upper and Lower limits of the 95% confidence interval respectively. * Total number of patients who contributed person-years to the Unexposed period was 32,157 patients who never used thiopurine and 4,140 patients who later on used thiopurine during the follow-up period. ** Among thiopurine users (4,734 patients), 4,662 patients had follow-up after stopping it.
ACCEPTED MANUSCRIPT
Thiopurine users
year of thiopurine use 1st (0-1) 2nd (1-2) 3rd (2-3) 4th (3-4) ≥ 4 years OVERALL
*No.
Thiopurine non-users
P-Y
E
IR
year of follow-up 1st (0-1) 2nd (1-2) 3rd (2-3) 4th (3-4) ≥ 4 years OVERALL
SC
Total No. included in the analysis
RI PT
Table 4: Life table analysis comparing the exposure and the follow-up periods
*No.
P-Y
E
IR
Age, Sex, Race adjusted HR on thiopurine users compared to non-users during each follow-up interval (Cox regression analysis)
HR
LCI
UCI
p
NNH
AC C
EP
TE D
M AN U
36,891 4,734 3,233 3 0.93 32,157 30,951 27 0.87 1.2 0.4 4.0 0.77 17,344 31,834 2,305 1,856 3 1.62 29,529 28,095 13 0.46 3.8 1.1 13.5 0.04 864 28,116 1,478 1,212 2 1.65 26,638 25,324 14 0.55 2.9 0.7 13.0 0.16 911 25,029 991 799 4 5.01 24,038 22,896 8 0.35 16.0 4.8 53.8 <0.0001 215 22,400 605 678 6 8.85 21,795 84,160 57 0.68 14.4 5.9 35.0 <0.0001 122 36,891 4,734 7,778 18 2.31 32,157 191426 119 0.62 4.2 2.5 6.8 <0.0001 591 Notes: IR: Incidence Rate, Rate is per 1000 person-years, *Total number of patients who contributed person-years to the follow-up or exposure interval, P-Y: person-years of exposure among thiopurine users and follow-up among thiopurine non-users, E: (Events) lymphoma cases observed during each interval, HR: Hazard Ratio, NNH: number needed to harm.
ACCEPTED MANUSCRIPT
Table 5: Incidence rate of lymphoma after different cumulative exposure rates.
E
After Stopping Thiopurine Year of Median FU stopping *No. of after stopping thiopurine stoppers (years) 1st (0-1) 2,429 3.6 2nd (1-2) 827 3.1 3rd (2-3) 487 3.6 4th (3-4) 386 3.3 ≥ 4 years 605 3.7 OVERALL 4,734 3.5
RI PT
While Using Thiopurine Year of thiopurine use *No. P-Y 1st (0-1) 4,734 3,233 2nd (1-2) 2,305 1,856 3rd (2-3) 1,478 1,212 4th (3-4) 991 799 ≥ 4 years 605 678 OVERALL 4,734 7,778
IR
P-Y of FU after stopping E
IR
AC C
EP
TE D
M AN U
SC
3 0.93 9,472 0 0 3 1.62 3,021 2 0.66 2 1.65 1,881 1 0.53 4 5.01 1,403 1 0.71 6 8.85 2,173 1 0.46 18 2.31 17,950 5 0.28 Notes: IR: Incidence Rate, Rate is per 1000 person-years, *Total number of patients who contributed person-years to the exposure or follow-up interval, P-Y: person-years of exposure while on thiopurine and follow-up after stopping it, E: (Events) lymphoma cases observed during each interval.