Systematic review: hepatosplenic T-cell lymphoma on biologic therapy for inflammatory bowel disease, including data from the Food and Drug Administration Adverse Event Reporting System

Abstract

BACKGROUND:

Hepatosplenic T-cell lymphoma (HSTCL) is a rare, poorly treatable malignancy associated with therapy for inflammatory bowel disease (IBD). Current knowledge of HSTCL risk in IBD comes from an era of step-up therapy, before earlier use of biologics or combination therapy were advocated to achieve deep mucosal healing. HSTCL risk among newer biologic classes has also not been evaluated.

AIMS:

To systematically characterize the association of HSTCL with biologic therapy for IBD.

METHODS:

We conducted a literature search and query of the Food and Drug Administration Adverse Event Reporting System to summarize HSTCL cases among IBD patients with prior biologic exposure. Demographics and immunosuppression exposure were extracted. Patients were stratified by current regimen (combination therapy, biologic monotherapy, or no biologic), and biologic class (anti-TNF, anti-integrin, anti-interleukin 12/23).

RESULTS:

Sixty-two cases of HSTCL were identified from 2,486 abstracts and 181 FAERS reports. The median age of affected patients was 28 years (range 12-81), and 83.6% were male, 84.7% had Crohn’s disease. Five of 62 patients had no reported azathioprine/mercaptopurine exposure. Three patients within the cohort developed HSTCL after exposure to natalizumab, vedolizumab, or ustekinumab; all three also had anti-TNF and azathioprine/mercaptopurine exposure. Forty-three of 49 (87.8%) patients with known outcomes died with median survival of 5 months.

CONCLUSIONS:

Consistent with existing data, almost all identified HSTCL cases among IBD patients on biologic therapy had azathioprine/mercaptopurine exposure, and all cases on patients exposed to biologics had anti-TNF exposure. These data suggest initiating the patient-centered discussion before starting anti-TNF therapy or other biologics.

INTRODUCTION

Hepatosplenic T-cell lymphoma (HSTCL) is a rare peripheral lymphoma potentially associated with tumor necrosis factor (TNF) antibody therapy for the treatment of inflammatory bowel disease (IBD)(1,2). Despite aggressive treatments including allogeneic stem cell transplantation(3), five-year survival is limited to an estimated 7-16.2%(4,5). Commonly accepted risk factors for HSTCL among IBD patients on anti-TNF agents include male sex, age under 35 years, and current/prior azathioprine or mercaptopurine (AZA/MP) therapy(6). The potential association between anti-TNF therapy and HSTCL in IBD patients was incorporated into product labeling through a boxed warning for all anti-TNF agents(7), based on epidemiological findings which now date back almost ten years(8,9).

In contrast to years past, the contemporary goal of therapy in IBD is to achieve deep mucosal healing rather than symptomatic improvement alone. With this more lofty goal, it is not uncommon to recommend early introduction of biologic therapy for treatment-naïve patients(10) or combination therapy with biologic and immunomodulator(11,12). In the long-term, patients are being continued on biologic therapy for years given the risk of relapse with discontinuation(13). While United States Food and Drug Administration (FDA) labelling recommends that providers counsel their patients on the potential association between anti-TNF therapy and the rare but potentially devastating development of HSTCL, there are no published data to guide the patient-centered discussion needed for contemporary and generally more intensive treatment strategies. Two major questions which are not yet systematically addressed in the literature include: (1) are there cases of HSTCL among IBD patients on biologic therapy without any prior AZA/MP exposure, and (2) are new classes of anti-integrin and anti-interleukin 12/23 biologic agents associated with HSTCL?

The aim of this systematic review was to characterize demographic and treatment exposure characteristics for all known HSTCL cases among IBD patients exposed to biologic therapy, stratified by AZA/MP exposure and class of biologic therapy.

METHODS

This systematic review was conducted in accordance with the PRISMA statement. A literature search was conducted using PubMED and EMBASE (inception to May 11, 2017) to identify case reports and case series of patients with HSTCL published in an abstract form or as a full manuscript (see Supporting Information). We then conducted a search of the FDA Adverse Events Reporting System (FAERS) which catalogs all reported adverse events that FDA has received for FDA-approved therapies. The use of this reporting system is voluntary for patients and providers but is a legal requirement for drug manufacturers if they receive a report from the healthcare professional or consumer. However, the FAERS database cannot be used to calculate the incidence of an adverse event due to limitations such as voluntary reporting, duplicate reporting, and that it does not require a causal relationship between the product and the adverse event to be proven.

Inclusion criteria and data extraction

Our inclusion criteria were: (1) patients must have been diagnosed with IBD, (2) patients must have developed HSTCL, and (3) patients must have been treated with infliximab, adalimumab, certolizumab, golimumab, natalizumab, vedolizumab, or ustekinumab. Duplicate cases in the literature and FAERS search were counted only once.

Demographic and clinical data were extracted, including age, sex, location, disease type (CD or UC), distribution of disease and associated complications. Exposure to IBD treatments (AZA/MP, methotrexate, infliximab, adalimumab, certolizumab, golimumab, natalizumab, vedolizumab, and ustekinumab) was categorized by the number of years exposed to therapy, time since last dose, and therapeutic goal (induction vs maintenance of remission). Active infections identified at time of HSTCL diagnosis were extracted. Treatment and outcomes related to HSTCL were recorded. By design, FAERS includes all case reports included in the literature (i.e. FDA reviewers regularly upload literature case reports to FAERS). To reduce the risk of reporting duplicate cases of the same patient, case reports were combined if the age, sex, exposure, and outcome characteristics matched across multiple case reports. Exposure history was reviewed independently by two authors (ES and EC) with discrepancies addressed by consensus among authors.

Statistical analysis

HSTCL cases were stratified by the ongoing treatment regimen at time of diagnosis, as well as exposure to biologics and AZA/MP: combination therapy with AZA/MP, biologic monotherapy with prior AZA/MP exposure, biologic monotherapy without prior AZA/MP exposure, and previous exposure to biologics. Within-strata mean, median, range, and standard deviation were calculated for continuous variables. Inter-observer variability in determining therapeutic exposure was calculated using Cohen’s kappa. An R² statistic was calculated to evaluate whether AZA/MP exposure or number of years with IBD might modulate the biologic exposure required to develop HSTCL. Statistical analysis was performed using STATA (version 15, STATACorp, College Station, Texas).

RESULTS

From 2,486 literature abstracts and 181 FAERS records, we identified 62 unique HSTCL cases among IBD patients who received at least one dose of biologic therapy (Figure 1). Of the 62 cases, 41 cases occurred in patients actively receiving biologic therapy while the remaining 21 cases occurred among patients with prior biologic exposure. The level of agreement between reviewers in assessing exposure history was good (κ statistic = 0.88). Demographics and clinical characteristics at the time of HSTCL diagnosis are reported in Table 1, stratified by current biologic regimen as well as AZA/MP exposure. A summary of case reports is displayed in Supporting Information. Among patients currently taking biologic therapy at the time of HSTCL diagnosis, the ages ranged from 15 to 81 years. Within-strata median ages were 27.5 years old (combination biologic and AZA/MP), 24.5 years old (biologic monotherapy with prior AZA/MP exposure), and 48.0 years old (biologic monotherapy with no reported prior AZA/MP exposure). Of the 59 cases in which sex was reported, 10 (17.0%) of cases were female. 50 (84.7%) patients had CD and nine (15.3%) patients had UC, while it was not reported in the remaining three cases whether the patient had UC or CD. Disease extent was reported in 17 CD patients: 12 (70.6%) with small bowel involvement and 15 (88.2%) with colonic involvement. The disease extent was not reported among patients with UC. The average and median duration of IBD was 12.2 and 8.0 years, respectively, preceding HSCTL diagnosis (minimum = 1.8 years, maximum = 35 years between IBD and HSTCL diagnoses).

Figure 1:

Flowchart of study inclusion

Table 1:

Demographics and clinical characteristics of HSTCL cases stratified by current IBD therapy.

Characteristic	Combo therapy at time of HSTCL diagnosis	Biologic monotherapy at time of HSTCL diagnosis		No biologic therapy at time of HSTCL diagnosis			Current therapy unknown
		Previous AZA/MP exposure	No previous AZA/MP exposure	Current AZA/MP	Previous AZA/MP exposure	No reported AZA/MP exposure	At least some AZA/MP exposure	No reported AZA/MP exposure
Number of patients	27	8	3	9	7	No cases identified in dataset	6	2
Median age (range)	27.5 years (15.0 to 81.0)	24.5 years (24.0 to 68.0)	48.0 years (<18.0 to 74.0)	28.0 years (19.0 to 39.0)	30.0 years (16.0 to 52.0)		35.0 years (12.0 to 60.0)	35.0 years (reported data available for 1 case)
% Male (n/N)	86.2% (25/29)	75.0% (6/8)	33.3% (1/3)	100.0% (9/9)	100.0% (7/7)		83.3% (5/6)	50.0% (1/2)
Disease
UC	11.1% (3/27)	12.5% (1/8)		22.2% (2/9)	28.6% (2/7)		16.7% (1/6)	Not specified
CD	88.9% (24/27)	87.5% (7/8)	100% (3/3)	77.8% (7/9)	71.4% (5/7)		83.3% (5/6)
Median disease duration (range)	8.0 years (3.0 to 34.0)	8.0 years (1.8 to 12.0)	Not reported	7.0 years	20.0 years (7.0 to 35.0)		22.0 years (reported data available for 1 case)	Not specified
Penetrating disease or prior surgery	4 CD cases (16.7% of 24 CD cases)	3 CD cases (42.9% of 7 CD cases)	Not reported	2 CD cases (28.5% of 7 CD cases)	3 CD cases (60.0% of 5 CD cases)		1 CD case (20.0% of 5 CD cases)	Not specified
Biologic exposure Median	3.8 years	1.0 year	2 years (reported data available for 1 case)	<0.1 years	1.9 years		4.2 years (reported data available for 2 cases)	Not specified
AZA/MP or methotrexate exposureMedian Minimum Time since last dose	5 years 1 year	4.0 years 3 months 1.6-11 years		5.0 years 1.7 years	6.5 years 1 year		4 years (reported data available for 1 case)	Not specified

HSTCL among patients exposed to AZA/MP or methotrexate

Fifty-seven of 62 cases (91.9%) reported at least some AZA/MP exposure. The median exposure to AZA/MP was 5.3 years prior to HSTCL diagnosis. Five cases had fewer than two years of AZA/MP exposure prior to diagnosis of HSTCL (FAERS ID# 6132748, 6950655, 7138070, 7410990, and 7114026), but only two of these patients were on combination therapy when HSTCL was diagnosed. All other cases had at least two years of exposure to AZA/MP prior to diagnosis of HSTCL. The overall duration of AZA/MP exposure prior to HSTCL diagnosis appeared unrelated to duration of IBD (R²=0.02) (Figure 2).

Figure 2:

Azathioprine/mercaptopurine and biologic exposure appear unrelated among HSTCL cases.

Only one patient had prior exposure to methotrexate in the dataset, with a total historical exposure of two years, but this patient had also been exposed to AZA/MP, anti-TNF-α, anti-integrin, and anti-interleukin 12/23 agents (FAERS ID# 7410990).

Cases of HSTCL without known exposure to AZA/MP

Five cases (8.0% of 62 total HSTCL cases) occurred among patients with at least some biologic exposure but with no reported exposure to AZA/MP (FAERS ID# 12537122, 9628730, 6950655, 5958596 and 12531313).

Of the five total cases, three patients were receiving biologic monotherapy at the time of HSTCL diagnosis and had no known AZA/MP exposure (FAERS ID# 6950655, 5958596 and 12531313). These patients were 22, 35, and 74 years of age at the time of HSTCL diagnosis, and two patients were female. All three patients had CD and were managed with infliximab monotherapy: one patient was exposed to infliximab for two years, with maintenance of remission as the intent of ongoing infliximab therapy. The length of infliximab exposure was unknown for the other two patients. There was no known exposure to other biologics among these patients.

The remaining two cases reported exposure to infliximab with no known exposure to AZA/MP, but whether these patients remained on infliximab monotherapy or were on no therapy at the time of HSTCL diagnosis was unclear in the dataset (FAERS ID# 12537122 and 9628730).

Cases of HSTCL stratified by exposure to patient classes of biologics

All 62 cases reported exposure to anti-TNF-α agents. Of these cases, 57 cases were associated with prior exposure to infliximab with an average exposure of 3.5 years. Fourteen cases had prior exposure to adalimumab (average exposure 1.4 years), although 11 (78.6%) of these cases had known previous exposure to infliximab. One patient had exposure to certolizumab, although the exposure to certolizumab and ongoing therapeutic regimen at the time of HSTCL diagnosis was unknown. There were no known cases of HSTCL exposed to golimumab.

One patient developed HSTCL having been previously exposed to natalizumab. This was a 52-year old male with CD with a history of prior bowel resection having completed induction therapy with natalizumab one year prior. This patient had 10 years of exposure to AZA/MP monotherapy. After this patient experienced disease flare on AZA/MP, the patient completed induction with infliximab but was changed to adalimumab on which remission was achieved for two years. The patient then experienced a disease flare and received natalizumab induction, however the patient had been completely off of immunosuppressive therapy for one year following natalizumab until diagnosis of HSTCL.

There was also one case of HSTCL in a 51-year male undergoing vedolizumab induction with continued combination therapy using AZA/MP (FAERS ID# 13213897). This patient had previously been maintained on adalimumab for 10 years. There was a 20-year exposure history to AZA/MP.

One case of HSTCL occurred in a 68-year old female with ileal CD diagnosed approximately two years earlier, having just completed four months of ustekinumab induction monotherapy (FAERS ID# 7410990). This patient was initiated on azathioprine at disease diagnosis but was changed to methotrexate after four months, which continued until initiation of ustekinumab. Regarding prior biologic exposure, this patient had received four months of infliximab followed by one year of adalimumab, both in combination with immunomodulators.

Biologic exposure and disease duration did not appear to be correlated among HSTCL cases (R²=0.02) (Figure 3).

Figure 3:

Biologic exposure and disease duration appear unrelated among HSTCL cases.

Association of HSTCL diagnosis with ongoing infection

Four cases of HSTCL were associated with active Epstein Barr Virus (EBV) infection. Two cases were diagnosed with a positive polymerase chain reaction (PCR) assay, one case with a positive EBV IgM antibody titer, and one case with seroconversion of EBV IgG antibody before and after diagnosis of HSTCL. No other active infections were reported in the remaining cases identified. Three cases of HSTCL associated with EBV infection occurred among patients on combination therapy (one case with adalimumab and two with infliximab). The fourth case was being maintained on infliximab monotherapy for one year, having taken AZA/MP for almost five years previously.

Treatment and outcomes

At least 43 of 49 patients with known outcome died after HSTCL diagnosis with a median survival of five months. The longest survival was 5.7 years in a 50-year old male with CD having had exposure to both infliximab and AZA/MP (length of exposure and ongoing therapy unknown).

DISCUSSION

We identified 62 cases of HSTCL occurring among patients with IBD exposed to biologic therapy. All patients had been exposed to anti-TNF agents. While 57 cases were exposed to AZA/MP previously, five cases were identified without any reported AZA/MP exposure. There was one case each of HSTCL with prior exposure to natalizumab, vedolizumab, or ustekinumab; however, these patients had at least some prior exposure to both anti-TNF and AZA/MP therapy.

Since the approval of infliximab for the treatment of moderate-to-severe CD in 1998 and UC in 2005, HSTCL appears to remain a rare malignancy based on our analysis. This is consistent with an estimated absolute risk reported previously of fewer than one in 21,947 patients on anti-TNF therapy for IBD (<5/100,000 patients)(6). This is despite more stringent treatment targets requiring deeper immunosuppression for longer periods of time. While HSTCL remains a devastating malignancy when it occurs, this likely balances against a risk of relapse of 44% for CD and 38% for UC following discontinuation of anti-TNF therapy(14). The benefits of long-term biologic therapy may outweigh the very low risk of a potentially devastating condition for most patients; however, current guidelines and product labeling ensure that patients receiving anti-TNF therapy are adequately informed of the potential risk of HSTCL in order to facilitate this benefit-risk discussion with their healthcare providers.

Our study demonstrated that HSTCL occurred most often among younger males exposed to AZA/MP, consistent with current knowledge. Specifically, almost all known HSTCL cases to date occurred in patients with at least two years of exposure to AZA/MP. This finding is important as we begin to understand when and how to appropriately step-down from combination therapy(15). The SPARE trial, which is currently enrolling(16), is randomizing patients on combination therapy to stop either their anti-TNF biologic or AZA/MP. If this trial shows no difference in rates of clinical remission, steroid use, or hospitalizations, and shows that risk of HSTCL likely increases after two years of exposure, these data may suggest that long-term monotherapy with a biologic may be a preferred treatment compared to long-term AZA/MP monotherapy, especially in young males. While we identified five cases of HSTCL with at least some biologic exposure but no reported AZA/MP exposure, the absolute risk of HSTCL in 20 years of marketed biologic therapy appears small for patients on anti-TNF therapy agents without AZA/MP exposure (i.e. less than < 5/100,000). Importantly, there were 10 cases among women (including two female cases occurring without known AZA/MP exposure) and six cases among patients at least 60 years of age, suggesting that HSTCL could occur in these populations.

Four patients on anti-TNF therapy had an active EBV infection at the time of HSTCL diagnosis. Anti-TNF therapy is known to cause lytic activation of EBV(17) which may support a mechanistic hypothesis for the development of HSTCL; however, whether EBV infection is necessary for the development of HSTCL in this patient population is unknown. It is also unknown whether differences in baseline serologic positivity might affect the risk of developing HSTCL. Current practice guideline recommendations do not suggest screening for EBV status before initiating biologic therapy.

The primary strength of our study is an up-to-date FAERS and literature search which captures all known HSTCL cases during the current era of the earlier use of biologic therapy, combination therapy, and therapy with anti-integrin or anti-interleukin 12/23 agents. Two well-conducted systematic reviews also addressed the association between T cell lymphomas and anti-TNF therapy for IBD(6,18); however, these studies included a FAERS search which now dates back eight years, before regulatory approvals of vedolizumab and ustekinumab. This study benefitted from the active efforts of the FDA in maintaining the FAERS database, combined with a legal requirement of drug sponsors to report known HSTCL cases, both of which can mitigate under-reporting.

This study has several limitations. The FAERS database cannot be used to calculate the incidence of an adverse event due to limitations such as voluntary reporting, duplicate reporting, and that it does not require a causal relationship between the product and the adverse event to be proven. Due to these limitations, we did not assess AZA/MP monotherapy as a comparative control. Specifically, variation in the number of HSTCL cases among the anti-TNF therapies may relate to differences in time on the market and frequency of use (i.e. differences in total patient-years of exposure among agents), therefore our study cannot make inferences toward comparative risks of HSTCL among the various anti-TNF agents. Moreover, underreporting of adverse events in FAERS limits quantification of risk. Potential duplicate cases were adjudicated based on comparison of content similarities among case reports and cross-referencing between FAERS and literature; however, FAERS often includes duplicate reports containing missing fields(19). While it may not be possible to ensure that each case in FAERS is completely unique compared to cases reported in the literature, the cases presented herein have unique demographic, exposure, and outcomes suggesting that they are indeed unique cases.

Consistent with existing labeling, the potential association of HSTCL with IBD biologic therapy is mostly among patients with prior AZA/MP exposure. Young age and male sex continue to appear as risk factors which is consistent with current knowledge. Most patients diagnosed with HSTCL report at least two years of AZA/MP exposure. There is one reported case each for natalizumab, vedolizumab, and ustekinumab exposure; however, all of these patients were previously exposed to both anti-TNF and AZA/MP. It is important to acknowledge that the small number of cases seen with these therapies (compared to anti-TNFs) may be due to less time on market and less frequent use compared to anti-TNFs and AZA/MP.

In conclusion, all patients being started on anti-TNF therapy should continue to be counseled on the risk of HSTCL, and the benefits of long-term immunosuppressive regimens should be considered on an individual patient basis. Careful attention regarding benefits and risks of biologic therapy should continue to be given, particularly to the young male population. This devastating malignancy continues to appear rare based on over 20 years of data in this population despite increasingly intensive management in contemporary practice of managing IBD.