Ustekinumab and Vedolizumab Are Not Associated With Subsequent Cancer in IBD Patients with Prior Malignancy

Abstract

Background

There is little data regarding the risk of new or recurrent cancer in patients with inflammatory bowel disease (IBD) and a prior history of cancer who are exposed to ustekinumab or vedolizumab. We assessed the risk of subsequent cancer in patients exposed to these agents.

Methods

We performed a retrospective cohort study of patients with IBD and a history of cancer at an academic medical center between January 2013 and December 2020. We collected data on demographics, IBD and cancer disease characteristics, and drug exposures. The primary exposure was immunosuppressive therapy after diagnosis of cancer. The primary outcome was interval development of new or recurrent cancer.

Results

Of 390 patients with IBD and a previous history of cancer, 37 were exposed to vedolizumab, 14 ustekinumab, 41 antitumor necrosis factor (anti-TNF), and 31 immunomodulator; and 267 were not exposed to immunosuppression following cancer diagnosis. During a total median follow-up time of 52 months, 81 (20%) patients developed subsequent cancer: 6 (16%) were exposed to vedolizumab, 2 (14%) to ustekinumab, 3 (10%) to immunomodulators, 12 (29%) to anti-TNF, and 56 (21%) with no immunosuppression (P = .41). In a multivariable Cox model adjusting for age, IBD subtype, smoking, cancer recurrence risk, and cancer stage, there was no increase in subsequent cancer with vedolizumab (adjusted hazard ratio, 1.36; 95% CI, 0.27-7.01) or ustekinumab (adjusted hazard ratio, 0.96; 95% CI, 0.17-5.41). Patients with more than 1 biologic exposure also did not have an increased risk of subsequent cancer.

Conclusions

Exposure to ustekinumab or vedolizumab in patients with IBD and a prior history of cancer does not appear to be associated with an increased risk of subsequent new or recurrent cancer.

Introduction

Immunosuppressive therapy is the mainstay of treatment for patients with moderate to severe disease inflammatory bowel disease (IBD).^1,2 Recently, new therapeutics with novel mechanisms of action have further advanced the ability to effectively treat IBD. However, immunosuppression is associated with treatment-related adverse events, particularly the risk of malignancy, which is of considerable concern to both physicians and patients.³

Much of our current understanding regarding the risk of malignancy associated with immunosuppressive therapy in IBD comes from studies of patients without prior history of cancer or from registry trials, in which patients with a prior cancer within 5 years are generally excluded.⁴ Furthermore, there is considerable literature describing the risk of lymphoma and skin cancer in patients exposed to thiopurines or antitumor necrosis factor (anti-TNF) therapies, which can often lead providers to suspend or withhold immunosuppression in patients with IBD who have recently been diagnosed with cancer.^5–11

More recently, reassuring data have shown that the risk of new or recurrent cancer is not increased in IBD patients with prior cancer who undergo treatment with immunomodulators or anti-TNFs.^12–14 A recent retrospective study reported that vedolizumab (VDZ) is not associated with subsequent cancer in patients with prior malignancy, but data do not exist yet regarding ustekinumab (UST), an anti-interleukin 12/23 agent.¹⁵ Although registry data for these agents reported overall low incidence rates of subsequent cancer, patients who were older or had a prior cancer were excluded from these studies.^16–19

The objective of this study was to assess whether patients with IBD and a history of cancer who were exposed to VDZ or UST had an increased risk of developing subsequent cancer compared with patients who received immunomodulators, anti-TNF, or no immunosuppression.

Materials and Methods

Study Population

We reviewed the medical records of all adult patients with IBD and cancer who received care at an academic medical center between January 2013 and December 2020. Patients were included in the study if they had (1) a diagnosis of IBD based on diagnostic criteria of clinical symptoms with supporting endoscopy, pathology, and radiologic findings; and (2) a confirmed diagnosis of previous cancer in remission based on pathology or oncology documentation with cancer occurring after diagnosis of IBD. The study entry point was date of diagnosis of the initial cancer. Patients were categorized into groups based on no exposure to immunosuppression or exposure to 1 immunosuppressive agent (immunomodulator [6-mercaptopurine or azathioprine], anti-TNF, VDZ, or UST) at any point after their prior cancer diagnosis. Patients with exposures to multiple agents were excluded from the primary analysis and were analyzed in a separate subcohort. This study was approved by the institutional review board at NYU Langone Health.

We recorded demographic information, age at diagnosis of IBD, type of IBD, date of diagnosis of prior cancer, cancer type and stage, cancer recurrence risk, cancer treatment and history, and duration of IBD medication exposure before and after cancer diagnosis. For patients with subsequent cancer, date of diagnosis, type, stage, and treatment of cancer were collected. Cancers were categorized into 4 types: gastrointestinal, hematologic (any leukemia or lymphoma), dermatologic (melanoma or nonmelanoma skin cancer), and all other solid organ tumors. Cancer risk was assigned based on Penn classification for recurrence risk of preexisting malignancies, in which a low, intermediate, and high risk are associated with recurrence rates of 0% to 10%, 11% to 25%, and >25%, respectively.²⁰

The primary outcome was development of subsequent cancer, classified as either new or recurrent. The time to subsequent cancer was calculated from the date of initial cancer diagnosis to the date of the first new or recurrent malignancy or date of last follow-up. Using the date of initial cancer diagnosis permitted the most uniform calculation of time to subsequent cancer. If the exact date or month of initial cancer diagnosis was not identifiable, the date of diagnosis was approximated using midpoint of the month or year. For patients with multiple new or recurrent cancers, only the interval between initial cancer diagnosis and the first subsequent malignancy was used for analysis.

Statistical Analysis

Continuous variables were reported as means with standard deviation (or medians with interquartile range) and were compared using the Student t test (or Wilcoxon rank-sum test). Categorical values were summarized as percentages and were compared using the χ² test. Cox proportional hazard models were used to assess the risk of subsequent cancer stratified by study arms and adjusted for IBD type, age at index cancer, smoking status, cancer recurrence risk, and prior cancer stage. Time to cancer was compared between study arms with Kaplan-Meier survival curves using log-rank testing. All statistical analyses were considered significant at a 2-sided P value of less than .05. A sensitivity analysis was performed including all patients with multiple biologic exposures who were excluded from the primary analysis. Statistical analysis was performed using IBM SPSS 25 (IBM, Armonk, NY).

Results

We identified a total of 390 patients with IBD and a prior history of cancer. Of these, 31 patients were exposed to immunomodulators at any timepoint after their initial cancer, 41 to anti-TNF, 37 to VDZ, and 14 to UST; 267 patients were not exposed to any immunosuppression. Patients who received no immunosuppression were more likely to have ulcerative colitis (P < .01) compared with all other study arms except VDZ and had a numerically higher rate of current smokers (P = .11; Table 1). No differences were seen in age at IBD diagnosis or cancer across all medication exposure groups (P = .95).

Table 1.

Demographic characteristics at baseline.

	No IS	IMM	TNF	VDZ	UST	P
No. patients (% of total)	267 (68.5)	31 (7.9)	41 (10.5)	37 (9.5)	14 (3.6)
IBD type
Crohn’s disease	121 (45.3)	18 (58.1)	23 (56.1)	10 (27.0)	10 (71.4)
Ulcerative colitis	143 (53.6)	12 (38.7)	17 (41.5)	26 (70.3)	2 (14.3)
IBD-U	3 (1.1)	1 (3.2)	1 (2.4)	1 (2.7)	2 (14.3)	<0.01
Sex
Male	118 (44.2)	14 (45.2)	24 (58.5)	21 (56.8)	8 (57.1)
Female	149 (55.8)	17 (54.8)	17 (41.5)	16 (43.2)	6 (42.9)	0.82
Race
White	248 (92.9)	30 (96.8)	38 (92.7)	35 (94.6)	12 (85.7)
Black	6 (2.2)	0 (0)	0 (0)	1 (2.7)	1 (7.1)
Asian	8 (3.0)	0 (0)	0 (0)	0 (0)	1 (7.1)
Other	5 (1.9)	d	3 (7.3)	1 (2.7)	0 (0)	0.41
Ethnicity
Hispanic	18 (6.9)	0 (0)	4 (10.0)	1 (2.8)	0 (0)
Non-Hispanic	241 (93.8)	30 (100)	36 (90.0)	35 (97.2)	12 (100)	0.31
Age at IBD diagnosis (mean ± standard deviation) (y)	44.9 ± 17.6	44.2 ± 21.3	43.5 ± 20.3	42.5 ± 18.6	46.8 ± 18.4	0.95
Smoking status
Never	148 (55.4)	18 (58.1)	26 (63.4)	16 (43.2)	7 (50.0)
Former	109 (40.8)	13 (41.9)	11 (26.8)	20 (54.1)	5 (35.7)
Current	10 (30.7)	0 (0)	4 (9.8)	1 (2.7)	2 (14.3)	0.11

Proportions reported as no. (%). Continuous variables reported as mean ± SD, or median (interquartile range) when data not distributed normally.

6-mercaptopurine, azathioprine

infliximab, adalimumab or golimumab.

In total, prior cancers included 50 gastrointestinal, 80 dermatologic, 16 hematologic, and 244 solid tumors (Table 2). Patients who did not receive immunosuppression were less likely to have had immunosuppression exposure prior to malignancy (P < .001) and more likely to have a prior solid malignancy (P < .01). Patients with no immunosuppression were more likely to have a higher prior cancer risk type than patients receiving UST but not VDZ (P < .01) and were numerically more likely to have more advanced stage primary cancers than patients with VDZ or UST exposure (P = .65).

Table 2.

Disease characteristics at baseline.

	No IS	IMM	TNF	VDZ	UST	P
No. patients (% of total)	267 (68.5)	31 (7.9)	41 (10.5)	37 (9.5)	14 (3.6)
IBD treatment before cancer
Systemic steroids	29 (11%)	9 (29%)	5 (12%)	11 (30%)	3 (21%)	<0.01
Immunomodulatora	9 (3%)	19 (61%)	3 (7%)	9 (24%)	4 (29%)	<0.001
Anti-TNFb	8 (3%)	4 (13%)	14 (34%)	7 (19%)	4 (27%)	<0.001
Combination therapy	3 (1%)	4 (13%)	0 (0%)	4 (11%)	4 (11%)	<0.001
VDZ	0 (0%)	2 (7%)	0 (0%)	6 (16%)	3 (21%)	<0.001
UST	0 (0%)	0 (0%)	0 (0%)	0 (0%)	3 (21)	<0.001
Antibiotics	6 (2%)	1 (3%)	0 (0%)	2 (5%)	2 (14%)	0.08
Surgery	38 (14%)	5 (17%)	3 (7%)	3 (8%)	1 (7%)	0.66
Prior cancer category
Gastrointestinal	42 (16%)	4 (13%)	1(2%)	1 (3%)	2 (14%)
Hematologic	9 (3%)	1 (3%)	0 (0%)	3 (8%)	3 (21%)
Dermatologic	46 (17%)	11 (36%)	14 (34%)	8 (22%)	1 (7%)
Solid	170 (64%)	15 (48%)	26 (63%)	25 (68%)	8 (57%)	<0.01
Prior cancer recurrence risk20
High	32 (12%)	5 (16%)	10 (24%)	6 (16%)	1 (7%)
Intermediate	114 (43%)	12 (39%)	13 (32%)	11 (30%)	4 (29%)
Low	25 (9%)	1 (3%)	3 (7%)	3 (8%)	6 (43%)
Undetermined	96 (36%)	13 (42%)	15 (37%)	17 (46%)	3 (21%)	<0.01
Age at prior malignancy diagnosis (mean ± standard deviation) (y)	57.3 ± 15.7	56.9 ± 16.2	51.5 ± 17.7	50.1 ± 18.6	53.7 ± 15.7	0.95
Stage of prior cancer
1	22/61 (36%)	2/4 (50%)	7/16 (44%)	6/13 (46%)	5/9 (56%)
2	9/61 (15%)	1/4 (25%)	4/16 (25%)	3/13 (23%)	1/9 (11%)
3	6/61 (10%)	0 (0%)	0 (0%)	2/13 (15%)	2/9 (22%)
4	24/61 (39%)	1/4 (25%)	5/16 (31%)	2/13 (15%)	1/9 (11%)	0.65
Prior cancer treatment
Radiation therapy	76 (29%)	7 (23%)	9 (22%)	12 (32%)	2 (14%)	0.59
Hormonal therapy	23 (9%)	1 (3%)	3 (7%)	2 (5%)	1 (7%)	0.83
Chemotherapy	93 (35%)	3 (10%)	8 (20%)	11 (30%)	5 (36%)	0.03
Immunotherapy	9 (3%)	0 (0%)	4 (10%)	1 (3%)	0 (0%)	0.18
Surgery	216 (81%)	26 (84%)	35 (85%)	29 (78%)	9 (64%)	0.50

Proportions reported as no. (%). Continuous variables reported as mean ± SD or median (interquartile range) when data not distributed normally.

6-mercaptopurine, azathioprine

infliximab, adalimumab, certolizumab, or golimumab.

The median time to initiation of new immunosuppression after initial cancer diagnosis was 5 months (range, 0-312) for immunomodulators, 30 months (range, 0-336) for anti-TNF, 29 months (range, 0-457) for VDZ, and 81 months (range, 0-133) for UST (P = .36). No statistically significant differences were seen in rates of therapy continuation through the prior cancer diagnosis: 44% (22 of 50) of patients on immunomodulators, 46% (19 of 41) on anti-TNF, 67% (8 of 12) on VDZ, and 100% (3 of 3) of UST (P = .64). The median duration of exposure to immunomodulators, anti-TNF, VDZ, and UST was 43 months (range, 1-292), 61 months (range, 4-402), 45 months (range, 4-457), and 48 months (5-140), respectively (P = .69).

Subsequent Cancer

During the total duration of follow-up, 79 patients (20%) developed subsequent cancer (Table 3); 61 (15.6%) had a new cancer, and 18 (4.6%) had a recurrent cancer. The subsequent cancer rate per 100 person-years for VDZ and UST was 1.9 and 3.0, respectively, compared with 1.6, 3.8, and 2.7 for patients in the immunomodulator, anti-TNF, and no immunosuppression groups, respectively. Time to subsequent cancer did not differ between groups (Figure 1; log-rank, P = .50). In a multivariable Cox model adjusting for IBD type, age at index cancer, smoking status, Penn classification of recurrence risk, cancer stage, time from cancer to treatment initiation and duration of exposure (Table 4), VDZ (adjusted hazard ratio [aHR], 1.36; 95% CI, 0.27-7.01), and UST (aHR, 0.96; 95% CI, 0.17-5.41) were not associated with subsequent cancer development.

Table 3.

Subsequent cancer outcomes.

	No IS	IMM	TNF	VDZ	UST	P
No. patients (% of total)	267 (66.7)	31 (7.7)	41 (10.2)	37 (9.2)	14 (3.5)
Median duration of follow-up (mo)	54	43	61	45	42	0.56
Subsequent cancer	56 (21%)	3 (10%)	12 (29%)	6 (16%)	2 (14%)	0.29
New	45 (80%)	3 (100%)	9 (75%)	3 (50%)	1 (50%)
Recurrent	11 (20%)	0 (0%)	3 (25%)	3 (50%)	1 (50%)	0.06
Subsequent cancer category
Gastrointestinal	8 (14%)	0 (0%)	1 (8%)	0 (0%)	0 (0%)
Hematologic	1 (2%)	1 (33%)	0 (0%)	0 (0%)	0 (0%)
Dermatologic	16 (28%)	2 (67%)	5 (33%)	3 (50%)	1 (50%)
Solid	32 (56%)	0 (0%)	6 (50%)	3 (50%)	1 (50%)	0.12
Subsequent cancer rate per 100-person years (no. of person-years)	2.67 (2102)	1.61 (186)	3.83 (313)	1.94 (309)	2.98 (67)

Figure 1.

Cancer-free survival based on drug exposure.

Table 4.

Cox proportional hazards model for new or recurrent cancer using person-years of follow-up since cancer.

Group	No. of new or recurrent	Person-years of follow-up since cancer	Adjusted HR (95% CI) for incident cancera
No IS	56 (45N/11R)	2102	Reference
Immunomodulator	3 (3N)	187	0.62 (0.26-2.18)
Anti-TNF	12 (9N/3R)	313	0.70 (0.10-4.74)
VDZ	6 (3N/3R)	309	1.36 (0.27-7.01)
UST	2 (1N/1R)	67	0.96 (0.17-5.41)

Model adjusted for age at index cancer, IBD subtype, smoking history, cancer recurrence risk, cancer stage, time from cancer to treatment initiation and duration of exposure. .

A secondary analysis of the subcohort of patients with sequential exposure to more than 1 biologic after diagnosis of cancer was performed. There was no increase in risk of subsequent cancer when comparing patients with any exposure to a given biologic agent or multiple exposures to those did not receive immunosuppression (Supplementary Tables 1 and 2).

Sensitivity Analyses

There were no significant differences in category or subtype of subsequent cancer among the exposure groups (Table 3, Supplementary Table 3; P = .12). After excluding nonmelanoma skin cancer, there was no increased risk of subsequent cancer between medication exposure groups (P = .59), and the adjusted risk for subsequent cancer remained similar for VDZ (aHR, 1.04; 95% CI, 0.16-2.67) and UST (aHR, 0.57; 95% CI, 0.03-11.50) compared with patients not on immunosuppression. Similarly, a sensitivity analysis excluding gastrointestinal (GI) malignancies did not show any differences in subsequent cancer risk (Supplementary Table 4). Restricting the analysis to those with biologics initiated within 5 years of a cancer diagnosis did not show any significant differences in subsequent cancer risk (VDZ: aHR, 0.69; 95% CI, 0.22-2.14; UST: aHR, 1.31; 95% CI, 0.23-7.53).

Discussion

Novel mechanisms of action have expanded the therapeutic options for patients with IBD. However, the aging IBD population has increased the likelihood that patients being considered for long-term immunosuppression will have had a history of cancer.²¹ Given the association of older immunosuppressive agents such as immunomodulators and anti-TNFs with malignancy, there is considerable interest in the risk of subsequent cancer associated with newer biologic agents in IBD patients with a history of cancer.^22–24

Recent studies investigating the risk of cancer recurrence have provided reassuring data. The Cancers Et Surrisque Associéaux Maladies Inflammatoires Intestinales En France (CESAME) study demonstrated that prior exposure to thiopurines was not associated with an increase in the risk of new or recurrent cancer in patients with prior cancer.¹² In a multicenter retrospective study of 333 patients, Axelrad et al showed that anti-TNF or immunomodulator exposure did not increase the risk of new or recurrent cancer, and a meta-analysis of immunosuppressant use in various immune-mediated diseases reported similar findings.^13,14

Current data on the risk of malignancy with VDZ and UST have generally been informed by the registry trials for these agents, in which rates of de novo cancer were generally low. In a pooled analysis of 2830 patients enrolled in the GEMINI clinical trials, only 18 (0.6%) patients with VDZ exposure developed malignancy over a total of 4811 person-years of follow-up.²⁵ In the 3-year follow-up study of the IM-UNITI registry trial for UST in Crohn’s disease, the incidence rate of malignancy was 1.9% (11 of 567). Similarly in the UNIFI study of UST for ulcerative colitis, only 0.8% (7 of 825) of patients in the study arm developed subsequent cancer.^18,26 More recently however, real-world data have begun to emerge regarding these agents. Vedamurthy et al reported that in 96 patients with a prior history of cancer, VDZ exposure was not associated with an increased risk of subsequent cancer (HR, 1.03; 95% CI, 0.65-1.64).¹⁵ In the psoriasis longitudinal assessment and registry (PSOLAR) registry, UST exposure in patients with psoriasis was not associated with an increased risk of malignancy when compared with exposure to infliximab, adalimumab, or nonbiologic therapies.²⁷

In this cohort of patients with IBD and a history of cancer, exposure to VDZ or UST did not increase the risk of new or recurrent cancer when compared with exposure to immunomodulators, anti-TNF, or no immunosuppression. This is the largest study to date to assess UST along with anti-TNF and VDZ; it showed that none of these biologic agents are associated with an increased risk of subsequent cancer.^13,15 Our findings suggest that monotherapy with anti-TNF, VDZ, or UST in IBD patients with a history of malignancy is safe and does not increase the risk of subsequent new or recurrent cancer. Furthermore, the duration of biologic exposure after a diagnosis of cancer was not associated with risk of subsequent cancer.

In addition to our primary cohort, which was limited to patients with exposure to a single biologic after cancer diagnosis, we also collected data on patients with exposure to more than 1 biologic after their initial cancer. We found that the incidence of new or recurrent cancer was not increased in this subgroup, both when assessing either any exposure to a given biologic or exposure to multiple sequential biologics. Although this cohort had a small sample size (total N = 39), these data are reassuring given the increasing likelihood that patients with treatment-refractory disease will be exposed to sequential biologic monotherapy or even combination biologic therapies as the number of therapeutic agents and classes for IBD treatment expands over time.²⁸

Our findings provide support for the clinical practice of continuing or starting new immunosuppressive IBD therapies in patients who are in remission from a prior malignancy. The current European Crohn’s and Colitis Organisation (ECCO) guidelines suggest a 2-5 year interval period for resuming biologics, but it is important to note that these recommendations are based on experience with cancer recurrence in organ transplant recipients.¹¹ In our study, when limiting analysis to patients with exposure to biologic within 5 years of prior cancer diagnosis, the risk of subsequent cancer was not increased. The median time to initiation was 24 months for VDZ and 5 months for UST, indicating that it may be safe to resume or initiate these agents earlier than 5 years after an initial cancer diagnosis.

There are several limitations to this study that are inherent to a retrospective study design. Limitations in the available data may not have fully accounted for unmeasured confounders that may have influenced cancer development and decision-making in IBD management, such as dose and duration of biologic exposures prior to the initial cancer. The stage and type of tumor may confound our results by introducing selection bias, as those with higher stage cancers may have been less likely to receive immunosuppression after cancer; and the risk of subsequent cancer may vary according to specific tumor biology. Our approach to measuring cancer-free survival based on date of diagnosis may introduce immortal time bias by including the time period during which patients had active cancer prior to achieving remission. The Penn classification is less accurate in defining recurrence risk compared with tumor characteristics such as molecular subtype, but these data were not readily available. Lastly, the relatively small sample size of patients with exposure to newer biologics may have underpowered our ability to detect small differences in subsequent cancers. Larger studies are needed to confirm our findings.

As the long-term safety profiles of VDZ and UST continue to be elucidated, our findings provide reassurance that exposure to these agents after a diagnosis of cancer is not associated with an increased risk of subsequent cancer compared with patients who receive anti-TNFs, immunomodulators, or no immunosuppression. These novel therapeutic classes, which have the presumed benefit of more selective targeting of immune pathways, appear to be safe in patients with a prior history of cancer. Further prospective studies, such as the ongoing SAPPHIRE registry, are required to confirm these data.²⁹

Author Contribution

Study concept and design: S.J.H., S.C., and J.E.A.

Data acquisition: S.J.H., C.Z., J.P., A.P., and M.V.

Statistical analysis: S.J.H.

Interpretation of data: S.J.H., D.P.H., S.C., and J.E.A.

Drafting of article: S.J.H.

Critical revision of article: S.J.H., C.Z., D.P.H., S.C., and J.E.A.

Funding

J.E.A. is funded by the Crohn’s and Colitis Foundation, the Judith & Stewart Colton Center for Autoimmunity, and the National Institute of Health-National Institute of Diabetes and Digestive and Kidney Diseases K23DK124570.

Conflicts of Interest

S.J.H., C.Z., J.P., A.P., and M.V. have no disclosures. S.C. has served as a consultant for Shire, Pfizer, and Oshi Health. J.E.A. has received research grants from BioFire Diagnostics; served as a consultant or advisory board for BioFire Diagnostics, Janssen, and Abbvie; and holds U.S. patent 2012/0052124A1. D.P.H. has received research grants from Pfizer and has served as a consultant for Abbvie, BMS, Janssen, Takeda, Pfizer, and Samsung.

Data Availability

The data underlying this article will be shared upon reasonable request to the corresponding author.