Anti-TNF Therapy and the Risk of Herpes Zoster Among Patients With Inflammatory Bowel Disease

Abstract

Background

The specific contribution of anti-TNF therapy to the onset of herpes zoster (HZ) in patients with inflammatory bowel disease (IBD) remains uncertain. Thus, the purpose of this nested case-control study was to explore whether the use of anti-TNF therapy is associated with an increased risk of HZ.

Methods

Using the Regie de l’Assurance Maladie du Québec, we identified incident cases of IBD between 1998 and 2015. We matched IBD cases of HZ with up to 10 IBD HZ-free controls on year of cohort entry and follow-up. Current use was defined as a prescription for anti-TNF therapy 60 days before the index date, with nonuse as the comparator. We conducted conditional logistic regression to estimate odds ratios (ORs) with 95% confidence intervals (CIs), adjusting for potential confounders.

Results

The cohort consisted of 15,454 incident IBD patients. Over an average follow-up of 5.0 years, 824 patients were diagnosed with HZ (incidence of 9.3 per 1000 person-years). Relative to nonuse, current use of anti-TNF therapy was associated with an overall increased risk of HZ (OR, 1.5; 95% CI, 1.1–2.1). The risk was increased among those older than 50 years (OR, 2.1; 95% CI, 1.2–3.6) and those additionally using steroids and immunosuppressants (OR, 4.1; 95% CI, 2.3–7.2).

Conclusions

Use of anti-TNF therapy was associated with an increased risk of HZ among patients with IBD, particularly among those older than 50 years and those on combination therapy. Prevention strategies for HZ ought to be considered for younger IBD patients commencing treatment.

Introduction

Inflammatory bowel disease (IBD), including Crohn’s disease (CD) and ulcerative colitis (UC), is a chronic gastrointestinal condition marked by an altered regulation of the immune system.^{1, 2} This altered immune response among patients with IBD might increase the risk of opportunistic infections, such as herpes zoster (HZ).³ Herpes zoster is an infection driven by the reactivation of the varicella zoster virus and can be associated with debilitating effects including depression and postherpetic neuralgia.^{2, 4} Studies have found that IBD patients are at a significantly increased risk of HZ compared with non-IBD patients,^{3, 5–7} with some observing up to a doubling of the risk.⁵ Treatment for IBD is aimed at suppressing the immune response, which may contribute to this increased risk.^8,9 Therefore, research has been geared toward identifying risk factors for HZ in the IBD population, including the role of biologic therapy.

Studies on the link between biologic therapy and HZ have been on the rise; however, study populations and treatment regimens of interest have varied across studies. Indeed, several found that treatment with biologics in IBD patients, alone or in combination with other immunosuppressive drugs, was associated with a heightened risk of HZ, with some observing this effect in the younger population (<50 years of age).^{6, 8, 10–12} On the other hand, one study found that combination therapy with anti-TNF and thiopurines was associated with an increased risk of HZ, whereas anti-TNF therapy alone was not.¹³ Furthermore, one study found that treatment with anti-TNF therapy was not associated with an increased risk of HZ compared with nonbiologic therapies in a population of patients with inflammatory diseases (including IBD).⁴

Taken together, though several studies have examined the role of overall biologic treatment for IBD in the risk of HZ, the specific contribution of anti-TNF therapy (ie, anti-TNF therapy alone while accounting for other IBD medications) to this risk is less certain. To date, studies investigating this association in Quebec, a Canadian province with a high incidence of IBD, are lacking.¹⁴ Thus, a better understanding of the potential role of anti-TNF therapy in the risk of HZ among IBD patients in this context is warranted. The present study therefore aimed to investigate whether the independent use of anti-TNF therapy is associated with an elevated risk of HZ among patients with IBD enrolled in the Quebec public drug plan.

MATERIALS AND METHODS

Data Source

This study was performed using administrative health care databases from Quebec, a Canadian province providing universal health care coverage for nearly all of its residents (over 8 million individuals). The Regie de l’Assurance Maladie Du Quebec (RAMQ) database holds information on patient demographics, records of physician billings on medical diagnoses, services, \ procedures, and prescription claims filed under the provincial public drug plan. In addition, the Maintenance et exploitation des données pour l’étude de la clientèle hospitalière (Med-Echo) database contains all hospital discharge records. The public drug plan insures over 40% of the Quebec population, including those age 65 years and older, those receiving social assistance along with their dependents, and all individuals not receiving private drug insurance through their employer. A prior study assessing the quality of drug information in the RAMQ prescription claims database found that it represents one of the most reliable sources for determining drugs dispensed to individuals in Quebec.¹⁵

Study Population

Using the RAMQ and Med-Echo databases, we downloaded data on all patients with a diagnosis of IBD between January 1, 1996, and December 31, 2015. Within this population, we identified patients meeting the IBD case-defining criteria, employing a previously validated algorithm.^{14, 16} To meet the IBD case definition, patients were required to have (1) at least 1 hospitalization discharge record with a primary diagnosis of CD or UC or (2) at least 4 physician billings with a diagnosis of CD or UC within a 2-year interval. The date of IBD diagnosis was considered as the date of the fourth physician billing within a 2-year time window or the date of the first hospitalization discharge, whichever occurred first. To limit the cohort to incident IBD, cases were counted as of January 1, 1998 (ie, patients meeting the IBD case-defining criteria before 1998 were excluded).

Diagnoses for CD and UC were identified using the relevant ICD-9 (CD, ICD-9 555; UC, ICD-9 556) and ICD-10 (CD, ICD-10 K50; UC, ICD-10 K51) codes. For patients with both CD and UC codes, an adapted algorithm¹⁶ was used, taking into account both hospitalizations and physician billings, to classify patients into 1 of 3 categories (CD, UC, or unclassified). As such, hospitalizations and physician billings were weighted using the following scoring system: −2 for hospitalization discharge with a diagnosis of CD, −1 for each physician billing for CD, +2 for hospitalization discharge with a diagnosis of UC, and +1 for each physician billing for UC. Patients with an overall score of −2 and below were classified as CD, those with an overall score of +2 and above were classified as UC, and those with a score between −2 and +2 were considered unclassified IBD.

The date of IBD diagnosis (ie, the date the case-defining criteria are met) was considered the date of cohort entry. We required patients to be registered with the RAMQ public drug plan for at least a year before cohort entry and have at least 1 day of follow-up to allow a minimum time for the event to occur. Furthermore, we excluded patients exposed to anti-TNF therapy in the year before cohort entry. We also excluded individuals with a diagnosis of HZ in the 2 years before cohort entry to account for those with a history of the disease. Patients were then followed starting 1 day after cohort entry up until the date of the outcome, end of drug insurance coverage, date of death, or end of study period (December 31, 2015), whichever came first.

Case-control Selection

We searched for all patients with a first time ICD-9 or ICD-10 code (053.xx/B02.x group) for incident HZ at any time after cohort entry. The date of the first recorded HZ code was defined as the index date. These codes have been used across several studies investigating HZ and have been shown to be valid.^{17, 18}

Each case was then matched with up to 10 HZ-free controls who were still at risk of the event in the risk set defined by the case (ie, those still being followed and event-free at the time of the case event). We matched on year of cohort entry (to account for changes in practice and drug availabilities) and duration of follow-up. The date corresponding to the same duration of follow-up for the cases and controls was set as the index date for the controls.

Exposure Definition

Given the acute nature of the study outcome, the definition of exposure focused on current use of anti-TNF therapy. Thus, for both cases and matched controls, we identified all those who were dispensed an anti-TNF prescription (adalimumab, infliximab, golimumab) in the 60 days preceding the index date. Nonuse was defined as no prescription for an anti-TNF dispensed in the 60 days preceding the index date and was used as the reference category for all analyses.

In secondary analyses, we considered exposure to multiple drug classes in the 60 days before index date. Exposure to multiple drug classes, which we considered as combination therapy, was defined as pharmacy dispensation for anti-TNF therapy and a pharmacy dispensation for either a steroid (eg, prednisone, budesonide, methylprednisone, hydrocortisone) or an immunosuppressant (eg, azathiopurine, 6-mercaptopurine, methotrexate) or both in the 60 days before index date.

Statistical Analyses

We conducted a nested case-control analysis. We performed conditional logistic regression to estimate odds ratios (ORs) with 95% confidence intervals (CIs) of HZ associated with the use of anti-TNF therapy relative to nonuse among patients with IBD.¹⁹

We included the following potential confounders in the model, measured in the 2 years before cohort entry: age (at cohort entry), sex, HIV/AIDS, cancer, diabetes mellitus, other immune-mediated inflammatory diseases (eg, rheumatoid arthritis, psoriasis, psoriatic arthritis, ankylosing spondylitis, systemic lupus erythematosus, carditis, multiple sclerosis, celiac disease), liver disease, chronic kidney disease, respiratory diseases (eg, asthma, bronchitis, chronic obstructive pulmonary disease), and cardiovascular disease. We also adjusted for use of other IBD medications in the year before index date, which included the following: 5-aminosalicylic acid (5-ASA, eg, sulphasalazine, mesalazine, olsalazine), steroids (eg, prednisone, budesonide, methylprednisone, hydrocortisone), immunosuppressants (eg, azathiopurine, mercaptopurine, methotrexate), and other biologics (eg, vedolizumab, ustekinumab). Finally, to account for disease severity, we adjusted for the following in the year before the index date: any surgery on the digestive system and/or abdominal region, number of hospital admissions, and number of physician billings for all causes.

Secondary and Sensitivity Analyses

We conducted 3 secondary analyses. To explore whether the association between anti-TNF therapy and HZ varied by age and sex, we stratified the cohort by predefined age categories (<50 and ≥50) and by sex. Second, to determine whether being exposed to multiple drug classes adds to the risk of HZ, we examined the risk of HZ in patients using both an anti-TNF and either an immunosuppressant or steroid, or all 3 in the 60 days before index date. Finally, to investigate whether the association varies by type of IBD, we stratified the cohort based on a diagnosis for UC or CD.

As a sensitivity analysis, we redefined the exposure definition to account for past users of anti-TNF therapy. Specifically, we lengthened the exposure assessment period to consider those with a last prescription for anti-TNF therapy dispensed within 6 months of the index date and within 1 year of the index date.

ETHICAL CONSIDERATIONS

This study was approved by the research ethics board of the Lady Davis Research Institute of the Jewish General Hospital, Montreal, Canada.

RESULTS

A total of 15,454 incident IBD patients were included in the cohort (Fig. 1). The mean (standard deviation) age at cohort entry was 60.1 (17.5) years. The median follow-up was 4.2 years, and the mean (standard deviation) follow-up was 5.0 (4.0) years. During 88,602 person-years of follow-up, 824 patients were diagnosed with HZ and matched to 8240 controls, generating a crude incidence rate of 9.3 per 1000 person-years. Out of 824 cases and 8240 matched controls, 5012 were diagnosed with CD, 3687 were diagnosed with UC, and 365 had unclassified IBD. Overall, 48 (5.8%) cases and 369 (4.5%) controls were exposed to anti-TNF therapy over our exposure window of interest (ie, 60 days preceding HZ diagnosis). Out of those exposed to anti-TNF, 91.7% of cases and 85.9% of controls were exposed to infliximab.

Table 1 displays the baseline characteristics of the cases and matched controls. Cases were predominantly female and were older on average. Furthermore, cases had a slightly higher number of previous immune-mediated inflammatory diseases, a higher proportion were prescribed other IBD medications including steroids, and a higher proportion underwent surgeries involving the abdominal region and/or digestive tract. Less than 5 cases or controls were exposed to other types of biologics in the 60 days before index date.

Table 2 presents the results of the primary analysis and the results of the analysis considering multiple drug classes. Relative to nonuse, current use of anti-TNF therapy was associated with an overall increased risk of HZ (adjusted OR, 1.5; 95% CI, 1.1–2.1). Furthermore, the risk increased in patients with prescriptions for both anti-TNF therapy and a steroid dispensed in the 60 days before index date (adjusted OR, 2.4; 95% CI, 1.2–4.7) and in patients with prescriptions for both anti-TNF therapy and an immunosuppressant dispensed in the 60 days before index date (adjusted OR, 2.6; 95% CI, 1.5–4.6). Finally, the risk substantially increased in patients with prescriptions for anti-TNF therapy, a steroid, and an immunosuppressant, all within the 60 days before index date (adjusted OR, 4.1; 95% CI, 2.3–7.2).

The results of the secondary and sensitivity analyses are summarized in Figure 2 and presented in the supplementary content (see supplementary data content Tables 1–4, which present the results of the secondary and sensitivity analyses). The risk associated with the current use of anti-TNF therapy relative to nonuse was slightly higher among males than females; however, the 95% CIs for both estimates crossed the null. Furthermore, the risk increased in patients under 50 years of age (adjusted ORs 2.0; 95% CI, 1.2–3.5), especially among females (adjusted ORs 2.6; 95% CI, 1.1–5.8) within this age group. Although, the risk was not increased in patients 50 years of age and older. When stratifying by IBD type, the risk was elevated among those with CD (adjusted OR, 1.9; 95% CI, 1.2–2.9). Finally, in sensitivity analyses, the risk decreased gradually when lengthening the exposure assessment period from 60 days to 6 months (adjusted OR, 1.4; 95% CI, 1.0–1.9) and 1 year (adjusted OR, 1.3; 95% CI, 0.9–1.8).

Discussion

Our study investigated the independent use of anti-TNF therapy and its potential association with the risk of HZ among patients with IBD. We assembled a cohort of 15,454 incident IBD patients from the Quebec administrative database between 1998 and 2015 and followed-up for an average of 5 years. We observed an HZ incidence rate of 9.3 per 1000 person-years, which was higher than what was reported in a previous study by our group, likely due to differing study entry criteria.³ Namely, our study required patients to be HZ-free in the 2 years before study entry and to be registered with the Quebec public drug plan, which may have generated an older and more at-risk cohort on average. Overall, current use of anti-TNF therapy was associated with an increased risk of HZ relative to nonuse. This risk increased when patients were additionally dispensed a steroid or an immunosuppressant or both within the 60 days preceding the HZ diagnosis. Furthermore, a heightened risk was observed among those younger than 50 years of age, especially in females of this age group. The risk was also increased among those with CD, likely due to higher likelihood of combination therapy among this subgroup.²⁰ In sensitivity analyses, extending the exposure window to 6 months and 1 year gradually pulled the effect estimate toward the null, suggesting that a residual effect of anti-TNF therapy may be present although not statistically significant. Finally, we did not explore the type-specific anti-TNF effect due to minimal exposure to those other than infliximab.

To date, several studies examined the association between biologic therapy (including anti-TNF) and the risk of HZ.^{4, 6, 8, 10–12,21}. A systematic review and meta-analysis of randomized controlled trials found that treatment with biologic agents for IBD increased the risk of opportunistic infections such as HZ.²² Moreover, several observational studies found that treatment with biologic therapy increased the risk of HZ in the IBD population.^{6, 8, 10, 12} Interestingly, some observed this effect in populations that were younger on average (<50 years old) (OR, 1.81; 95% CI, 1.48–2.21), and some also observed that combination therapy (with steroids or immunosuppressants) further increased the risk (range OR, 1.65–14.5).^{6, 8, 10, 12} On the other hand, another study found that although combination therapy with immunosuppressants and anti-TNF agents increased the risk of HZ (hazard ratio [HR], 1.65; 95% CI, 1.22–2.23), treatment with anti-TNF alone did not increase this risk (HR, 1.15; 95% CI, 0.96–1.38).¹³ Similarly, another study found that patients treated with anti-TNF therapy were not at a heightened risk for HZ than patients treated with other nonbiologic treatment regimens (HR, 1.00; 95% CI, 0.77–1.29); however, this study included patients with IBD and other select inflammatory diseases, such as rheumatoid arthritis.⁴ Finally, 1 study observed a decreased risk of opportunistic infections (including HZ) in anti-TNF users compared with users of immunosuppressants (HR, 0.57; 95% CI, 0.38–0.87).¹² Taken together, although there is evidence on the association between anti-TNF and HZ, the findings vary depending on the specific population and comparator under study and thus are difficult to interpret. Importantly, few have isolated the effect of anti-TNF therapy alone while adjusting for other IBD medications, and many did not consider adjusting for IBD disease severity or proxies for disease severity.^{4, 6, 11, 12} Nevertheless, the results of our study are consistent with the majority of prior studies on the topic, in that anti-TNF therapy is likely predisposing IBD patients to an increased risk of HZ.

Our study points to an increased risk in IBD patients under 50 years of age—an age group that is currently not considered for HZ preventive measures. The increased risk observed in this subgroup may be attributable to the lower likelihood of prior HZ vaccination in younger patients compared with older age groups.²³ Moreover, it may also be explained by the higher incidence of CD in the younger population, which typically presents as a more severe form of IBD and will more often require combination therapy.^14,23

In Quebec, the current vaccination guidelines for HZ recommend vaccinating immunocompetent or immunosuppressed individuals 50 years of age and older, including those who are or will be commencing treatment with immunosuppressants or biologics, using preferentially the inactivated form.²³ This recommendation is based on evidence that the risk of HZ increases dramatically with older age, starting at around age 50.²⁴ However, based on our results and the current literature,^{3, 6, 12} accumulating evidence suggests that the younger IBD population (namely, those under 50 years of age) may be considered as candidates for the inactivated HZ vaccine. Indeed, our study indicates that anti-TNF therapy, alone and in combination with other IBD drug classes, may be contributing to the increased risk of HZ reported in the younger IBD population. Thus, given recent research showing safety and efficacy of the HZ vaccine in immunocompromised patients aged 18 years and older,^25–28 preventive strategies for HZ could be further extended to younger IBD patients undergoing such therapies if shown to be cost-effective.

Our study presents some key strengths. First, to our knowledge, this is the first study investigating the association between anti-TNF therapy and the risk of HZ among patients with IBD in Quebec on the public drug plan. Given the high incidence of IBD in this Canadian province, determining the risk of HZ in users of anti-TNF therapy is of clinical importance. Furthermore, a prior study using the RAMQ data set observed an increased risk of HZ among IBD patients³; however, the contribution of anti-TNF therapy to this risk was not investigated. Second, we conducted a population-based study using the RAMQ data set that enrolled all patients eligible for the Quebec public drug plan over the study period. Importantly, use of the Quebec public drug plan database was shown to be the most effective way of determining drugs dispensed to individuals.¹⁵ Third, few prior studies investigating this association adjusted for IBD disease severity or proxies for disease severity. Though we did not have any direct information on disease severity, we adjusted our models for surrogate markers of overall health and proxies for disease severity such as steroid use, surgeries on the abdominal and digestive system, and number of physician billings and hospital admissions. Finally, we enrolled IBD patients of all age groups, as evidence for the risk of HZ in younger IBD patients treated with anti-TNF therapy is limited.

Our study also has several limitations. First, the Quebec public drug plan covers individuals and their dependents who are not covered by private insurance, thus consisting primarily of older individuals often of lower socioeconomic status.²⁹ Therefore, as individuals covered by the public drug plan represent a very specific demographic of the overall Quebec population, the generalizability of our results may be limited.²⁹ Second, we did not have access to information on HZ vaccination history, which may have differed among our specified age subgroups. Third, a significant proportion of IBD patients treated with anti-TNF therapy undergo dose escalations, which may affect the risk of HZ. However, our database did not contain information on anti-TNF dosing. Finally, given the administrative nature of the RAMQ database, we were unable to obtain information on lifestyle factors such as smoking and body mass index (BMI).

In this population-based study, current use of anti-TNF therapy was associated with an increased risk of HZ among patients with IBD who were enrolled in the Quebec public drug plan during the study period. This risk was especially elevated among those older than 50 years of age and those on combination therapy. Given the significant complications of HZ that can arise in this immunocompromised patient population, prevention strategies for HZ should be further examined for younger patients prescribed such treatments.

Supported by: This work obtained CAS Research support from the Division of Gastroenterology and Hepatology McGill University health center.

Conflicts of Interests: PLL has been a speaker and/or advisory board member for AbbVie, Arena Pharmaceuticals, Celltrion, Falk Pharma GmbH, Ferring, Genetech, Janssen, Merck, Pharmacosmos, Pfizer, Roche, Shire and Takeda and has received unrestricted research grants from AbbVie, MSDs and Pfizer. TB has been a speaker and/or advisory board member for AbbVie, Janssen, Takeda, Pfizer, Merck, Ferring, and Shire and has received unrestricted research grants from AbbVie and Janssen. AB has been a speaker and/or advisory board member for AbbVie, Roche, Takeda, Pfizer, and Janssen. All other authors declare no conflicts.

References