Abstract
Goals:
Characterize and compare risk of CDI and CMVC after initiation of vedolizumab or anti-tumor necrosis factor (TNF)α agents for UC.
Background:
Immunosuppression is a risk factor for gastrointestinal infections including Clostridioides difficile infection (CDI) and cytomegalovirus colitis (CMVC) among patients with ulcerative colitis (UC), however the risk according to biologic class is poorly understood.
Study:
Retrospective cohort study of adults with UC initiating vedolizumab or anti-TNFα agents 6/1/14–12/31/20 at a large academic health system. The primary outcome for both CDI and CMVC analyses were first CDI or CMVC after biologic initiation. The secondary outcome for the CDI analysis was severe CDI (>10,000 WBC or serum creatinine >1.5 mg/dL). Independent variables included demographics and UC history/severity factors. Inverse-probability of treatment weighted (IPTW) Cox regression was performed to assess hazard of CDI by biologic group. Due to few outcomes, CMVC was reported descriptively.
Results:
805 UC patients initiated vedolizumab (n=195) or anti-TNFα agents (n=610). There were 43 CDIs and 11 severe CDIs over 1,436 patient-years. IPTW Cox regression demonstrated no association between CDI and vedolizumab vs anti-TNFα (HR 0.33, 95% CI 0.05–2.03) but identified a significantly lower hazard of severe CDI for vedolizumab vs anti-TNFα (HR 0.10, 95% 0.01–0.76). There were 5 cases of CMVC, all in the anti-TNFα group.
Conclusions:
There was a lower adjusted risk of severe CDI but not total CDI associated with vedolizumab. CMVC was not observed after initiating vedolizumab. These findings may provide reassurance regarding the use of vedolizumab when also considering risk of gastrointestinal infections.
Keywords: inflammatory bowel disease, anti-TNF, anti-integrin, Clostridioides difficile, cytomegalovirus
Introduction
Clostridioides difficile infection (CDI) and cytomegalovirus colitis (CMVC) are known complications of ulcerative colitis (UC) that are commonly evaluated during acute disease exacerbations.1,2 CDI has been independently associated with an increased risk of both colectomy and mortality among patients with UC.1,3–7 CMVC has been independently associated with increased inpatient mortality, hospital length of stay, cost of care, and need for rescue therapy in this population.8,9 Meta-analytic data has identified a lower risk of colectomy when antiviral therapy is administered to patients with both steroid-refractory UC and tissue evidence of CMVC.10 Therefore, identifying risk factors to facilitate prompt recognition and treatment of coexisting gastrointestinal (GI) infections during a UC exacerbation has the potential to improve patient outcomes.
Immunosuppressive therapy is a known risk factor for both CDI and CMVC among patients with UC.11,12 Vedolizumab, an α4β7 integrin antagonist which selectively inhibits lymphocyte trafficking to the intestine, is often preferred over more systemically active biologics for patients who are at greater risk for infections.13 While the overall safety profile of vedolizumab is favorable14, its gut-specific activity raises concern for an increased risk of GI infections relative to other biologic classes. Recent data has identified a significantly greater risk of serious GI infections associated with vedolizumab compared to anti-tumor necrosis factor (TNF) α agents among patients with inflammatory bowel disease (IBD).15 However, the composite definition of GI infection for this study included biliary tract and hepatic infections that are not known to be associated with worse UC-specific outcomes.
Despite the clinical significance of CDI and CMVC for the UC population, the risk of these specific pathogens according to class of biologic therapy remains unknown. We therefore performed a retrospective cohort study to characterize and compare the individual risks of CDI and CMVC after initiation of vedolizumab or anti-TNFα agents for UC.
Materials and Methods
Study design and patient selection
This study included adult, biologic-naïve patients with UC (ICD-10-CM 51x) initiating vedolizumab, infliximab, or adalimumab 6/1/14–12/31/20 at Brigham and Women’s Hospital (Boston, MA, USA), Massachusetts General Hospital (Boston, MA, USA), and affiliated hospitals (MA, USA). Electronic health records and all available outside hospital records were manually reviewed. Patients with CD, indeterminate colitis, prior colectomy or biologic exposure, and initiation of biologics for non-UC indications were excluded. Patients were followed until first CDI or CMVC, biologic discontinuation/switch, colectomy, death, or last gastroenterology encounter through 8/1/21. CDI and CMVC were assessed separately, and the first diagnosis of either infection did not preclude subsequent assessment for the other infection.
Outcomes
For the CDI analysis, the primary outcome was first CDI episode after biologic initiation. CDI was defined as a positive stool C. difficile toxin or toxigenic C. difficile polymerase chain reaction (PCR) assay with associated oral vancomycin, fidaxomicin, or metronidazole prescriptions. The secondary outcome was severe CDI (WBC >15,000 or serum creatinine >1.5 mg/dL). Additional outcomes included CDI hospitalization, colectomy, or death within 30 days after CDI diagnosis. CDI hospitalizations included those where CDI was listed as a hospital discharge diagnosis. Discharge summaries were manually reviewed to confirm that patients received CDI treatment (oral vancomycin, fidaxomicin, or metronidazole) during the hospitalization. A subgroup analysis excluding patients with any history of pre-biologic CDI was also performed for the primary outcome.
For the CMVC analysis, the primary outcome was first CMVC episode after biologic initiation, defined as the presence of characteristic CMV viral inclusions or positive immunohistochemical staining for CMV on colon histopathology followed by administration of appropriate antiviral therapy. CMV serologies and CMV polymerase chain reaction assays were not considered in the diagnosis of CMVC. Patients with positive CMV histopathology findings that were not treated with antiviral therapy were not considered to have CMVC, but these cases were reported separately. Secondary outcomes included CMV hospitalization, colectomy, or death within 30 days of CMVC diagnosis. CMV hospitalizations included those where positive CMV biopsies were obtained during a hospitalization or CMVC was listed as a discharge diagnosis and appropriate anti-viral therapy was administered, confirmed by review of discharge summaries.
Additionally, we assessed all available stool microbial assays for other (i.e. non-C. difficile and non-CMV) bacterial, viral, or parasitic infections. We also recorded any documented episodes of acute diverticulitis during the study period.
Independent variables
The primary independent variable was vedolizumab vs anti-TNFα (infliximab or adalimumab) therapy. Other baseline variables included age, sex, race, body mass index (BMI), disease duration, immunomodulator use, corticosteroid use, substance (cigarette smoking, cannabis, or opioids) use, most recent Mayo endoscopic subscore and Montreal disease extent, UC hospitalization within the preceding 12 months, history of pre-biologic CDI or CMVC, history of cholecystectomy, history of malignancy, and the most recent values within the preceding 12 weeks for the following: serum albumin, C-reactive protein (CRP), fecal calprotectin, daily bowel movement frequency, and simple clinical colitis activity index (SCCAI; when documented in gastroenterology notes). UC hospitalization was defined as a hospitalization with UC listed as the primary discharge diagnosis or secondary discharge diagnosis with a UC-related complication as the primary diagnosis, based on prior definitions.16
Statistical analysis
Categorical and continuous variables were compared using Fisher’s exact test and the Wilcoxon rank sum test, respectively. Incidence rates for CDI and CMVC were calculated by biologic group. Kaplan-Meier analysis with the log-rank test was performed for CDI stratified by biologic group. Patients were censored at loss of follow-up, biologic discontinuation, colectomy, or death. Due to few events for CMVC and other GI infections, outcomes were reported descriptively and statistical tests were not performed to compare biologic groups.
For the CDI analysis, propensity scores (PSs) were calculated to control for differences in baseline characteristics using logistic regression of vedolizumab vs anti-TNFα on the following covariates: age, sex, Caucasian race, body mass index (BMI), disease duration, current systemic corticosteroid use, UC hospitalization within prior 12 months, last Mayo endoscopic subscore (3 vs <3), Montreal disease extent (>E1 vs ≤E1), albumin, and history of malignancy. CRP, fecal calprotectin, and SCCAI were excluded due to >20% missing data. Inverse probability of treatment weighting (IPTW) was performed using PSs. Covariate balance in the weighted sample was confirmed with <10% absolute standardized differences for all covariates (Figure 1).
Figure 1.
Standardized mean differences for covariates before and after inverse probability of treatment weighting.
An unweighted univariable Cox proportional hazards model was fit to calculate the unadjusted hazard ratio (HR) of CDI for vedolizumab vs anti-TNFα. A multivariable, IPT-weighted Cox model was then fit with two additional covariates extrinsic to the PS: pre-biologic CDI and immunomodulator exposure modeled as a time-varying covariate. For the analysis of severe CDI, which had fewer outcomes, these two covariates were incorporated into the PS calculation to avoid model overfitting.
Antibiotic and corticosteroid exposure were not modeled as time-varying covariates due to uncertainty in discontinuation dates. These were instead examined as corticosteroid and antibiotic exposures at any point during follow-up and also within the 30 days preceding CDI diagnosis, which is the interval of greatest CDI risk.17 Antibiotic exposure was defined as a minimum of one oral antibiotic prescription or 24-hour duration of intravenous antibiotics, excluding oral vancomycin, fidaxomicin, and metronidazole. P<0.05 was considered significant and the proportional hazards assumption was tested using Schoenfeld residuals. STATA/SE 17.0 (College Station, TX) was used for all analyses.
Results
Cohort characteristics
The cohort included 805 patients with UC who initiated vedolizumab (n=195) or anti-TNFα agents (n=610; 332 infliximab, 278 adalimumab) between 6/1/14–12/31/20. Compared to patients in the anti-TNFα group, patients in the vedolizumab group were older (median age 48 y vs 35 y, p<0.001), had longer disease duration (median 7 y vs 4 y, p<0.001), less commonly received concomitant systemic corticosteroids (43.1% vs 70.5%, p<0.001), less commonly had a Mayo endoscopic subscore of 3 (20.2% vs 32.8%, p<0.001), less commonly had a prior UC hospitalization within 12 months (13.8% vs 42.0%, p<0.001), and more commonly had a history of malignancy (31.3% vs 9.5%, p<0.001) (Table 1).
Table 1.
Patient characteristics by biologic group
| Characteristic | Vedolizumab | Anti-TNFα | P-value† |
|---|---|---|---|
| N‡ | 195 | 610 | |
| Follow-up time until CDI or censoring, d, median (IQR) | 583 (243, 956) | 363.5 (125, 1064) | |
| Follow-up time until CMVC or censoring, d, median (IQR) | 587 (254, 956) | 384 (134, 1076) | |
| Age, y, median (IQR) | 48 (31, 64) | 35 (26, 51) | <0.001 |
| Female sex | 108 (55.4%) | 333 (54.6%) | 0.87 |
| Race | |||
| White, non-Hispanic | 177 (90.8%) | 529 (86.7%) | 0.48 |
| Hispanic | 5 (2.6%) | 16 (2.6%) | |
| Black, non-Hispanic | 5 (2.6%) | 21 (3.4%) | |
| Asian | 4 (2.1%) | 21 (3.4%) | |
| Native American or Hawaiian | 1 (0.5%) | 1 (0.2%) | |
| Unknown | 3 (1.5%) | 22 (3.6%) | |
| BMI, kg/m2, median (IQR), n=194, 583 | 24.45 (21.8, 27.6) | 24.6 (21.5, 28.15) | 0.63 |
| Disease duration, y, median | 7 (2, 18) | 4 (1, 10) | <0.001 |
| Current immunomodulator | 0.035 | ||
| Thiopurine | 18 (9.2%) | 95 (15.6%) | |
| Methotrexate | 5 (2.6%) | 8 (1.3%) | |
| Current corticosteroids | <0.001 | ||
| Systemic (prednisone or methylprednisolone) | 84 (43.1%) | 430 (70.5%) | |
| Oral budesonide | 26 (13.3%) | 62 (10.2%) | |
| Corticosteroid enemas only | 8 (4.1%) | 14 (2.3%) | |
| Current smoking | 3 (1.5%) | 27 (4.4%) | 0.08 |
| Current cannabis | 16 (8.2%) | 75 (12.3%) | 0.12 |
| Current opioids | 10 (5.1%) | 44 (7.2%) | 0.31 |
| Mayo endoscopic subscore, n=193, 600 | <0.001 | ||
| 0 (normal mucosa) | 12 (6.2%) | 20 (3.3%) | |
| 1 (mild inflammation) | 47 (24.4%) | 94 (15.7%) | |
| 2 (moderate inflammation) | 95 (49.2%) | 289 (48.2%) | |
| 3 (severe inflammation) | 39 (20.2%) | 197 (32.8%) | |
| Montreal disease extent >E1 (proctitis), n=194, 604 | 158 (81.4%) | 544 (90.1%) | 0.002 |
| Albumin, g/dL, median (IQR), n=188, 566 | 4.2 (3.8, 4.5) | 4 (3.5, 4.4) | <0.001 |
| CRP, mg/dL, median (IQR), n=161, 496 | 2.7 (1, 8.5) | 7.5 (1.6, 31.9) | <0.001 |
| Fecal calprotectin > 120 ug/g, n=74, 146 | 59 (79.7%) | 127 (87.0%) | 0.17 |
| Daily bowel frequency, median (IQR), n=170, 470 | 4 (2, 6) | 6 (4, 10) | <0.001 |
| SCCAI, median (IQR), n=168, 452 | 4 (2, 6) | 6 (3, 9) | <0.001 |
| History of CDI (pre-biologic) | 14 (6.5%) | 84 (12.1%) | 0.023 |
| History of pre-biologic CMVC | 0 (0.0%) | 5 (9.5%) | 0.34 |
| History of cholecystectomy | 6 (3.1%) | 33 (5.4%) | 0.25 |
| History of malignancy | 61 (31.3%) | 58 (9.5%) | <0.001 |
| UC hospitalization within prior 12 months | 27 (13.8%) | 256 (42.0%) | <0.001 |
| Dose escalation of biologic during follow-up | 79 (40.5%) | 311 (51.0%) | 0.013 |
| Systemic corticosteroids during follow-up | 103 (47.9%) | 366 (52.7%) | 0.123 |
| Any antibiotic prescription during follow-up | 71 (33.0%) | 246 (35.5%) | 0.286 |
| >1 antibiotic prescription during follow-up | 40 (18.6%) | 125 (18.0%) | 0.457 |
All p-values were calculated using Wilcoxon rank-sum test for continuous data and Fisher’s exact test for categorical data.
Total sample size unless otherwise specified due to missing data
Abbreviations: TNF = tumor necrosis factor, IQR = interquartile range, BMI = body mass index, CRP = C-reactive protein, SCCAI = simple clinical colitis activity index, CDI = C. difficile infection
CDI outcomes
There were 43 CDIs, 11 severe CDIs, and 25 CDI hospitalizations over 1,436 patient-years of follow-up. All of these outcomes occurred less commonly in the vedolizumab vs anti-TNFα group (incidence rates per 1000 patient-years: 5.6 vs 37.9 for CDI, 2.8 vs 10.2 for severe CDI, and 5.6 vs 21.3 for CDI hospitalization). There were no significant differences in CDI-associated colectomies or deaths, exposure to antibiotics or corticosteroids during follow-up, or exposure to antibiotics or corticosteroids within the 30 days preceding CDI (Table 2). Kaplan-Meier analysis demonstrated separation in CDI survival functions by biologic group (Figure 2).
Table 2.
C. difficile and CMV outcomes by biologic group
| CDI Outcomes | Vedolizumab | Anti-TNFα | P-value† |
|---|---|---|---|
| CDI, incidence rate (per 1,000 patient-years) | 5.6 | 37.9 | 0.001 |
| Severe CDI, incidence rate (per 1,000 patient-years) | 2.8 | 10.2 | 0.18 |
| CDI hospitalization, incidence rate (per 1,000 patient-years) | 5.6 | 21.3 | 0.042 |
| Colectomy within 30 days after CDI, fraction (%)‡ | 0/2 (0.0%) | 1/41 (2.4%) | 1.00 |
| Death within 30 days after CDI, fraction (%)‡ | 0/2 (0.0%) | 0/41 (0.0%) | 1.00 |
| Antibiotics within 30 days prior to CDI, fraction (%)‡ | 0/2 (0.0%) | 3/41 (7.3%) | 1.00 |
| Systemic corticosteroids within 30 days prior to CDI, fraction (%)‡ | 1/2 (50.0%) | 28/41 (68.3%) | 1.00 |
| CMV Outcomes | Vedolizumab | Anti-TNFα | -- |
| CMVC, incidence rate (per 1,000 patient-years) | 0 | 4.5 | -- |
| CMVC hospitalization, incidence rate (per 1,000 patient-years) | 0 | 3.6 | -- |
| Colectomy within 30 days after CMVC, fraction (%)§ | -- | 1/5 (20.0%) | -- |
| Death within 30 days after CMVC, fraction (%)§ | -- | 0/5 (0.0%) | -- |
All p-values were calculated using Fisher’s exact test for categorical data except for CDI, severe CDI, CDI hospitalization, which were calculated using the log-rank test. Due to the small number of CMVC outcomes, p-values were not calculated to compare biologic groups.
Denominator for these outcomes includes only patients who developed CDI (i.e. n=2 for vedolizumab, n=41 for anti-TNFα)
Denominator for these outcomes includes only patients who developed CMVC (i.e. n=5 for anti-TNFα only)
Abbreviations: TNF = tumor necrosis factor, CDI = C. difficile infection, CMVC = cytomegalovirus colitis
Figure 2. Kaplan-Meier analysis of time to post-biologic C. difficile infection.
Patients were censored at loss of follow-up, biologic discontinuation, or total colectomy. P-value was calculated using the log-rank test.
Complete covariate data was available for 749/805 (93.0%) patients, among which 35/43 (81.4%) CDI events were maintained. The unadjusted Cox model demonstrated a significantly lower hazard of CDI for vedolizumab vs anti-TNFα (HR 0.17, 95% CI 0.04–0.71). The multivariable IPT-weighted Cox model demonstrated no association between CDI and vedolizumab vs anti-TNFα (HR 0.33, 95% CI 0.05–2.03) or immunomodulator exposure (HR 1.01, 95% CI 0.41–2.40). Pre-biologic CDI was associated with an increased hazard of CDI (HR 5.95, 95% CI 2.93–12.09). Among patients in the full cohort who developed CDI, 17/43 (39.5%) had a history of pre-biologic CDI diagnosed a median of 227 days (IQR 160–550 days) prior to CDI. No CDIs met criteria for recurrence, as all cases occurred >8 weeks after the last episode of pre-biologic CDI.18
For severe CDI, pre-biologic CDI and baseline immunomodulator use were incorporated into the PS. IPTW was repeated with <10% absolute standardized differences confirmed across all covariates. After IPTW Cox regression, there was a significantly lower adjusted hazard of severe CDI for vedolizumab vs anti-TNFα (HR 0.10, 95% 0.01–0.76). In the subgroup analysis excluding patients with pre-biologic CDI, there were no significant associations between CDI and vedolizumab vs anti-TNFα (HR 0.63, 95% CI 0.10–3.98) or immunomodulator exposure (HR 1.30, 95% CI 0.47–3.63).
CMVC outcomes
There were 5 cases of CMVC over 1,470 patient-years (follow-up differs slightly from CDI analysis as patients in the CMVC analysis were not censored at time of CDI), all in the anti-TNFα group. Incidence rates of CMVC were 0 per 1,000 patient-years for vedolizumab and 4.5 per 1,000 patient-years for anti-TNFα. CMVC episodes occurred at a median of 9 days (range 6–382 days) after anti-TNFα initiation. Of the 5 patients who developed CMVC, 5 (100.0%) were receiving concomitant systemic corticosteroids at the time of CMVC diagnosis, 4 (80.0%) were hospitalized, one (20.0%) had colectomy within 30 days, and no patients died within 30 days (Table 2). However, one patient with CMVC died 34 days after CMVC diagnosis (21 days after colectomy due to post-operative complications). One additional patient had evidence of CMV viral inclusions on colon histopathology but did not undergo antiviral therapy. This patient was also in the anti-TNFα group and was receiving corticosteroids at the time of colon mucosal biopsies. This patient did not require colectomy and did not die within 30 days after the positive CMV biopsies. Additional characteristics for all 6 patients with positive CMV histopathology are presented in Table 3.
Table 3.
Detailed characteristics of patients with positive CMV histopathology on colonic biopsies
| Biologic | Sex | Prior history of CMVC | Age at biologic initiation | Mayo endoscopic subscore prior to biologic initiation | Mayo endoscopic subscore at time of biopsies for CMV | Days from biologic initiation to CMV biopsies | Colon CMV histopathology | Corticosteroids at the time of colonic biopsies for CMV | CMV biopsies taken during a hospitalization | Received antiviral therapy for CMVC | Colectomy within 30 days of CMVC | Death within 30 days of CMVC |
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| IFX | F | No | 78 | 3 | 3 | 9 | Numerous CMV inclusions are present. | Yes | Yes | Yes | No | No |
| ADA | M | No | 32 | 2 | 2 | 382 | Immunohistochemical stain for CMV shows scattered positive cells. | Yes | No | Yes | No | No |
| IFX | M | No | 78 | 3 | 3 | 7 | Viral cytopathic effect consistent with cytomegalovirus infection (H&E stain). | Yes | Yes | Yes | Yes | No† |
| IFX | M | No | 24 | 3 | 3 | 6 | Numerous enlarged cells with viral inclusions are identified. These cells are positive for CMV on immunochemistry. | Yes | Yes | Yes | No | No |
| IFX | M | No | 32 | 3 | 3 | 17 | CMV positive on immunohistochemical staining | Yes | Yes | Yes | No | No |
| ADA | F | No | 19 | 3 | 2 | 55 | Rare endothelial cells immunohisto-chemically positive for CMV | Yes | Yes | No | No | No |
Death did occur 34 days after CMV colitis diagnosis (due to post-operative complications after colectomy).
Abbreviations: CMVC = cytomegalovirus colitis, ADA = adalimumab, IFX = infliximab, F = female, M = male
Other GI infections
No positive results for other stool microbial assays were observed for either biologic group during the study period. There were two episodes of acute diverticulitis in the anti-TNFα group and none in the vedolizumab group.
Discussion
Our study identified incidence rates of 5.6/1000 patient-years for CDI and 0/1000 patient-years for CMVC with vedolizumab therapy. In contrast, incidence rates were 37.9/1000 patient-years for CDI and 4.5/1000 patient-years for CMVC with anti-TNFα therapy. Our CDI analysis attempted to identify differences in the risk of CDI by biologic group. After balancing covariates related to UC severity, there was no significant difference in the adjusted risk of CDI among those initiating vedolizumab vs anti-TNFα agents or among those receiving concomitant immunomodulators. However, vedolizumab was associated with a lower adjusted risk of severe CDI, which is an important risk factor for surgery and mortality.19,20
For our CMV analysis, there were too few events observed to make statistical comparisons by biologic group. However, it is notable that 4/5 CMVC cases occurred in the setting of hospitalization and all 5 patients were receiving corticosteroids at the time of CMVC diagnosis. We therefore suspect that CMV may be a bystander of severe UC exacerbations rather than a direct consequence of either vedolizumab or anti-TNFα therapy. This is further supported by the occurrence of CMVC at a median of only 9 days after initiation of biologic therapy, reducing the likelihood that chronic immunosuppression from biologics precipitated CMVC in our cohort.
These results provide additional insight regarding the risk of specific GI infections associated with vedolizumab and anti-TNFα therapy. Two recent claims-based studies have attempted to compare the risk of severe infections requiring hospitalization between vedolizumab and anti-TNFα agents.13,15 Singh et al. evaluated a cohort of nearly 6,000 patients with IBD, in which vedolizumab was associated with a greater risk of severe GI infections compared to anti-TNFα agents (HR 1.82, 95% CI 1.08–3.07).15 After stratification by IBD type, there was a greater risk of GI infections associated with vedolizumab for both Crohn’s disease (HR 2.90, 95% CI 1.21–6.94) and UC (HR 1.20, 0.57–2.53), though the result was statistically significant only for Crohn’s disease. In the study by Kirchgesner et al., the risk of GI infections was not separately assessed in a cohort of more than 35,000 patients with IBD. There was a lower risk of all severe infections associated with vedolizumab for UC, but this risk was reduced further after GI infections were excluded.13 Therefore, the findings of both of these large studies serve to elevate concerns regarding the risk of GI infections with vedolizumab therapy. It critical to note, however, that the definition of “GI” utilized in these studies encompassed a broad range of conditions, including gastroenteritis, cholangitis, peritonitis, and viral hepatitis, among others that are not associated with adverse IBD outcomes. In contrast, our study focused on CDI and CMVC individually, both of which have potential implications for the disease course of UC.5,7–9 Our study was also able to assess infections that were identified in both ambulatory and inpatient settings.
To our knowledge, no studies to-date have specifically evaluated the risk of CDI by biologic class. One study has assessed the risk of CMVC by biologic class, identifying a greater risk of CMV reactivation among UC patients receiving vedolizumab compared to anti-TNFα agents.12 However, the results were not statistically significant and the total sample size was limited to 33 patients. While the hazard of CMVC was higher among vedolizumab-treated patients (HR 2.3) in this study, clinical and endoscopic disease activity in this group were also significantly greater at the time of CMVC diagnosis. In the context of our findings, we speculate that these episodes of CMVC may also be due to severe flares of UC rather than an independent effect of biologic therapy.
The strengths of the study include assessment of granular variables such as endoscopic severity, disease extent, and biochemical markers, which are typically unavailable in claims-based studies. For the CDI analysis, IPTW achieved appropriate balance of relevant covariates despite baseline differences in these factors while maximizing the analyzed sample size. This obviated the need to perform a matched analysis which would have reduced statistical power. The study also provided patient-level clinical details regarding CMVC cases that support the hypothesis that positive histopathologic findings for CMV are typically observed during severe UC exacerbations and are less likely due to prolonged exposure to biologic therapy. By limiting the study to patients who were biologic-naïve, we have eliminated the possibility that infections could be due to residual effects of other biologics, which is often a limitation of studies that assess complications in a bio-exposed population.
Limitations of the study include the retrospective design and relatively modest number of CDI and CMVC outcomes with the potential for type II error. Greater UC severity in the anti-TNFα group may have led to more frequent UC exacerbations, CDI testing, and mucosal biopsies for CMV, increasing the potential for detection relative to the vedolizumab group. However, this study relies on the assumption that CDI and CMVC are symptomatic diseases that prompt patients to seek medical attention. It is standard of care in our health system for UC patients on biologic therapies with acute or worsening symptoms to have CDI testing and an endoscopic evaluation to exclude CMV. Our suspicion for significant under-detection of clinically relevant CDI or CMV in the vedolizumab group is therefore low. However, because assessments of other GI infections are not routinely performed for all symptomatic patients with UC at our institution, there may be under-detection of other potentially pathogenic organisms for both biologic groups in our cohort. There is data to suggest that anti-TNF therapy may have a beneficial impact on colonic dysbiosis in IBD, which is a known risk factor for CDI.21,22 Our study does not compare changes in fecal microbial composition after anti-TNF and vedolizumab treatment, though this is an important area of future research. Our CDI analysis is also limited by the inability to adjust for corticosteroid and antibiotic exposures as time-varying covariates. However, when these were examined as binary exposures, no differences by biologic group were observed. In our CMV analysis, we were unable to obtain baseline data regarding CMV serologic status, and thus are unable to differentiate between CMV reactivation and primary infection among the 5 identified cases of CMVC. This small number of cases also precluded statistical comparisons and subgroup analyses.
In summary, our study identified a significantly lower adjusted risk of severe CDI associated with vedolizumab compared to anti-TNFα agents but no significant difference in adjusted risk of any CDI by biologic group. There may also be a lower risk of non-severe CDI associated with vedolizumab in UC, but larger studies are needed to investigate this possibility. We observed that CMVC was rare after initiating either vedolizumab or anti-TNFα agents, typically occurring in the setting of a disease exacerbation requiring corticosteroids without prolonged biologic exposure. While prospective studies with standardized infection assessments are needed, these findings should serve to reassure UC patients and providers who are considering both vedolizumab therapy and the risk of GI infections.
Acknowledgements:
We would like to thank Harvard Catalyst for statistical support.
Financial Support
1. Authors’ declaration of personal interests:
JRA serves as a consultant for Takeda, Janssen, Pfizer, Pandion, Servatus, Finch Therapeutics, Iterative Scopes and Artugen and has grant support from Merck. RSD, JM, and HG have no financial or personal conflicts of interest to disclose.
2. Declaration of funding interests:
This work was supported in part by the National Institute of Diabetes and Digestive and Kidney Diseases (5T32DK007533-35 to R.S.D.)
Writing assistance: none
References
- 1.Singh H, Nugent Z, Yu BN, Lix LM, Targownik LE, Bernstein CN. Higher Incidence of Clostridium difficile Infection Among Individuals With Inflammatory Bowel Disease. Gastroenterology. 08 2017;153(2):430–438.e2. doi: 10.1053/j.gastro.2017.04.044 [DOI] [PubMed] [Google Scholar]
- 2.Domènech E, Vega R, Ojanguren I, et al. Cytomegalovirus infection in ulcerative colitis: a prospective, comparative study on prevalence and diagnostic strategy. Inflamm Bowel Dis. Oct 2008;14(10):1373–9. doi: 10.1002/ibd.20498 [DOI] [PubMed] [Google Scholar]
- 3.Jodorkovsky D, Young Y, Abreu MT. Clinical outcomes of patients with ulcerative colitis and co-existing Clostridium difficile infection. Dig Dis Sci. Feb 2010;55(2):415–20. doi: 10.1007/s10620-009-0749-9 [DOI] [PubMed] [Google Scholar]
- 4.Shrestha MP, Taleban S. Colectomy Rates Are Increasing Among Inpatients With Concomitant Ulcerative Colitis and Clostridioides difficile. J Clin Gastroenterol. Aug 2020;doi: 10.1097/MCG.0000000000001412 [DOI] [PubMed] [Google Scholar]
- 5.Law CC, Tariq R, Khanna S, Murthy S, McCurdy JD. Systematic review with meta-analysis: the impact of Clostridium difficile infection on the short- and long-term risks of colectomy in inflammatory bowel disease. Aliment Pharmacol Ther. 04 2017;45(8):1011–1020. doi: 10.1111/apt.13972 [DOI] [PubMed] [Google Scholar]
- 6.Chen Y, Furuya-Kanamori L, Doi SA, Ananthakrishnan AN, Kirk M. Clostridium difficile Infection and Risk of Colectomy in Patients with Inflammatory Bowel Disease: A Bias-adjusted Meta-analysis. Inflamm Bowel Dis. 02 2017;23(2):200–207. doi: 10.1097/MIB.0000000000000998 [DOI] [PubMed] [Google Scholar]
- 7.Tariq R, Law CCY, Khanna S, Murthy S, McCurdy JD. The Impact of Clostridium difficile Infection on Mortality in Patients With Inflammatory Bowel Disease: A Systematic Review and Meta-Analysis. J Clin Gastroenterol. 02 2019;53(2):127–133. doi: 10.1097/MCG.0000000000000968 [DOI] [PubMed] [Google Scholar]
- 8.Hendler SA, Barber GE, Okafor PN, Chang MS, Limsui D, Limketkai BN. Cytomegalovirus infection is associated with worse outcomes in inflammatory bowel disease hospitalizations nationwide. Int J Colorectal Dis. May 2020;35(5):897–903. doi: 10.1007/s00384-020-03536-8 [DOI] [PubMed] [Google Scholar]
- 9.Lee HS, Park SH, Kim SH, et al. Risk Factors and Clinical Outcomes Associated with Cytomegalovirus Colitis in Patients with Acute Severe Ulcerative Colitis. Inflamm Bowel Dis. Apr 2016;22(4):912–8. doi: 10.1097/MIB.0000000000000675 [DOI] [PubMed] [Google Scholar]
- 10.Shukla T, Singh S, Loftus EV, Bruining DH, McCurdy JD. Antiviral Therapy in Steroid-refractory Ulcerative Colitis with Cytomegalovirus: Systematic Review and Meta-analysis. Inflamm Bowel Dis. Nov 2015;21(11):2718–25. doi: 10.1097/MIB.0000000000000489 [DOI] [PubMed] [Google Scholar]
- 11.Balram B, Battat R, Al-Khoury A, et al. Risk Factors Associated with Clostridium difficile Infection in Inflammatory Bowel Disease: A Systematic Review and Meta-Analysis. J Crohns Colitis. Jan 2019;13(1):27–38. doi: 10.1093/ecco-jcc/jjy143 [DOI] [PubMed] [Google Scholar]
- 12.Hommel C, Roblin X, Brichet L, Bihin B, Pillet S, Rahier J-F. P579 Risk of CMV reactivation in UC patients with previous history of CMV infection following infliximab or vedolizumab treatments. Journal of Crohn’s and Colitis. 2018;12(supplement_1):S400–S400. doi: 10.1093/ecco-jcc/jjx180.706 [DOI] [Google Scholar]
- 13.Kirchgesner J, Desai RJ, Beaugerie L, Schneeweiss S, Kim SC. Risk of Serious Infections With Vedolizumab Versus Tumor Necrosis Factor Antagonists in Patients With Inflammatory Bowel Disease. Clin Gastroenterol Hepatol. Dec 2020;doi: 10.1016/j.cgh.2020.12.030 [DOI] [PubMed] [Google Scholar]
- 14.Colombel JF, Sands BE, Rutgeerts P, et al. The safety of vedolizumab for ulcerative colitis and Crohn’s disease. Gut. 05 2017;66(5):839–851. doi: 10.1136/gutjnl-2015-311079 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Singh S, Heien HC, Herrin J, et al. Comparative Risk of Serious Infections With Tumor Necrosis Factor α Antagonists vs Vedolizumab in Patients With Inflammatory Bowel Diseases. Clin Gastroenterol Hepatol. Feb 25 2021;doi: 10.1016/j.cgh.2021.02.032 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Ananthakrishnan AN, McGinley EL, Binion DG. Does it matter where you are hospitalized for inflammatory bowel disease? A nationwide analysis of hospital volume. Am J Gastroenterol. Nov 2008;103(11):2789–98. doi: 10.1111/j.1572-0241.2008.02054.x [DOI] [PubMed] [Google Scholar]
- 17.Hensgens MP, Goorhuis A, Dekkers OM, Kuijper EJ. Time interval of increased risk for Clostridium difficile infection after exposure to antibiotics. J Antimicrob Chemother. Mar 2012;67(3):742–8. doi: 10.1093/jac/dkr508 [DOI] [PubMed] [Google Scholar]
- 18.McDonald LC, Gerding DN, Johnson S, et al. Clinical Practice Guidelines for Clostridium difficile Infection in Adults and Children: 2017 Update by the Infectious Diseases Society of America (IDSA) and Society for Healthcare Epidemiology of America (SHEA). Clin Infect Dis. 03 2018;66(7):987–994. doi: 10.1093/cid/ciy149 [DOI] [PubMed] [Google Scholar]
- 19.Bhangu A, Nepogodiev D, Gupta A, Torrance A, Singh P, Collaborative WMR. Systematic review and meta-analysis of outcomes following emergency surgery for Clostridium difficile colitis. Br J Surg. Nov 2012;99(11):1501–13. doi: 10.1002/bjs.8868 [DOI] [PubMed] [Google Scholar]
- 20.Cheng YW, Fischer M. Clinical management of severe, fulminant, and refractory. Expert Rev Anti Infect Ther. 04 2020;18(4):323–333. doi: 10.1080/14787210.2020.1730814 [DOI] [PubMed] [Google Scholar]
- 21.Schierova D, Roubalova R, Kolar M, et al. Fecal Microbiome Changes and Specific Anti-Bacterial Response in Patients with IBD during Anti-TNF Therapy. Cells. 11 16 2021;10(11)doi: 10.3390/cells10113188 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Sanchis-Artero L, Martínez-Blanch JF, Manresa-Vera S, et al. Evaluation of changes in intestinal microbiota in Crohn’s disease patients after anti-TNF alpha treatment. Sci Rep. 05 11 2021;11(1):10016. doi: 10.1038/s41598-021-88823-2 [DOI] [PMC free article] [PubMed] [Google Scholar]