Tuberculosis and Immune Reconstitution Inflammatory Syndrome in Patients With Inflammatory Bowel Disease and Anti-TNFα Treatment: Insights From a French Multicenter Study and Systematic Literature Review With Emphasis on Paradoxical Anti-TNFα Resumption

Abstract Background The advent of anti–tumor necrosis factor α (anti-TNFα) has revolutionized the treatment of inflammatory bowel disease (IBD). However, susceptibility to active tuberculosis (TB) is associated with this therapy and requires its discontinuation. The risk of immune reconstitution inflammatory syndrome (IRIS) in this population is poorly understood, as is the safety of resuming anti-TNFα. Methods This French retrospective study (2010–2022) included all TB cases in patients with IBD who were treated with anti-TNFα in 6 participating centers. A systematic literature review was performed on TB-IRIS and anti-TNFα exposure. Results Thirty-six patients were included (median age, 35 years; IQR, 27–48). TB was disseminated in 86% and miliary in 53%. IRIS occurred in 47% after a median 45 days (IQR, 18–80). Most patients with TB-IRIS (93%) had disseminated TB. Miliary TB was associated with IRIS risk in univariate analysis (odds ratio, 7.33; 95% CI, 1.60–42.82; P = .015). Anti-TB treatment was longer in this population (median [IQR], 9 [9–12] vs 6 [6–9] months; P = .049). Anti-TNFα was resumed in 66% after a median 4 months (IQR, 3–10) for IBD activity (76%) or IRIS treatment (24%), with only 1 case of TB relapse. Fifty-two cases of TB-IRIS in patients treated with anti-TNFα were reported in the literature, complicating disseminating TB (85%) after a median 42 days (IQR, 21–90), with 70% requiring anti-inflammatory treatment. Forty cases of TB-IRIS or paradoxical reaction treated with anti-TNFα were also reported. IRIS was neurologic in 64%. Outcome was mostly favorable (93% recovery). Conclusions TB with anti-TNFα treatment is often complicated by IRIS of varying severity. Restarting anti-TNFα is a safe and effective strategy.

The advent of anti-tumor necrosis factor α (anti-TNFα) therapies has revolutionized the management of chronic inflammatory bowel disease (IBD) [1].However, since the early 21st century, evidence has emerged of an inherent susceptibility to active tuberculosis (TB) associated with their administration [2].This susceptibility is primarily due to reactivation of latent TB [3], with relative risk ranging from 2 to 16 in the literature [4][5][6].
This observation has led to the recommendation of screening and treating latent TB infection (LTBI) before initiating Open Forum Infectious Diseases M A J O R A R T I C L E anti-tumor necrosis factor (anti-TNF) therapy [7].In particular, TB under anti-TNF therapy manifests as aggressive forms, often extrapulmonary and disseminated [8,9].Unfortunately, TB under anti-TNF treatment mandates discontinuation of this therapeutic [10], even in patients with severe gastrointestinal pathology.The optimal time frame for resuming such treatment remains unclear, although several guidelines suggest a provisional period of 2 months, without clear justification [10].
The ensuing immune reconstitution associated with the cessation of anti-TNFα, coupled with the propensity for disseminated TB manifestations, creates increased susceptibility in patients to the potential onset of immune reconstitution inflammatory syndrome (IRIS), a phenomenon that can exhibit varying degrees of clinical severity [11,12].
Management of IRIS lacks standardized protocols and is primarily dictated by the severity of its manifestations.It ranges from symptomatic treatment to predominantly antiinflammatory approaches, often including corticosteroid therapy [13].Paradoxically, a handful of studies report the use of anti-TNFα agents in this context [14,15].
The aim of our study was to delineate the clinical features of TB in patients undergoing anti-TNFα therapy for IBD and to assess IRIS within this demographic, as well as its severity and subsequent management.
To achieve this objective, we conducted a retrospective multicenter study in the Ile-de-France region, France, involving patients with IBD undergoing anti-TNF therapy who experienced active TB.Additionally, we performed a systematic literature review on TB-IRIS in this population receiving anti-TNFα therapy, as well as cases of TB-IRIS treated with anti-TNFα agents.

Cohort
We performed a 12-year multicenter retrospective study in 6 French teaching hospitals in Ile-de-France.All consecutive patients were included if hospitalized in the participating centers between 2010 and 2022 with proven TB disease during treatment with anti-TNFα for IBD.
Patients were identified at each center through a request for the following ICD-10 codes: IBD (K50, K51, K52) and TB (A15-A19).Medical records were reviewed to verify the inclusion criteria: age >18 years, diagnosis of IBD, treatment with anti-TNFα at the time of TB diagnosis, and proven TB (positive culture for Mycobacterium tuberculosis complex or granulomatous lesion with caseous necrosis).

Collection of Data.
Data were collected on demographics, comorbidities, clinical examination, laboratory findings, imaging findings, microbiologic testing, and therapeutic management during hospitalization and follow-up.In this research, IRIS was defined in operational terms as the appearance of new lesions or the deterioration of preexisting ones after a period of improvement, with no proof of TB recurrence or noncompliance with medication or adverse drug reactions.The investigation conforms with the principles outlined in the Declaration of Helsinki.In accordance with the ethical standards of our hospital's institutional review board, the Committee for the Protection of Human Subjects, and French law, written informed consent was not needed for demographic, physiologic, and hospital outcome data analyses because this observational study did not modify existing diagnostic or therapeutic strategies; however, patients were informed of their inclusion in the study.

Systematic Review
Search Strategy and Selection Criteria.This study was performed in accordance with the PRISMA statement (Preferred Reporting Items for Systematic Reviews and Meta-analyses) [16].Electronic databases were searched August 2023, including MEDLINE via the OVID platform (1946-present), Embase (1980-present), Cochrane, and PubMed.The search strategies are detailed in Supplementary Table 1.Studies were screened by 2 independent reviewers in separate databases.We included all studies about TB-IRIS and anti-TNFα treatment.We focused on studies written in English or French.After the removal of the duplicates, 2 reviewers independently screened titles and abstracts to obtain relevant articles for fulltext analysis.Eligible articles were then selected for the review.Any disagreement was resolved by discussion with a third reviewer.Ethical approval was not required.
Screening for LTBI prior to anti-TNFα initiation was negative in 20 (67%) of 30 patients; among the other 10 patients, 6 (21%) were treated for LTBI, 1 (3%) for pulmonary TB, and 3 (8%) had no screening.Among the patients with negative LTBI screening, 14 (70%) were treated with azathioprine prior to the screening.Out of 36 patients, 19 (53%) came from a TB-endemic country; 3 (8%) had been exposed to a patient with TB; and 1 (3%) had a personal history of TB.The 6 patients treated for LTBI reported TB after 20 months (IQR, 11.75-56.25)-1 of them only 2 months after treatment of latent TB, raising the suspicion of poor treatment compliance.Among the 5 others, 4 (80%) had been reexposed to the disease by contact or travel to endemic countries.

Systematic Literature Review
Studies Characteristics.After removal of duplicates, 269 records were screened, and 64 were assessed for eligibility.After verification of inclusion criteria, 61 studies were selected for the systematic review (see flow diagram, Figure 1).Studies are listed in Supplementary Table 2. Most studies were case reports (n = 44) or case series (n = 11) reporting either cases of IRIS in patients treated for TB who were taking anti-TNFα agents for inflammatory/autoimmune disease or TB-IRIS cases treated with anti-TNFα.Others were articles mentioning the risk of IRIS in patients being treated with anti-TNFα.
IRIS in Anti-TNFα TB Cases.Sixty-seven patients were studied (65% male; median age, 41 years [IQR, 28-56]): 52 from the literature and 15 from the current study.The main characteristics are listed in Table 3.All patients were treated with anti-TNFα at the time of TB diagnosis: 42% for IBD, 20% for rheumatoid arthritis, 17% for spondyloarthropathy, 12% for psoriasis.TB occurred after a median 9 months (IQR, 2-36) of anti-TNFα treatment and was disseminated in most cases (85%).Thirty-six (54%) patients had miliary TB, and 8 (12%) had central nervous system TB.Anti-TNFα was discontinued in all patients at TB diagnosis, and all patients received ATT.The median time between TB diagnosis and IRIS was 42 days (IQR, 21-90).Two patients required intensive care unit admission and orotracheal intubation for TB-IRIS with acute respiratory distress.Patients presented with fever (45%), worsening or new onset of lymphadenopathy (43%), and worsening of pulmonary lesions (34%).Fourteen patients (21%) had neurologic IRIS.Of note, 2 had acute kidney injury attributed to IRIS.
Twenty patients (30%) received symptomatic treatment only (paracetamol, nonsteroidal anti-inflammatory) or no treatment, with a favorable outcome.Thirty-nine (58%) patients were treated with corticosteroids and 16% with anti-TNFα.Most patients (87%) recovered without sequelae.A 75-year-old female patient died.She had disseminated TB with miliary involvement.Three weeks after starting ATT, she presented with respiratory distress due to worsening of pulmonary lesions and was intubated.Although a paradoxical reaction was diagnosed (with very high serum TNF levels), she did not receive specific treatment, and her pulmonary lesions progressed to fibrosis and death [17].TB relapse occurred in 2 cases, 1 of which was previously described in our cohort.The second had enlargement of cervical adenopathy attributed to IRIS and was treated surgically.He received ATT for 9 months, and despite a complete recovery and no resumption of anti-TNF, he relapsed 10 months later with abdominal TB [12].
IRIS occurred at a median 36 days (IQR, 21-63) after TB diagnosis.Most patients with IRIS (64%) had neurologic tropism.Six patients had worsening of pulmonary lesions.Three patients (7%) had worsening or new-onset spondylodiscitis, and 2 (4%) had psoas abscess.Most patients (93.3%) were treated with corticosteroids in the first instance, and all received anti-TNFα treatment (87% infliximab and 20% adalimumab) for IRIS for corticosteroid failure, due to relapse during corticosteroid tapering, or as corticosteroid-sparing treatment.
Twenty-eight patients (62%) recovered without sequelae and 33% with sequelae.One patient with central nervous system TB and IRIS did not improve despite corticosteroids and anti-TNF   and remained in a deep coma; after 4 months of treatment, palliative care was decided upon and she died few days after [18].Psoas abscess persisted despite anti-TNFα treatment in 1 patient.Manesh et al recently performed a matched retrospective cohort study to evaluate neurologic improvement in patients with central nervous system TB receiving corticosteroids alone or anti-TNF.Forty of 90 patients had a paradoxical reaction, 19 of whom were treated with infliximab.Overall, infliximab treatment was the only parameter associated with disability-free survival at 6 months [15].

DISCUSSION
In this study, we describe 36 cases of TB in patients receiving anti-TNFα to treat IBD in Ile-de-France, the region with the highest prevalence of TB in metropolitan France [19].We confirmed that in patients treated with anti-TNFα, TB cases are characterized by few isolated pulmonary forms but rather disseminated forms, as noted in several previous studies [2,4].A notable result is the occurrence of IRIS in our patients, with 15 of 32 experiencing it.This number is higher than the 7% reported in the RATIO cohort [11], in a different study population with more rheumatologic diseases, raising the question of a potentially higher proportion of IRIS in IBD or granulomatous diseases (eg, Crohn disease).Only 1 patient had sarcoidosis in our literature review, but anti-TNFα prescription is less common in this indication.However, in the Abitbol et al study on TB exclusively in patients with IBD, only 1 case of IRIS was reported out of 44 patients [20].The only clearly identified risk factor in univariate analysis for IRIS was the presence of miliary TB, with an odds ratio of 7.33.Rivoisy et al also noted the prescription of corticosteroids for TB and disseminated forms [11].We did not find these results.Yet, the majority of patients in our cohort had a disseminated form of TB at diagnosis, and very few received corticosteroids.
We aimed to study IRIS under anti-TNFα more precisely and conducted a systematic literature review, which included 52 cases in addition to the 15 cases in our cohort.This review indicates that IRIS appears to be prevalent in this population, with presentations of varying severity, ranging from increased lymphadenopathy to marked general symptoms with exacerbation of pulmonary lesions potentially leading to respiratory failure.In cases with mild symptoms, given the spontaneously favorable course, symptomatic treatment only seems justified [21].Corticosteroids appear effective in cases that require treatment in a majority of situations, although there is a risk of relapse upon tapering or corticosteroid dependency.In cases of corticosteroid therapy failure or the most severe cases, resuming anti-TNFα treatment could be an interesting option.In the literature, 40 TB-IRIS cases (plus 5 in our cohort) were treated with anti-TNFα, with favorable outcomes in most cases.Interestingly, the majority of patients had severe neurologic forms of IRIS.A recent Indian cohort study confirmed the benefits of anti-TNFα in neurologic IRIS [15]; in fact, the prescription of infliximab was the only protective factor in the multivariate analysis for neurologic prognosis at 6 months.No adverse effects related to anti-TNFα prescribing have been reported in our patients or in the literature.The role of TNFα in granuloma formation [22] and the studies showing that T lymphocytes produce TNFα in quantity during IRIS [23] provide pathophysiologic arguments for the use of these molecules in these indications (Figure 2).None of the patients in our cohort who had resumed anti-TNFα for digestive disease had IRIS after restarting treatment.
For patients in whom TB occurs under anti-TNF treatment, the fear is TB relapse.In our cohort, anti-TNFα was resumed in 21 cases and relatively quickly (median, 4 months), with only 1 case of TB relapse after anti-TNF resumption.Other studies had shown that it seemed reasonable to resume such treatment without the risk of relapse [20,24], but the resumption was later than in our study (median, 12 and 11 months, respectively).A recent systematic literature review assessed the risk of TB relapse in patients who started or restarted TNF antagonists of JAK inhibitors.Of 368 patients with TB who initiated or reinitiated anti-TNF, only 14 (3.8%) relapsed after a median 8.5 months following the initiation/reinitiation [25].Thus, it seems reasonable to restart treatment as soon as the bacillary load has been reduced by anti-TB treatment, either in the case of IBD activity requiring anti-TNF or in the case of symptomatic or severe IRIS.
Many patients in our study were originally from TB-endemic countries or had been exposed and developed active TB despite treated LTBI or negative screening.This raises the question of reexposure and how to proceed.Once anti-TNF treatment is started or after the treatment of initial LTBI, interferon γ release assays are no longer interpretable [7,26].It seems necessary to be particularly vigilant with these patients, providing information and preventive measures regarding the risk of TB and maintaining a perpetual concern about the occurrence of active TB.
Finally, our study has limitations, mainly due to its retrospective nature and the small number of patients, which makes generalization difficult.Because the study was retrospective and observational, we may have failed to collect important data, and some bias may have occurred.Because follow-up was not standardized, we may have missed late relapses, although the median follow-up after anti-TNFα resumption was 27 months (IQR, 17-82).However, the cases from the systematic literature review strongly support our results, even if there is a publication bias that cannot be separated from case reports.Randomized studies will be needed to draw definitive conclusions regarding the effectiveness and safety of anti-TNF treatment in TB-IRIS.
Overall, this work confirms that the presentation of TB in patients with IBD who are treated with anti-TNFα is often disseminated.It shows that IRIS is a common complication of varying severity.Anti-TNFα reintroduction appears to be a safe and effective strategy for the treatment of IRIS and IBD.
TB and IRIS in Patients With IBD and Anti-TNFα Treatment • OFID • 1
Statistical Analysis.Quantitative data were expressed as median (IQR) according to their nonnormal distribution.Quantitative data were compared by a nonparametric Mann-Whitney test according to their distribution.Categorical data were compared by a Fisher exact test.Factors associated with IRIS were analyzed by univariate logistic regression.The tests were 2-tailed, and P < .05 was considered statistically significant.Statistical analysis was performed with R version 4.2.2 (R Foundation for Statistical Computing).The database is registered at the Commission Nationale de l'Informatique et des Libertés (2225933) and was approved by the Ethics Committee of the French Infectiology Society (CER-MIT 2023-0904).

Table 2 . Demographic and Clinical Characteristics of 17 Non-IRIS and 15 IRIS-TB Cases From the Current Cohort and Univariate Analysis for the Risk of IRIS
Bold indicates P < .05.Abbreviations: anti-TNFα, anti-tumor necrosis factor α; IBD, inflammatory bowel disease; IRIS, immune reconstitution inflammatory syndrome; TB, tuberculosis.