In contrast to the aforementioned clinical trial data, a nationwide, register-based, propensity score-matched cohort study from Denmark (2000–2012; final cohort n = 3086, with 1543 anti-TNF users and 1543 anti-TNF nonusers) demonstrated a 63% increase in the risk of serious infections (associated with hospitalization) for anti-TNF users within the first 3 months of treatment compared to anti-TNF nonusers, followed by a subsequent risk decline. Over a 1-year risk period, the HR decreased and was no longer significant [13]. Similarly, increased infection risk was detected in an analysis of prospective observational safety data from the Crohn’s Therapy, Resource, Evaluation, and Assessment Tool (TREAT) registry, evaluating 6273 patients with Crohn’s disease (3420 who received infliximab with a total of 17,712 person-years and 2853 who received conventional nonbiological medications only [other-treatments-only group] with a total of 13,251 person-years) over a mean follow-up of 5.2 years. This study found an increased risk for serious infection in patients treated with infliximab (HR 1.43, 95% CI 1.11–1.84, P = 0.006). Almost 90% of infliximab-treated patients received at least two infusions, a majority (81.5%) of whom were dosed at 5 mg/kg. There was no evidence to support that greater numbers of infliximab infusions or infliximab dose escalation (from 5 mg/kg to 10 mg/kg) influenced serious infection risk [8]. Multivariate regression analysis from the TREAT registry found that moderate-to-severe Crohn’s disease activity was the strongest significant predictor for serious infection (HR 2.24, 95% CI 1.57–3.19, P < 0.001), while isolated colonic Crohn’s disease involvement (compared to both ileum and colon involvement) appeared to be protective against serious infection (HR 0.73, 95% CI 0.54–1.00, P = 0.046) [8].

The overall safety profile of adalimumab in global clinical trials of Crohn’s disease (involving 3160 patients representing 3401.9 patient-years of drug exposure) was reported to be comparable to that of other anti-TNF agents used for the same indication. Adverse event rates were similar to those described in other studies of adalimumab used for alternate approved indications covering a clinical follow-up period over 10 years. The most frequently reported serious adverse event was serious infection, most commonly due to abscess (intra-abdominal and gastrointestinal related). The incidence of opportunistic infections including tuberculosis was low [27].

A systematic review and meta-analysis including three randomized controlled trials (conducted from drug inception to January 2015) comparing adalimumab with placebo for moderate-to-severely active ulcerative colitis reported no significant difference in adverse events (over 8 weeks, including infection and tuberculosis) or serious adverse events when comparing induction therapy with adalimumab (dosed 160/80 mg at weeks 0/2 and then 40 mg at weeks 4 and 6) versus placebo [28]. Adalimumab maintenance therapy (40 mg every other week) increased the risk of adverse events (over 1 year) compared with placebo (risk ratio 1.28, 95% CI 1.06–1.54) [28].

Safety data pertaining to infection risk with the use of certolizumab is limited. Three meta-analyses using randomized controlled trial data for certolizumab suggested that the risk of infection with long-term therapy was not clearly increased [29–31]. A pooled analysis showed that the incidence rate for serious infectious complications was higher in short-term studies of certolizumab treatment versus placebo, but the risk did not heighten with long-term certolizumab therapy (up to 7 years) [31].

Golimumab safety evaluated in the PURSUIT trials revealed that adverse events in golimumab treatment groups appeared similar to those observed with other anti-TNF agents and with golimumab used for other approved indications [32, 33]. Results from the PURSUIT-SC induction study found the overall incidence of adverse events through week 6 was similar for golimumab- and placebo-treated patients, with serious infection reported in 0.5% versus 1.8%, respectively [32]. Overall rates of infections, serious infections, and infections warranting antimicrobial therapy per 100 patient-years of treatment did not increase with continued golimumab exposure [33, 34].

Natalizumab , a humanized monoclonal antibody against alpha-4 integrin, received initial FDA approval for use in multiple sclerosis but was temporarily withdrawn from the market in 2005 due to the risk of progressive multifocal leukoencephalopathy (PML), a serious opportunistic infection of the central nervous system caused by reactivation of the John Cunningham (JC) virus in chronically treated patients. One case of PML was reported in a Crohn’s disease patient on combination therapy with azathioprine [38]. In 2008, natalizumab was reapproved in the USA under a specialized distribution program (TOUCH, Tysabri Outreach: Unified Commitment to Health) and FDA approved for the treatment of active Crohn’s disease. Natalizumab-associated risk of PML has been most dramatically demonstrated in the multiple sclerosis literature. Among 99,571 multiple sclerosis patients treated with natalizumab (representing 209,123 patient-years), there were 212 reported PML cases (2.1 cases per 1000 patients); 22% of the affected patients died [39]. The risk of PML was lowest among patients who tested negative for anti-JC virus antibodies (estimated incidence 0.09 cases per 1000 patients, 95% CI 0–48). The highest estimated risk was seen among patients with the following factors (alone or in combination): positive anti-JC virus antibody status, immunosuppressant use prior to natalizumab initiation, and increasing duration of therapy (with greatest risk at 25–48 months and very few infections under 12 months). In the highest-risk subgroup of patients with all three risk factors, the estimated incidence was 11.1 cases per 1000 patients (95% CI 8.3–14.5) [39]. Natalizumab is generally prescribed with reservation due to this risk profile, and the availability of vedolizumab has further limited natalizumab use in IBD.

The advent of vedolizumab presented a favorable alternative to natalizumab as a gut-selective anti-alpha-4 beta-7 integrin agent. Integrated safety data from six trials of vedolizumab used in Crohn’s disease and ulcerative colitis (2380 patients with 4811 person-years of vedolizumab exposure) found no associated increased risk of infection or serious infection, reinforcing the presumed gut specificity of the therapy [40]. Systemic infections may still be a concern, however, with gastrointestinal infections as a potential risk. Serious infections including clostridial infections, sepsis, and tuberculosis were rarely reported in ≤0.6%. Independent risk factors for serious infection were corticosteroid use, narcotic analgesic use and younger age in Crohn’s disease, and narcotic analgesic use and prior anti-TNF failure in ulcerative colitis [40]. A retrospective cohort study assessing vedolizumab safety for moderate-to-severe Crohn’s disease from seven medical centers (May 2014–December 2015) reported 21 serious infections (requiring antibiotics or resulting in discontinuation of vedolizumab, hospitalization, or death) [41]. There have been no associated reports of PML.

Safety data of ustekinumab for induction of remission in Crohn’s disease indicated no significant difference in adverse events or serious adverse events when comparing ustekinumab to placebo; based on limited data, the assessment of rare adverse events could not be determined [42]. A multicenter, double-blind, placebo-controlled phase 3 study of ustekinumab for the treatment of moderate-to-severely active Crohn’s disease refractory to anti-TNF therapy (741 patients, 51% of whom had previously failed two or more anti-TNFs) reported similar proportions of patients with infections in ustekinumab versus placebo groups. Tuberculosis was not reported to have occurred in ustekinumab-treated patients through week 20 [43].

The risk of tuberculosis is increased with anti-TNF agents. Infection typically presents within the first few months of initiating anti-TNF therapy but may occur up to 2–3 years later or even following treatment for tuberculosis. Although pulmonary infections are classic, atypical sites can be involved [50, 51].

Detection of latent tuberculosis infection or active disease among patients receiving anti-TNF therapy became an issue of notable importance after the US FDA Adverse Events Reporting System found higher tuberculosis rates among patients exposed to infliximab compared to background population rates [52]. Most patients (56%) had extrapulmonary tuberculosis, and 24% had disseminated disease. Not only did the frequency of tuberculosis infection appear increased compared to other opportunistic infections reported in association with infliximab but also 64/70 cases (91%) manifested in countries with a low incidence of tuberculosis suggesting disease reactivation [52]. The risk of tuberculosis has been confirmed in other studies of TNF-alpha antagonist exposure, particularly with the use of monoclonal antibodies [53, 54].

Not only is the risk for reactivation of latent tuberculosis increased among anti-TNF-treated patients but the disease may also be more severe than in the general population [1]. Active tuberculosis can present in IBD patients undergoing anti-TNF therapy despite negative screening tests for latent tuberculosis and can also be seen in those who have completed tuberculosis treatment or received latent tuberculosis prophylaxis [55]. A recent retrospective study conducted at GETAID centers investigated all IBD patients undergoing anti-TNF therapy who developed tuberculosis despite negative screening tests. Among 44 patients identified, the median interval from initiation of anti-TNF therapy to diagnosis of tuberculosis was 14.5 months (interquartile range 25–75, 4.9–43.3). Tuberculosis involvement included pulmonary site in 57% with extrapulmonary involvement in 91%. Tuberculosis exposure was thought to be implicated in 32% of the cases. Anti-TNF therapy was re-initiated in 27 patients approximately 11.2 months (interquartile range 25–75: 4.4–15.2) following tuberculosis diagnosis, and infection recurrence was not detected [56].

Generally, standard and complete treatment for latent tuberculosis infection (such as isoniazid for 6–9 months) [57] should be commenced prior to anti-TNF therapy, which should not be initiated until at least 3–4 weeks after introduction of the antituberculosis agent(s). Anti-TNF therapy should be stopped if active tuberculosis is detected and may be resumed after approximately 2 months of antituberculosis therapy [1, 57]. Restarting anti-TNF therapy following adequate treatment for tuberculosis appears safe [56].

Immunosuppression is a predisposing factor for the development of Pneumocystis jiroveci pneumoni a , previously known as Pneumocystis carinii pneumonia (PCP) . Pneumocystis pneumonia (PCP) infection appears increased among IBD patients, particularly in association with combination immunosuppressive therapy including infliximab [79–82]. The mean time from infliximab infusion to pneumonia symptom onset was 21 ± 18 days (n = 40), and patients had an average of 2.1 ± 1.3 infusions (n = 76) prior to symptom development. The mortality rate was 27% [82].

Prophylactic treatment for PCP (such as trimethoprim/sulfamethoxazole) should be considered for patients on triple immunosuppression (i.e., corticosteroids, biologic, and immunomodulator therapy). Additional risk factors for the development of PCP that may necessitate prophylaxis include lymphopenia (total lymphocyte count <600 cells/mm) and age over 55 years [83]. Primary chemoprophylaxis is not recommended for fungal infections other than Pneumocystis jiroveci, and there are no vaccinations available for disease prevention [1].

Patients on anti-TNF therapy appear to be at heightened risk for Legionella pneumonia infection, particularly with combination immunosuppressant therapy and among elderly populations aged over 65 years [36]. The relative risk of L. pneumophila infection was increased in patients exposed to anti-TNF therapy (relative risk 16.5–21) compared with that in the overall population in France [84]. Cases of legionella have similarly been reported in association with anti-TNF therapy used for the treatment of IBD [6, 85–88]. In 2011, the FDA issued a boxed warning regarding the risk of Legionella for the TNF-alpha inhibitor class [36]. Immunosuppressant therapy should be held until the acute infection has resolved. Recurrent Legionella infection has also been reported and may influence reintroduction of immunosuppressant therapy [1, 89].

Patients on anti-TNF therapy appear to be at heightened risk for Listeria infection , particularly with combination immunosuppressant therapy and elderly populations aged over 65 years [36, 90]. Several cases of listeriosis have been reported among patients treated with anti-TNF therapy for IBD [91–96] and rheumatoid arthritis [97, 98]. In 2011, the FDA issued a boxed warning regarding the risk of Listeria for the TNF-alpha inhibitor class [36].

The risk of systemic and cutaneous nocardia infection has been recognized in association with anti-TNF therapy [99], particularly with concomitant corticosteroid therapy. A review of the literature (1980–2014) pertaining to nocardial infections among immunosuppressed IBD patients reported nine cases (six associated with anti-TNFs, two associated with prednisone plus thiopurine, one associated with cyclosporine).

CDI has become an epidemiologic phenomenon as a leading cause of hospital-associated gastrointestinal illness [100]. It is well established that patients with IBD, particularly those on chronic immunosuppressive therapy with certain agents such as corticosteroids, are at increased risk for the development of CDI [101]. Furthermore, IBD patients who develop CDI have increased risks for severe infection, gastrointestinal surgery, and greater hospital length of stay compared to IBD patients without CDI along with increased inhospital mortality compared to C. difficile-infected patients without underlying IBD and IBD patients without CDI [100, 102, 103].

CDI should be excluded (or empirically treated in some cases) prior to initiation of biologic or other immunosuppressant therapy. No meaningful association linking infliximab with serious bacterial infections including CDI was seen in a large retrospective cohort study involving 10,662 patients with IBD, while corticosteroid therapy was associated with an over threefold increased relative risk for CDI (RR 3.4, 95% CI 1.9–6.1) compared with other immunosuppressants [101]. A subsequent retrospective cohort study of 503 patients with CDI identified IBD patients as 33% more likely than the general population to experience recurrent infection. Among this IBD cohort (n = 110), patients with recurrent CDI were significantly more likely than those without recurrent CDI to have reported exposure to biologic therapy (48.6 versus 40.0%, P < 0.01). Infliximab use (compared to nonuse) significantly elevated the risk of recurrent CDI (34.3% versus 17.3%, respectively, P < 0.01), while adalimumab use did not. Steroid therapy, recent antibiotic exposure, and 5-aminosalicylic acid use also significantly increased the risk for recurrent CDI, while immunomodulators (azathioprine, methotrexate, and cyclosporine) did not appear to influence this risk [5]. Treatment with two or three immunomodulators increased the risk, independent of disease severity at presentation [104]. Chemoprophylaxis for CDI is not recommended [1].

An increased risk of Streptococcal pneumoniae has been established in association with anti-TNF agents, as demonstrated in several large studies in Denmark and in the USA [6, 105]. The risk of invasive pneumococcal disease appears increased among IBD patients compared to controls, not only following but also in years prior to IBD diagnosis [6]. The risk of invasive pneumococcal disease in IBD versus control groups was increased twofold for Crohn’s disease and 1.5-fold for ulcerative colitis; this risk was greatest during the first year after IBD diagnosis and decreased 2–4 years after IBD diagnosis. Exposure to anti-TNF agents did not influence the risk of invasive pneumococcal disease in the IBD population (nor did exposure to oral or topical corticosteroids or 5-aminosalicylates/sulfasalazine) [6]. Anti-TNF treat ment, either alone or in combination with immunomodulator therapy, has been associated with diminished antibody response to pneumococcal vaccination [106, 107].

Immunosuppressed patients may be at increased risk for developing complications related to influenza infection [1, 108]. Influenza virus infection may be severe or fatal and may be complexed by secondary bacterial infection(s). Additional risk factors for influenza-related mortality include extremes of age (young and elderly) as well as medical comorbidities [109]. Inactivated trivalent influenza vaccine is recommended for patients undergoing immunosuppressant therapy. Lower immune response rates to vaccination [110–112] and persistence of seroprotection have been detected in IBD populations, particularly in association with anti-TNF treatment [113, 114] or combination immunosuppressant therapy [111, 112, 115]. Timing of influenza vaccination relative to infliximab dosing in pediatric and adult patients receiving maintenance IBD therapy does not appear to influence immune response [116]. Influenza vaccination appears safe and well-tolerated among IBD patients [117] and does not appear to be associated with disease flare [111, 113, 117, 118].

The prevalence of hepatitis B virus (HBV) infection among patients with IBD appears similar to that of the general population in some studies [119–121] and increased in IBD patients compared to non-IBD patients in others [122, 123]. Hepatitis B reactivation is an important concern among immunosuppressed populations, widely reported among patients undergoing cytotoxic chemotherapy (particularly for hematologic malignancies) and solid organ or stem cell transplantation and also reported in association with biologic treatments for autoimmune conditions and IBD [124]. Immunosuppressive treatment (e.g., with TNF inhibition) can reduce viral clearance, exhaust HBV-specific T-cell responses, and enhance viral load, leading to immune-mediated liver damage particularly after immunosuppression is withdrawn [125–127]. Reactivation of viral replication and flares of HBV thus reflect immune reconstitution and can occur even after short courses of immunosuppression. Occurrence of this preventable consequence has been associated with significant morbidity and mortality that may be mitigated by the use of prophylactic antiviral therapy among at-risk patients [1, 128].