Pharmacoepidemiology

There are several challenges in determining the infectious risk of any given therapy. Briefly, the determination of whether a particular adverse event is related to a specific treatment is difficult if the event in question occurs infrequently. Serious and opportunistic infections in patients with IBD fall into this category. In addition, in patients with IBD, the underlying disease process can complicate interpretation because the disease itself may increase the risk of developing an adverse event.

The strongest evidence that an adverse event is related to a particular therapy is best demonstrated through randomized controlled trials. These randomized controlled trials are not powered to assess adverse events, however, especially those that occur rarely or over a longer period of time. Large observational cohort data or registry data can help to detect infrequent events, but they are neither randomized nor controlled and can be subject to bias. Case-control studies are an effective tool to study the risk of adverse events, but the identification of proper controls can be difficult and time-consuming.

Corticosteroids

Population-based studies have shown that 42% of patients with Crohn disease are on CS within the first 3 years of diagnosis, and that 39% to 56% are on CS at some point during the course of their disease. The absolute infectious risk of CS in IBD is difficult to quantify, but there is significant evidence that they increase the risk of infection.

A meta-analysis of 71 controlled trials assessed the association between CS and subsequent infections. The overall rate of infectious complications was 12.7% in the CS group and 8% in the controlled group (relative risk [RR], 1.6; 95% confidence interval [CI], 1.3–1.9). In those receiving CS, there was a significantly increased risk of death (RR, 2.6; 95% CI, 1.2–5.3). The risk of infections in the intestinal disease subgroup was lower but still significant (RR, 1.4; 95% CI, 1.1–1.7). The rate was not increased in those patients receiving less than 10 mg/day of prednisone or those who had a cumulative dose of less than 700 mg.

Increased risks of serious infection and mortality secondary to CS use were demonstrated specifically in patients with Crohn disease through the Crohn Therapy, Resource, Evaluation, and Assessment Tool (TREAT) registry. This registry was initially created to monitor the toxicity of therapy with infliximab, and is an ongoing, prospective, observational, multicenter registry of North American patients with Crohn disease. There are approximately 6000 patients in the registry, half of whom are on other therapies aside from infliximab. Analysis of the TREAT registry demonstrated that in multivariate logistical analysis, CS were associated with an increased mortality risk (odds ratio [OR], 2.10; 95% CI, 1.15–3.83). Out of the 2142 patients on CS, 2.85% developed serious infections compared with 1.09% of patients not on CS. Serious infection risk was independently associated with CS use (OR, 2.21; 95% CI, 1.46–3.34), but not with infliximab treatment. Although this estimate of increased risk seems quite robust, it is important to note that this registry had a dropout rate of 20%, and that the reporting of adverse events was based primarily on the review of gastroenterologists’ medical records, raising concerns about the completeness of safety data.

Several studies have characterized the risk of CS in specific clinical circumstances. The European Cooperative Crohn Disease Study demonstrated that 3 out of the 43 patients who presented with an abdominal (phlegmonous) mass and were treated with CS died from sepsis. A retrospective case-control study examined the effect of CS on intra-abdominal and pelvic abscess formation. Systemic CS therapy was associated with an increased rate of intra-abdominal and pelvic abscesses in patients with perforating disease (OR, 9.03; 95% CI, 2.40–33.98) and in patients with nonpenetrating disease (OR, 9.31; 95% CI, 1.03–83.91). There was a trend toward an increased rate of abscess formation in patients receiving more than 20 mg per day of prednisone (OR, 2.81; 95% CI, 0.99–7.99).

Combined, all of these findings demonstrate the increased infectious risk with CS use and support the recommendation that patients who require frequent CS to control their disease should be considered for steroid-sparing therapy with immunomodulators or biologic agents.

Corticosteroids

Population-based studies have shown that 42% of patients with Crohn disease are on CS within the first 3 years of diagnosis, and that 39% to 56% are on CS at some point during the course of their disease. The absolute infectious risk of CS in IBD is difficult to quantify, but there is significant evidence that they increase the risk of infection.

A meta-analysis of 71 controlled trials assessed the association between CS and subsequent infections. The overall rate of infectious complications was 12.7% in the CS group and 8% in the controlled group (relative risk [RR], 1.6; 95% confidence interval [CI], 1.3–1.9). In those receiving CS, there was a significantly increased risk of death (RR, 2.6; 95% CI, 1.2–5.3). The risk of infections in the intestinal disease subgroup was lower but still significant (RR, 1.4; 95% CI, 1.1–1.7). The rate was not increased in those patients receiving less than 10 mg/day of prednisone or those who had a cumulative dose of less than 700 mg.

Increased risks of serious infection and mortality secondary to CS use were demonstrated specifically in patients with Crohn disease through the Crohn Therapy, Resource, Evaluation, and Assessment Tool (TREAT) registry. This registry was initially created to monitor the toxicity of therapy with infliximab, and is an ongoing, prospective, observational, multicenter registry of North American patients with Crohn disease. There are approximately 6000 patients in the registry, half of whom are on other therapies aside from infliximab. Analysis of the TREAT registry demonstrated that in multivariate logistical analysis, CS were associated with an increased mortality risk (odds ratio [OR], 2.10; 95% CI, 1.15–3.83). Out of the 2142 patients on CS, 2.85% developed serious infections compared with 1.09% of patients not on CS. Serious infection risk was independently associated with CS use (OR, 2.21; 95% CI, 1.46–3.34), but not with infliximab treatment. Although this estimate of increased risk seems quite robust, it is important to note that this registry had a dropout rate of 20%, and that the reporting of adverse events was based primarily on the review of gastroenterologists’ medical records, raising concerns about the completeness of safety data.

Several studies have characterized the risk of CS in specific clinical circumstances. The European Cooperative Crohn Disease Study demonstrated that 3 out of the 43 patients who presented with an abdominal (phlegmonous) mass and were treated with CS died from sepsis. A retrospective case-control study examined the effect of CS on intra-abdominal and pelvic abscess formation. Systemic CS therapy was associated with an increased rate of intra-abdominal and pelvic abscesses in patients with perforating disease (OR, 9.03; 95% CI, 2.40–33.98) and in patients with nonpenetrating disease (OR, 9.31; 95% CI, 1.03–83.91). There was a trend toward an increased rate of abscess formation in patients receiving more than 20 mg per day of prednisone (OR, 2.81; 95% CI, 0.99–7.99).

Combined, all of these findings demonstrate the increased infectious risk with CS use and support the recommendation that patients who require frequent CS to control their disease should be considered for steroid-sparing therapy with immunomodulators or biologic agents.

Immunomodulator therapy

Anti–TNF-α therapy

At present there are three different anti–TNF-α agents approved for the treatment of Crohn disease: (1) infliximab (Remicade, Centocor, Horsham, Pennsylvania); (2) adalimumab (Humira, Abbott Laboratories, North Chicago, Illinois); and (3) certolizumab pegol (Cimzia, UCB, Smyrna, Georgia). Infliximab is the only anti–TNF-α agent approved for use in the treatment of UC. There has been increasing use of these agents over the last decade, and multiple studies have examined the risk of infectious complications.

There is definite evidence that anti–TNF-α agents are associated with an increased rate of serious infections in patients with RA. A meta-analysis of nine placebo-controlled trials of infliximab and adalimumab in RA involving over 5000 patients yielded a pooled OR for serious infection of 2.01 (95% CI, 1.31–3.09).

In terms of patients with IBD, the association between anti-TNF therapy and serious infection is not as clear and may relate to the fact that the patients with IBD are generally younger and have less comorbid disease, compared with patients with RA. Individual randomized trials in Crohn disease have not demonstrated an elevated risk of serious infection. In a recently published meta-analysis, 21 trials with a total of 5356 patients were reviewed to assess the safety of anti–TNF-α agents. There were 3341 patients in the anti–TNF-α group and 2015 patients in the control group, with a median follow-up of 24 weeks (range, 4–60 weeks). There was no significant difference in the risk of serious infection in the anti-TNF group versus the control group (2.09% versus 2.13%, respectively; 95% CI, 0.45–0.65). In subgroup analysis, there were no differences in infectious risk between anti–TNF-α and control groups whether they were induction or maintenance trials.

Data from the TREAT registry assessed the infectious complications of IBD therapeutic medications. The risk of serious infection 3 months after an infliximab infusion was compared with the risk of infection 3 months before an infusion. In univariate analysis, serious infections were more than twice as likely to occur in the 3 months after an infliximab infusion (OR, 2.15; 95 CI, 1.44–3.21). After controlling for other factors, however, such as disease duration and severity, concurrent use of CS, immunomodulators, and narcotic analgesics, the association between infliximab and serious infection was no longer significant (OR, 0.99; 95% CI, 0.64–1.53). An update of the TREAT registry, now with 24,575 patient years of follow-up, was recently presented in abstract form. Infliximab-treated patients had an increased risk of serious infections (RR, 1.57; 95% CI, 1.16–2.14); however, multivariate analysis again suggested that this increased risk was independently associated with prednisone and not infliximab.

Several retrospective referral center–based studies have assessed the risk of anti–TNF-α agents in patients with IBD. In one study, among the first 500 Crohn disease patients treated with infliximab at Mayo Clinic, serious infections occurred in 20 (4%), including four deaths caused by infection (0.8%). In a study of 202 patients with UC and Crohn disease who had a median follow-up of 2.4 years (range, 1–4.9; 620 patient-years), 42 patients (20.8%) experienced an infectious event including 22 patients (10.9%) with a serious infection. The safety of more than 3000 patients treated with adalimumab in clinical trials was recently assessed. The incidence of opportunistic infections was 1.8%, with oral candidiasis being the most commonly reported infection (1.2%). Serious infections were seen 5.8% of patients (incidence rate of 6.6 cases per 100 person-years) but almost half of those were intra-abdominal (2.5%). It is important to note that these three studies had no control group and did not take into account concomitant medications and so direct causality cannot be established.

More recently, in a single-center experience from Leuven, Belgium, the long-term safety of anti–TNF-α agents was assessed in 1400 patients with Crohn disease (of whom 734 received infliximab). Patients who received anti-TNF therapy were followed for a median of 58 months (interquartile ratio, 33–88 months). The percentage of patients who developed serious infections was 6% in the infliximab group and 9% in the control group. The incidence rate of serious infections in the two groups was not significantly different (1.6 per 100 patient-years in infliximab group versus 1.1 per 100 patient-years in control group).

Two population-based studies have reported on the safety of infliximab. The first, a population-based cohort from Stockholm County, examined 217 patients who received infliximab between 1999 and 2001. Eighteen patients (8.3%) developed severe infectious complications (seven were postoperative) and there were two deaths related to infection (both on CS, none on AZA and 6MP). In this cohort, 51% of the patients were on concomitant immunomodulators, 54% were being treated with CS, and 25% were on both immunomodulators and CS. Given that both of these agents increase infectious risk, the exact role of infliximab with regards to infection can not be determined from this study. The second population-based cohort examined the association between infliximab and other immunosuppressive agents in a cohort of 10,662 IBD patients in British Columbia, Canada. In patients receiving infliximab, the event rate of serious bacterial infection was 4.28 per 1000 patient-years (95% CI, 0.11–23.8) but this was not significantly increased compared with patients receiving CS or immunomodulators.

The risk of infectious complications with anti–TNF-α monotherapy in patients with Crohn disease does not seem to be increased in terms of overall infections. These studies may not be large enough, however, to detect rarer opportunistic infections (see later sections on TB and fungal infections). In a referral center–based case-control study of risk factors for opportunistic infections in patients with IBD, infliximab was associated with a fourfold increased risk of opportunistic infection (OR, 4.4; 95% CI, 1.2–17.1).

Combination therapy

A recent randomized controlled trial published in abstract form demonstrated that in patients with moderate to severe Crohn disease, combination therapy with AZA and infliximab was superior in achieving clinical and endoscopic remission compared with either drug alone. Other studies have shown that the combination therapy reduces the immunogenicity of anti–TNF-α agents, especially when the latter are administered episodically. Although combination therapy with biologic therapy and concomitant immunomodulators may be more effective, this increased efficacy may come at the cost of increased infectious complications.

A randomized trial of infliximab with concomitant MTX in more than 1000 RA patients was designed specifically to assess safety end points. Patients in the study were randomized to placebo infusion to week 22, then 3 mg/kg every 8 weeks to week 54; 3 mg/kg infliximab infusions to week 22 with escalation by 1.5 mg/kg as needed to control symptoms; and lastly to 10 mg/kg infliximab infusions for the length of the study. There was greater than a threefold increased risk of serious infections in the 10 mg/kg group at week 22 compared with those patients who received MTX alone ( P = 0.013), but not in those who received 3 mg/kg on infliximab.

In patients with IBD, the evidence of increased risk with concomitant therapy is unclear. In one study, data from four prospective randomized phase 3 trials in IBD patients were used to assess the efficacy of concomitant immunomodulators with infliximab. In the pooled safety analysis of 1007 infliximab-treated patients, similar proportions of patients with and without combination therapy had infections (44.1% and 44.5) or serious infections (3.7% and 3.2%). In the earlier referenced population-based study out of British Columbia, an increased risk of serious infections was not found in patients receiving concomitant therapy with infliximab and an immunomodulator. In the study by Fidder and colleagues, only the concomitant use of steroids (and not immunomodulators) was associated with an increased risk of serious infection (OR, 2.7; 95% CI, 1.2–6.1). Throwing caution to these results, the case-control study by Toruner and colleagues demonstrated the OR for opportunistic infections in patients on two or more immunosuppressive medications simultaneously was 14.5 (95% CI, 4.9–43) and was “infinite” in patients with three immunosuppressive medications (CS, immunomodulators, and anti–TNF-α agent). Further examination of this study suggests, however, that much of the increased risk was driven by the use of CS.

Although concomitant therapy may increase the risk of infection, it seems that it is the use of that CS plays the largest role in the development of serious or opportunistic infections.

Cyclosporine

The use of cyclosporine in IBD is at present limited to rescue therapy for patients with fulminant UC. There are limited data on the safety of cyclosporine in this setting, and underlying disease severity complicates the assessment of safety.

In patients receiving cyclosporine for the prevention of liver allograft rejection, therapy was associated with a risk of infection of approximately 40%, sepsis in 20%, and cytomegalovirus in 15% to 25%. A study assessing the long-term outcome of treatment with intravenous cyclosporine in patients with severe UC demonstrated that in 86 patients treated with cyclosporine between 1992 and 2000, five (5.8%) patients had opportunistic infections, and three patients (3.5%) died (one from Pneumocystis jiroveci (carinii) pneumonia [PCP] and two from Aspergillus pneumonia). In addition, there were catheter-related infections in eight patients (9.3%). In a separate retrospective study, 19 patients with CS-refractory UC who received cyclosporine after infliximab or infliximab after cyclosporine were assessed. In total, three patients (15.8%) developed severe complications and one patient died (gram-negative bacterial sepsis with death, herpetic esophagitis, and pancreatitis with Enterococcus and Klebsiella bacteremia). Overall, these limited data suggest that treatment with cyclosporine leads to an increased risk of serious infections.

Natalizumab

Natalizumab (Tysabri, Biogen Idec, Cambridge, Massachusetts, and Elan Pharmaceuticals, South San Francisco, California), a humanized IgG4 monoclonal antibody against α ₄ integrin–mediated leukocyte migration, was approved by the FDA in 2008 for use in patients with Crohn disease with evidence of active inflammation who have had an inadequate response to or cannot tolerate anti-TNF therapy. In the ENACT-1 and ENACT-2 trials of natalizumab in Crohn disease, subjects in the treatment arm had a significantly increased risk of developing influenza or an influenza-like illness (12% versus 5%; P <.05). One case of varicella pneumonia and cytomegalovirus hepatitis was also reported in the natalizumab group. In two earlier trials with natalizumab in Crohn disease and in one smaller trial in patients with UC, an increased risk of serious infections was not found.

After publication of the ENACT trials, a fatal case of JC virus–induced progressive multifocal leukoencephalopathy was reported in one patient who had been enrolled in this trial. Two further cases were identified in patients receiving natalizumab for multiple sclerosis, and the drug was withdrawn temporarily from commercial use. Natalizumab was reintroduced under restrictive rules prohibiting concurrent therapy with an immunomodulator or long-term CS therapy. Since being reintroduced to the market, of the 52,000 patients who have taken the drug, there are, as of early June 2009, six additional cases of progressive multifocal leukoencephalopathy (four cases occurred in patients not on any concurrent therapy) resulting in an incidence of approximately 1.2 cases per 10,000 patients treated.