• Crohn’s disease
• inflammatory bowel disease
• infliximab
• lymphoma
• mortality
• ulcerative colitis

Infliximab, a monoclonal antibody to tumour necrosis factor (TNF) α, is an important advance in the treatment of Crohn’s disease.^1–5 The efficacy of infliximab for the treatment of ulcerative colitis is still unclear.^6,7 Infliximab was approved for the treatment of Crohn’s disease in 1998 based on a 12 week phase 2 trial in 108 patients¹ (followed by a 36 week extension trial)³ and a small phase 3 trial in 94 patients,² both of which showed compelling efficacy. Because of the possibility of open label crossover at week 4, 102 of 108 patients in the phase 2 trial received infliximab by week 12. Thus only six patients in the phase 2 study and 31 patients in the phase 3 study who received placebo were available for safety follow up without crossover to infliximab. Although not all patients enrolled in these studies were receiving concomitant corticosteroids and/or azathioprine or 6-mercaptopurine, regulatory approval in the USA was granted only in patients with moderate to severely active Crohn’s disease unresponsive to conventional therapy (not defined) and for fistulising Crohn’s disease; and in Europe in patients with moderate to severely active Crohn’s disease or fistulising Crohn’s disease unresponsive to a full and adequate course of corticosteroids and immunosuppressive therapy. These restrictions were, at least in part, a reflection of the uncertainties regarding safety.

Subsequently, two blinded, placebo controlled, phase 4 maintenance trials (designed as infliximab withdrawal trials) were conducted in 573 patients with moderate to severely active Crohn’s disease⁴ and in 306 patients with fistulising Crohn’s disease.⁵ All patients in these two maintenance studies initially received at least one induction dose of infliximab. Thus safety data in patients with Crohn’s disease treated only with placebo who were naïve to infliximab was available in only 34 of 1081 patients enrolled in placebo controlled trials. Potential treatment emergent safety issues in the other 1047 patients treated with infliximab in these clinical trials included non-Hodgkin’s lymphoma in 2/1047 (0.2%),^3,4 opportunistic infections in 3/1047 (0.3%) (tuberculosis n = 1; cytomegalovirus n = 1; cutaneous Nocardia n = 1),^4,5 serious infections which occurred at rates of 4% and 4.6% in the two large maintenance trials,^4,5 serum sickness-like reactions in 19/1047 (1.8%),^4,5 drug induced lupus in 2/1047 (0.2%),^3,4 and death in 4/1047 (0.4%) patients, including two deaths from lymphoma and a death from sepsis that were potentially related to infliximab^3–5 (one patient developed lymphoma during a maintenance trial and then died during follow up^4,8). Post-marketing safety reports or warnings from the US Food and Drug Administration have yielded additional information regarding the potential for opportunistic infections, including tuberculosis,⁹ disseminated histoplasmosis,¹⁰ coccidioidomycosis, listeriosis, and Pneumocystis carinii pneumonia,¹¹ and possibly non-Hodgkin’s lymphoma.¹² Many of these post-marketing events occurred in patients with rheumatoid arthritis who have a median age 20 years older than that of patients with Crohn’s disease.

In this issue of Gut, Ljung and colleagues⁸ report on the clinical benefit and toxicity associated with the use of infliximab in a population based cohort of 217 patients with inflammatory bowel disease (Crohn’s disease n = 191, ulcerative colitis n = 22, indeterminate colitis n = 4) in Stockholm County, Sweden [see page 849]. Patients received a mean of 2.6 infusions of infliximab (range 1–11). Fifty four per cent of patients were also receiving azathioprine or 6-mercaptopurine, 51% were receiving corticosteroids, and 25% were receiving both. The authors reported that the clinical benefit observed in patients with Crohn’s disease was comparable with that reported in the controlled trials referenced above and in other uncontrolled reports of infliximab in clinical practice.¹³ The authors also reported that severe adverse events occurred in 41 (18.9%) patients (Crohn’s disease n = 35, ulcerative colitis n = 6) including: non-Hodgkin’s lymphomas in 3/217 (1.4%) patients (Crohn’s disease n = 3) of whom two died; severe infections in 18/217 (8.3%) patients (Crohn’s disease n = 11, ulcerative colitis n = 5); serum sickness-like reaction in 5/217 (2.3%) (Crohn’s disease n = 5); and drug induced lupus in 1/217 (0.5%) (Crohn’s disease n = 1). The severe infections included opportunistic infection in 2/217 (0.9%) patients (listeriosis n = 1 in a patient with Crohn’s disease, Pneumocystis carinii n = 1 in a patient with ulcerative colitis who died) and fatal sepsis in 2/217 (0.9%) patients (Crohn’s disease n = 1, ulcerative colitis n = 1). In total, 6/217 (2.8%) patients died in 28 months for a crude annual mortality of 1.2%: 3/191 (1.6%) patients with Crohn’s disease died (non-Hodgkin’s lymphoma n = 2, sepsis n = 1); 3/22 (13.6%) patients with ulcerative colitis died (sepsis n = 1, Pneumocystis carinii n = 1, pulmonary embolus n = 1). All patients who died were receiving corticosteroids and most were elderly; at least two of the three patients who developed lymphoma were receiving or had previously received azathioprine.

Another study recently reported the safety experience in 500 consecutive patients with Crohn’s disease treated with infliximab at the Mayo Clinic.¹⁴ Patients received a median of three infusions and had a median follow up of 17 months. Forty three patients (8.6%) experienced a serious adverse event of which 30 (6%) were considered to be possibly related to infliximab. Serum sickness-like disease occurred in 19/500 patients and was attributed to infliximab in 14 (2.8%). Three patients (0.6%) developed drug induced lupus. One patient (0.2%) developed a new demyelination disorder. Forty eight patients had an infectious event of which 41 (8.2%) were attributed to infliximab. Twenty patients (0.4%) had a serious infection: two fatal sepsis, eight pneumonias of which two were fatal, six viral infections, two abdominal abscesses requiring surgery, one arm cellulitis, and one histoplasmosis (opportunistic infection). Nine patients had a malignant disorder, three of which were possibly related to infliximab, including one lymphoma (0.2%). A total of 10 deaths were observed over a median of 17 months, yielding a crude annual mortality of 1.3%. For five of these patients (1%), the events leading to death were possibly related to infliximab. Most of the patients who died were elderly.

These three data sets (controlled clinical trials, Ljung et al study,⁸ Colombel et al study¹⁴) show remarkable convergence for the frequency of the most important adverse events. Serious or severe infections occurred at a rate of 4.0–4.6% in clinical trials, 8.3% in the Ljung et al study, and 8.2% in the Colombel et al study. Opportunistic infection occurred at a rate of 0.3% in the clinical trials, 0.9% in the Ljung et al study, and 0.2% in the Colombel et al study. Serum sickness-like reactions occurred at a rate of 1.9% in the clinical trials, 2.3% in the Ljung et al study, and 2.8% in the Colombel et al study. Drug induced lupus occurred at a rate of 0.2% in the clinical trials, 0.5% in the Ljung et al study, and 0.6% in the Colombel et al study. Finally, death in patients with Crohn’s disease occurred at a crude annual rate of 0.4% in the clinical trials, 1.2% of patients in the Ljung et al study, and 1.3% of patients in the Colombel et al study. The mortality rate in these three data sets is comparable with what has previously been described in several studies of the natural history of Crohn’s disease.^15–17 Non-Hodgkin’s lymphoma occurred at a rate of 0.2% in the clinical trials and in the Colombel et al study, and at a rate of 1.4% in the Ljung et al study. Based on these results from clinical trials, a referral centre, and a population based cohort, we can conclude that patients with moderate to severely active Crohn’s disease treated with infliximab may have a small but real risk of developing severe infections, opportunistic infections, and non-Hodgkin’s lymphoma. However, it must be pointed out that all three data sets lack adequate controls, and one cannot be certain to what degree the potential bias of infliximab being given to the most refractory patients, and concomitant immunosuppressive therapy, may contribute to any possible risk.

Thus the important unanswered question is to what degree infliximab therapy caused or contributed to these serious adverse events and to the observed crude annual mortality rates? Population based studies in patients with Crohn’s disease have shown only a slightly increased mortality^15,18–22 (with three exceptions where no increased mortality was reported)^23–25 and no increased risk of non-Hodgkin’s lymphoma^26–29 (with one exception).³⁰ However, these population based studies have not provided mortality or lymphoma rates adjusted for patient age, disease severity, or concomitant therapy with corticosteroids and/or azathioprine or 6-mercaptopurine. Clinical trials and observational studies have reported that corticosteroids^31,32 and azathioprine³³ may result in abdominal abscess, sepsis, and death; and that azathioprine and 6-mercaptopurine may be associated with non-Hodgkin’s lymphoma.^34–37 Thus the data reported in the clinical trials and by Ljung et al and Colombel et al must be interpreted with caution. At the present time all that we can really say is that patients with moderately to severely active Crohn’s disease who are failing therapy with corticosteroids and/or immunosuppressive therapy and are subsequently selected for therapy with infliximab have a small but apparently real risk of serious infection, opportunistic infection, and possibly non-Hodgkin’s lymphoma. It is unclear if the crude annual mortality rate is increased or not. Whether these adverse outcomes are directly caused by or exacerbated by infliximab, or are the result of other important confounders such as age, severity of illness, concomitant therapy with corticosteroids, concomitant therapy with azathioprine or 6-mercaptopurine, and combination therapy with corticosteroids and azathioprine or 6-mercaptopurine, is impossible to determine from the available data. Preliminary data from one large registry study in 5000 patients has reported that the severity of illness and corticosteroids account for virtually all of the infections and any excess mortality in patients with Crohn’s disease treated with infliximab.³⁸

To definitively address these issues, population based studies of patients treated with infliximab adjusted for important confounders are needed. Patient registries containing comparator patients not treated with infliximab will also be useful in this regard. Placebo controlled trials with infliximab (without crossover of placebo patients to infliximab) in patients with mild to moderately active Crohn’s disease who are not receiving concomitant therapy with corticosteroids and/or immunosuppressives would also give a more clear signal regarding safety outcomes, but such studies may no longer be feasible. In general, clinicians should restrict the use of infliximab to patients with moderate to severely active Crohn’s disease who have failed conventional therapy such as corticosteroids and immunosuppressive therapy with azathioprine, 6-mercaptopurine, or methotrexate. Patients with ulcerative colitis and indeterminate colitis should not be treated with infliximab until definitive evidence of efficacy from placebo controlled trials is available.