Infliximab, a chimeric mouse-human antibody which targets tumour necrosis factor α (TNFα), has been used in the treatment of inflammatory bowel diseases (IBD) for over a decade. It is effective in the treatment of refractory luminal and fistulising Crohn’s disease (CD)1^–5 and ulcerative colitis (UC).6 7 Infliximab induces a rapid clinical response, has a steroid-sparing effect, increases quality of life, promotes mucosal healing and reduces hospitalisation rates.8 Although these obvious benefits have led to the widespread use of infliximab in the treatment of IBD, concern still exists about its long-term safety, which includes the risk of developing cancer, serious infections and demyelinating disorders.

Randomised controlled trials have not shown important increases in serious adverse events.1 3 6 However, clinical trials have stringent inclusion and exclusion criteria and a limited follow-up and sample size, so extrapolation of possible hazards may be problematic. In addition, in most trials the control subjects also received at least one dose of infliximab. Post-marketing surveillance, in contrast, provides data from large numbers of patients and may detect serious low background events but is hampered by incomplete reporting and documentation. To date, two post-marketing cohort studies have addressed the long-term safety of infliximab in patients with CD, reporting on the occurrence of opportunistic infections, tuberculosis, autoimmune phenomena, malignancies and deaths. However, the duration of follow-up in these studies was relatively short9 10 and registries such as the Therapy, Resource, Evaluation and Assessment Tool (TREAT) typically include selected patient populations.10 Since only limited information is available on the long-term safety of infliximab in patients with IBD in a real-life setting, we report our experience in clinical practice in all patients treated with infliximab in our centre since 1994.

METHODS

This a retrospective cohort study of all patients with IBD treated with infliximab at the Gasthuisberg University Hospital, Leuven, Belgium between December 1994 and December 2008. All infusions were administered at the outpatient infusion centre of the hospital or during hospitalisation. Records of the infusion unit were used to identify patients who received infliximab during the study period. Patients with IBD followed up in the Gasthuisberg Hospital during the same period who had not been treated at any time with infliximab or any other biological therapy served as a control group. These patients were identified through our VLECC database, which has existed since 1997 and contains all patients with IBD being followed up at our hospital. Patients were only entered in the database after they had given informed consent. The VLECC register has been approved by the ethics committee of our hospital.

Although the exact age and gender of each patient was known and also analysed as such, further in this study, age is reported in age groups (20–39, 40–59 and 60+ years) and gender has been anonymised for ethical reasons.

The diagnoses of CD and UC were determined according to conventional endoscopic, radiological and histological criteria.11 Cases with clinical and endoscopic evidence of IBD limited to the colon but no definitive histological or other evidence of CD or UC were classified as inflammatory bowel disease, type unclassified.12 The medical records of these patients were reviewed. For patients with no follow-up in 2006, the currently treating gastroenterologist or family physician and/or the patients themselves were contacted. The following baseline data were collected for each patient: age, date of birth, gender, diagnosis, disease location, disease behaviour including perianal disease, extraintestinal manifestations, duration of disease and body weight.

Potential adverse events included any cancer or dysplasia, death, serious and/or opportunistic infections, autoimmune disorders (infusion reactions, drug-induced lupus erythematosus, demyelinating disorders) and cardiovascular complications. For all adverse events, the time of onset and resolution, seriousness, concomitant medication, outcome and relationship to infliximab were recorded. A serious adverse event was defined as any adverse event that resulted in hospitalisation or prolonging hospitalisation, was fatal or life-threatening or led to significant disability. Serious infections were defined likewise. The occurrence of serious infections, death and malignancies between 1994 and April 2008 were retrospectively recorded by review of all clinical charts. Whereas the control group could reliably be screened for the occurrence of malignancies and deaths, infections might have been underestimated in this cohort because of recall bias and the fact that the follow-up in these patients (often on milder treatment) was less standardised and frequent than in the patients treated with infliximab.

Follow-up was started at the earliest in 1994, the year the first patients were treated with infliximab, or from the time of diagnosis of IBD if later than 1994. Patients in the infliximab group were followed from the start of their first infusion.

Acute infusion reactions were defined as any adverse experience that occurred during or within 1 h of the infusion, and serum sickness-like reaction was defined as the occurrence of myalgia, arthralgia, fever or rash, or a combination of symptoms 1–14 days after an infusion. The diagnosis of drug-induced lupus erythematosus was made when at least one of the following clinical symptoms was present: muscle and joint pain and swelling, flu-like symptoms or fever associated with the presence of antinuclear antibodies and either anti-double-stranded DNA or antihistone antibodies. For infusion reactions, only patients who received more than one infusion were analysed.

Infliximab (Remicade; Centocor, Malvern, Philadelphia, USA) was administered at a dose of 5 mg/kg body weight as a 2 h intravenous infusion. Patients with fistulising disease received three infusions at weeks 0, 2 and 6 as induction therapy. Maintenance therapy for CD and for UC was defined as scheduled infusions every 8 weeks or less. All other treatment schedules were considered episodic and were usually individually tailored according to clinical considerations or preference of the treating physicians. Before 2002, patients were mostly treated episodically owing to a lack of published data regarding optimal treatment regimens and funding constraints. Concomitant immunosuppressive medication (systemic corticosteroids, azathioprine or 6-mercaptopurine, methotrexate), indication for infliximab treatment (luminal disease, fistulas or extraintestinal manifestations), dates and number of infusions, dosage of infliximab and, if applicable, the reason for discontinuation of treatment were recorded. If there was a loss of response, patients usually received a higher dose of 10 mg/kg body weight and/or the interval between infusions was shortened.

Statistical analysis

GENMOD in the statistical software SAS Version 9.1 (SAS Institute, Cary, North Carolina, USA) and SPSS Version 15.0 (SPSS, Chicago, Illinois, USA) were used for appropriate statistical methods. Descriptive statistics were calculated as percentages for discrete data and medians with interquartile ranges (IQR) for continuous data. Both univariate (χ² and Mann-Whitney tests) and backward Wald multiple regression analysis were performed to assess the associations between possible risk factors and the occurrence of adverse events. The individual potential risk factors included gender, age, weight and duration of disease at the time of the first infliximab infusion, concomitant immunosuppressive therapy and treatment schedules (induction and average treatment intervals).

A Poisson regression model with the natural logarithm of the length of the time periods as offset was used to compare the infection rates between infliximab periods, characterised by the absence or presence of methotrexate, azathioprine and corticosteroids. Since each patient could have various periods with and without methotrexate, azathioprine or corticosteroids, generalised estimating equations were used to account for possible within-patient correlations. Only infections occurring within 12 weeks of the last infliximab infusion were included in the analysis. p Values <0.05 were considered significant.

RESULTS

Patient characteristics

The characteristics of infliximab-treated patients and controls are shown in table 1. A total of 743 patients with IBD were treated with infliximab at the Gasthuisberg University Hospital between December 1994 and April 2008; 48 (6%) were adolescents or children at the time of the first infliximab administration. Total follow-up until April 2008 was 3775 patient-years, with a median of 58 months (IQR 33–88). Complete follow-up until 2008 was obtained in 737 patients (99%). The control group comprised 666 patients with IBD. Total follow-up was 6704 years with a mean follow-up of 144 months (IQR 83–163). As a consequence of the study design, the follow-up in the control patients was significantly longer than in the infliximab group.

The control group included slightly more men, was older and had a later age of onset; 42% (277/667) of the patients had never been treated with immunomodulators (azathioprine, 6-mercaptopurine or methotrexate).

Infliximab treatment

A total of 7276 infusions were administered. Patients received a median of 6 infusions (IQR 3–12); 92% of the patients received more than one infusion. The median treatment interval, including induction schedules, was 56 days (IQR 36–88). The indications for treatment are listed in table 1; luminal disease was the primary indication. Fourteen percent of the patients received infliximab at a dose of 10 mg/kg at some point in their treatment. A total of 112 serious adverse events occurred in 94 patients (fig 1).

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

Serious adverse events in patients treated with infliximab. Three patients died as a result of malignant disease and one of infection and are therefore listed twice.

Mortality

Twelve of the 743 patients treated with infliximab (1.6%; 0.3/100 patient years; 6 women and 6 men) died during follow-up, two within 3 months of the last infusion (table 2). The median age at death was 53 years (range 25–82). The only identified predictor of death was older age at first infusion (p<0.01). The death of a patient aged ⩾60 years following an Aspergillus infection was considered directly related to infliximab treatment. Four deaths, including three malignancies, were considered possibly to be related to infliximab treatment and seven deaths were unrelated to treatment. Three sudden deaths were recorded. One patient in the 40–59-year age group with psoriatic arthropathy and newly diagnosed CD died suddenly 2 weeks after the first infusion. Another patient, also in the 40–59-year age group, who had stopped responding to infliximab and had switched to certolizumab and subsequently to adalimumab collapsed following dental treatment 2 weeks after the start of adalimumab. The third patient (in the 20–39-year age group) had a long history of intractable CD and was on the active small bowel transplant list. This patient had been on total parenteral nutrition for the previous 7 years. The disease course had been complicated by recurrent catheter infections and electrolyte disturbances. The patient suddenly collapsed during an elective admission for the administration of parenteral nutrition. Another patient (in the 40–59-year age group) with an 11-year history of UC committed suicide 4 months after the last infusion but had not been known to have a psychiatric disorder, and a patient from the same age group with a 14-year history of CD died 4 years after the last infusion of unknown cause. There is no information available about the circumstances of this death. One patient in the 40–59-year age group who had been treated with infliximab 6 years previously was diagnosed with secondary amyloidosis 1 month after reinstitution of treatment and died 3 years later from complications of hepatic and renal failure. A patient with a 28-year history of CD in the 40–59-year age group developed progressive multifocal leukoencephalopathy on natalizumab. The last infusion of infliximab was administered 1.5 years before diagnosis. A somewhat older patient in the ⩾60-year age group with a 25-year history of CD underwent surgery 7 weeks after the last infusion for a sealed perforation of the colon. The hospitalisation was complicated by pulmonary emboli, aspiration pneumonia and gastric ulceration. The patient died in the postoperative period of respiratory insufficiency. Three patients died from malignancies.

In the control group 16 patients died (2.4%, 0.2/100 patient-years; 3 women and 13 men, table 3). The death rate did not differ from the infliximab group (OR 1.33 (95% CI 0.56 to 3.00, p = 0.45). The median age at death was 61 years (range 33–85) and did not differ significantly from the median age at death in the patients treated with infliximab. The only predictive factor of death was age (p<0.001).

Neoplasia

Fifteen malignancies and eight non-melanoma skin cancers were diagnosed in 21 patients (16 women and 5 men) during 3775 patient-years of follow-up (0.4/100 patient-years) (table 4). None of the lesions was present at the start of infliximab therapy. One patient was diagnosed with pancreatic cancer and died soon afterwards and another patient with primary sclerosing cholangitis (PSC) died of a cholangiocarcinoma. A third patient, also with PSC, had recently been diagnosed with metastatic adenocarcinoma, most probably of intestinal origin. Since colonoscopy did not reveal a primary tumour or dysplasia, a primary cholangiocarcinoma was thought to be the most likely origin. The patient succumbed to his disease 1.5 years after diagnosis. One patient aged ⩾60 years with breast cancer was diagnosed with bone metastasis and two cases of lymphoma were observed. One patient was diagnosed with a non-Hodgkin’s lymphoma and one with a Hodgkin’s lymphoma. Both patients underwent treatment and are doing well. All other patients are alive and well. Three malignancies developed under adalimumab therapy but these patients had also been treated with infliximab previously. Two patients had undergone liver transplantation before infliximab treatment. The median age at diagnosis was 42 years (range 23–73). The median time intervals from the first and last infliximab infusions to the diagnosis of malignancy were 41 months (range 3–88) and 4 months (range 1–75), respectively. The median number of infusions before diagnosis was 6 (range 1–56). Although all but one of the patients who developed malignancies received methotrexate and/or azathioprine during infliximab treatment, concomitant immunosuppressive medication was not identified as a risk factor for the development of malignancies in univariate or multivariate analysis. The occurrence of malignancies was only associated with a longer disease duration (p<0.001).

In addition, two patients developed meningeomas. One of the patients has also been treated with adalimumab. An elderly patient with a 9-year history of UC was diagnosed with severe dysplasia of the colon. In the light of co-morbidity and advanced age, the patient was conservatively treated and is still alive with no signs of overt colorectal cancer on recent colonoscopy.

In the control group 37 malignancies and five non-melanoma skin cancers were diagnosed in 30 patients (10 women and 20 men) during 6704 years of follow-up (0.5/100 patient-years) (table 5). The median age at diagnosis was 55 years (range 34–87), which was significantly higher than in the infliximab group (p<0.01). Three of the malignancies were non-Hodgkin’s lymphomas, and there were eight colorectal cancers (four in patients with CD and four in patients with UC). Twenty-four of the 35 patients with malignancies (67%) had never been treated with immunosuppressive therapy other than corticosteroids. The only predictive factor for malignancies in the control group was older age (p<0.01). There was no significant difference in the incidence of cancer between patients treated with infliximab and the control group (OR 0.97 (95% CI 0.56 to 1.65, p = 0.91).

Infections

Fifty-nine serious infections occurred in 48 patients (6%) during treatment with infliximab (1.6/100 patient-years). Thirteen infections were disease-related and included seven perianal and seven intra-abdominal abscesses. One fatal infection occurred in a patient with CD aged 60+ years with a history of diabetes, pulmonary emboli and severe aortic stenosis who had been treated with infliximab intermittently over the previous 6 years. The patient was admitted with an Aspergillus infection and several days later succumbed to sepsis with multiorgan failure. Another patient (in the 40–59-year age group) presented with miliary tuberculosis after 17 infusions. The patient had been concomitantly treated with azathioprine. A young patient (in the 20–39-year age group) treated with methotrexate was diagnosed with caseating granulomas of the liver. Although Mycobacterium was isolated, the lesions resolved following treatment with antituberculosis drugs. Both patients had negative tuberculin skin tests. Sixteen patients had positive skin tests and/or fibrotic lesions on the chest radiograph and received antituberculosis chemoprophylaxis with nicotibine and pyridoxine for 6 months. None of these patients developed tuberculosis during follow-up. One patient was diagnosed with cytomegalovirus pneumonitis and myocarditis complicated by cardiac failure and respiratory insufficiency shortly after the third infusion. Seven patients were admitted for bacterial gastroenteritis and eight for respiratory tract infections or pneumonia. Other serious infections included appendicitis, pelvic inflammatory disease and skin-related infections. Ten cases of herpes zoster and four additional cases of cytomegalovirus may be listed as opportunistic infections but were all managed on an outpatient basis and not regarded as serious infections. Concomitant treatment with methotrexate or azathioprine had no effect on the occurrence of infections. The infection risk was significantly increased in periods where concomitant corticosteroids were administered (p = 0.018; OR 2.69, 95% CI 1.18 to 6.12). One patient had eight infections during infliximab treatment (six of which were catheter sepsis which frequently occurred before starting and after stopping of infliximab). Since the mean (SD) number of infections in patients with infections was 1.25 (0.44) (0.09 (0.35) for the whole cohort) and this patient skewed the results (data not shown), he was not included in the final analysis.

Of the 48 patients aged <18 years, five severe adverse events were documented in four patients, all of which were infections (8%). Two of them were concomitantly treated with corticosteroids at the time of the infection.

The number of severe infections in the control group (77 in 62 patients; 1.1/100 patient-years) did not differ significantly from the number of infections in the infliximab group. Nine of the severe infections may be regarded as opportunistic including four cytomegalovirus, one Cryptosporidium, one Pneumocystis and two generalised herpes simplex infections. All these infections occurred during treatment with azathioprine; the patient with Pneumocystis received triple therapy with cyclosporine, corticosteroids and azathioprine.

Infusion reactions

A total of 115 of 682 patients (17%) at risk experienced acute infusion reactions after a median of 2 infusions (IQR 1–5). Although in most patients infusion reactions could be managed with intravenous corticosteroids, cetirizine dihydrochoride and slowing of the infusion, the treatment had to be discontinued in 19 patients. Induction therapy (p = 0.03, OR 0.56, 95% CI 0.34 to 0.95), scheduled maintenance therapy (p = 0.03, OR 0.56, 95% CI 0.34 to 0.94) and concomitant azathioprine therapy (p<0.001, OR 0.42, 95% CI 0.27 to 0.67) were independent protective factors for the occurrence of acute infusion reactions. In two patients the acute reaction necessitated hospital admission and was defined as a serious adverse event.

In 50 patients (7%) a delayed-type hypersensitivity reaction occurred after a median of 3 infusions (IQR 1–8). In 46 of these patients (92%) infliximab was eventually discontinued, either because of the delayed-type reaction itself, loss of response or a combination of the two. In 10 cases the delayed-type hypersensitivity reaction was an indication for admission to hospital. Serum sickness-like disease occurred more frequently in women than in men (9% vs 4%; p<0.01; OR 3.74, 95% CI 1.55 to 9.04). Similar to acute infusion reactions, induction schedules (p = 0.03; OR 0.33, 95% CI 0.12 to 0.89) and scheduled maintenance therapy (p = 0.01; OR 0.37, 95% CI 0.17 to 0.79) were independent protective factors but concomitant immunosuppressive medication was not. Patients who developed delayed-type hypersensitivity had a longer median treatment interval, reflecting treatment holidays, than patients without (18 weeks (IQR 8–34) vs 9 weeks (IQR 5–12), p<0.001).

Autoimmune phenomena

Two patients developed arthritis and a butterfly facial erythema in the presence of anti-double-stranded DNA antibodies consistent with drug-induced lupus erythematosus, in both instances after four infusions. One patient developed Coombs negative autoimmune haemolytic anaemia 6 months after a single administration of infliximab for which corticosteroids were given.

Another patient developed sensory loss in the lower left extremity. MRI examination of the spinal cord showed a hyperintense lesion at the level of the cone compatible with demyelinisation. Infliximab treatment was discontinued after four infusions and the symptoms and radiographic lesions resolved. A 33-year-old patient was diagnosed with an optic neuritis after the second infusion which resolved with discontinuation of treatment. The diagnosis was made on clinical grounds and MRI was not performed.

Dermatological symptoms

One hundred and fifty patients (20%) reported a wide variety of skin eruptions during the course of infliximab treatment. In 64 cases the lesions were judged by the patient or treating physician to be sufficiently severe or incapacitating for referral to a dedicated dermatologist in our centre. Of these 64 patients, 39 (61%) were diagnosed with psoriasiform dermatitis and eczema was also frequently diagnosed. The lesions were most often located in the face (26%), pubic region (20%), groin (15%) and hand palms and soles (16%). The most frequent concomitant symptom was itching (40%) followed by pain in (30%). Biopsy samples were not taken systematically in all patients. In 15% of patients, biopsy samples were taken to obtain a definite diagnosis. In these biopsies, three types of lesions could be distinguished: eczema with spongiotic dermatitis; psoriasis with psoriasiform acanthosis; and parakeratosis, dilated capillaries and subcorneal pustule formation. Most skin lesions responded well to topical steroids, and treatment with infliximab had to be discontinued because of skin problems in only two patients. Patients who reported skin lesions were younger (p = 0.012) and more often women than patients without skin lesions (p<0.01; OR 1.9, 95% CI 1.24 to 2.85).

DISCUSSION

Although the use of infliximab has an excellent short-term safety profile, only limited data are available on its long-term safety in patients with IBD. The present study, which comprises the largest single-centre cohort study on patients with IBD to date, provides long-term safety data in an unselected group of consecutive patients with IBD (UC and CD). Complete follow-up was obtained in almost all patients. Our results confirm that infliximab is generally well tolerated and that severe side effects are no more frequent than in patients with IBD not receiving infliximab.

The mortality rate of 0.3/100 patient-years is slightly lower than the 0.53/100 patient-years reported in the large case-control study of the TREAT registry10 and did not differ from the mortality rate in the control group. Although the median age of the control group was higher, the median age at the time of death was comparable in the two groups. Of the 12 patients in the infliximab group who died, 5 died >2 years after the last infusion was administered. Only one death—an elderly patient who died of aspergillosis—was considered directly related to infliximab. The only risk factor associated with death was older age at the time of the first infusion.

A major concern regarding the long-term use of infliximab is the potential predisposition to the development of malignancies. Although placebo controlled trials have not shown an increased incidence of cancer in patients treated with infliximab,1 3 6 13 follow-up has usually been short. The observed malignancy rate in our population is similar to that reported in previous studies of infliximab in patients with IBD.9 14 15 A possible link between non-Hodgkin’s lymphomas and anti-TNF therapy has been suggested, but the association is not supported by our data. Non-melanoma skin cancers were frequently seen in the infliximab-treated patients, but not significantly more than in the controls. Findings from a large study in patients with rheumatoid arthritis have identified treatment with TNF inhibitors as a risk factor for the development of non-melanoma skin cancers.16 Most data regarding cancer risk and the use of immunosuppressive medication are derived from post-transplantation studies.17 Although the mean follow-up period in our infliximab cohort was only 5.1 years, the fact that only 1% of our patients developed skin cancer, two of whom had a history of liver transplantation, makes recommendations for intensified screening not justified at present. However, the age at diagnosis of non-melanoma skin cancers in the infliximab group was younger, and one should be aware of the possibility of skin cancer in this patient population.

Although our malignancy data cannot be directly compared with other cohort studies of patients with IBD, the rate per patient-year of follow up is very similar to rates reported in other studies in patients with IBD not treated with infliximab18^–20 and did not differ from the incidence of malignancy in the control group. Taken together, infliximab does not seem to confer an increased risk for the development of cancer.

Acute infusion reactions occurred in 17% of patients. This is similar to previously reported incidences in the major clinical trials.1 3 6 However, a retrospective Mayo Clinic study observed acute reactions in <4% of patients that necessitated discontinuation of treatment in over half of the group. In our cohort, infusion reactions could usually be managed easily and required discontinuation of treatment in 2.3% of the patients. This is in agreement with the discontinuation rate of 3% in combined safety data from clinical trials (Remicade package insert, Centocor, Malvern, Philadelphia, USA, 2007). The discrepancy with the results reported by the Mayo Clinic can be explained by the fact that, in the latter study, only severe reactions were recorded. Their overall discontinuation rate of 1.8% is comparable to our results. It has been suggested that concomitant treatment with immunomodulators prevents the development of antibodies to infliximab and therefore reduces the incidence of infusion reactions.1 21 Indeed, concomitant azathioprine therapy was associated with a reduced incidence of acute infusion reactions. Serum sickness-like disease was seen in 50 patients (7%) and led to the discontinuation of infliximab treatment in most of these patients. Although the occurrence of delayed-type hypersensitivity was uncommon in the clinical trials, it happens frequently in episodically treated patients with long treatment holidays.22 We confirm that delayed systemic reactions are associated with episodic infliximab treatment and long treatment intervals. Also, an induction series of three infusions at the start of treatment reduced the incidence of delayed reactions. Taking account of the fact that infusion-related reactions are strongly associated with loss of response,21 our results offer a clear mandate for induction therapy followed by bimonthly scheduled infusions and the use of concomitant immunosuppressive medication, at least at the start of treatment.

Reflecting the role of TNF in the systemic inflammatory response to infection, specifically those involving intracellular pathogens,23 various opportunistic infections have been reported.9 24 There is particular concern regarding the reactivation of mycobacteria.25 In this cohort, which involved a population with a low endemic incidence of tuberculosis, two cases of extrapulmonary tuberculosis were reported. Both infections occurred in patients who had negative pretreatment screening for tuberculosis. This should be viewed in the context that anergic tuberculin skin testing may reach up to 83% in patients with CD on immunosuppressive therapy.15 Consequently, skin testing on its own is an unreliable indicator for previous exposure to tuberculosis.15 Contrary to previous data which have reported failure of tuberculosis prophylaxis in one-quarter of patients,26 none of our 16 patients with a positive skin test developed active mycobacterial infection. These results suggest that infliximab can be safely administered in positive patients under appropriate chemoprevention.

Whether infliximab itself poses an additional risk factor for infections in patients already treated with the traditional immunosuppressive medications has been debated. Several studies reported an increased risk of serious infections associated with anti-TNF use in patients with rheumatoid arthritis.27 28 In contrast, a large British registry for patients with rheumatoid arthritis and the American TREAT registry for patients with IBD did not identify anti-TNF therapy as an independent risk factor for the development of serious infections.10 29 In the present study, multivariate analysis yielded only the concomitant use of corticosteroids as an independent predictor for serious infections associated with infliximab treatment. There was no evidence that the effect of administering corticosteroids was mediated by the presence or absence of methotrexate or azathioprine. These findings are consistent with the results of the TREAT registry, which found that only disease severity and prednisone use were associated with an increased risk for serious infections.10 Taken together, it is recommended that the combination of corticosteroids and infliximab should be avoided.

The high incidence of skin eruptions during treatment with infliximab affecting more than one-fifth of patients was unexpected. New onset of skin lesions occurring during anti-TNF therapy has been sporadically reported in patients with rheumatoid arthritis, with psoriasis and eczema-like manifestations representing the majority of cases.30^–32 Skin eruptions have been attributed to all three commercially available anti-TNF treatments (etanercept, adalimumab and infliximab) and appear to be a class effect of these drugs. Systematic reports addressing the issue of skin lesions in large cohorts are not available and there are only very limited data on skin reactions in patients with IBD. The occurrence of new onset psoriasiform lesions is particularly surprising since anti-TNF agents have been successfully used in the treatment of psoriasis. The aetiology of this paradoxical reaction remains unclear. It has been postulated that excessive TNF inhibition may lead to activation of autoreactive T cells, leading to tissue damage through autoimmune mechanisms.33 In most of our patients the eruptions responded well to topical treatment and necessitated discontinuation of treatment in only two patients.

Although infliximab induces the formation of antinuclear antibodies (ANA) in over half of the patients,34 clinical autoimmune disease is rarely seen. The two cases of drug-induced lupus erythomatosus and the patient with haemolytic anaemia have been reported previously and occurred in patients with positive ANA titres.34 Most of our patients have not since been systematically analysed for autoantibodies, but the fact that all autoimmune disease became manifest after a maximum of four infusions is compatible with the fact that seroconversion in almost all instances occurred before the fifth infusion and after a median of 12 months.34 The development of autoimmune disease associated with prolonged infliximab use is therefore not expected.

The present study comprises the largest single-centre cohort study to date and provides complete long-term safety data in an unselected group of patients with IBD. Unlike clinical trials that employ stringent inclusion criteria and the TREAT registry in which selected patients were included and withdrawn at the discretion of the participating clinicians, this study included all patients ever treated with infliximab at our centre. In addition, the follow-up period in our cohort was more than three times longer than in the TREAT registry and Mayo Clinic series.9 10 This is particularly important when evaluating life-threatening side effects such as malignancies which may become apparent only after a long latency period.

In conclusion, the findings of this study show that long-term treatment with infliximab is safe and well tolerated and is not associated with excess mortality or malignancies. Strict inclusion and exclusion criteria should be used within the label, with careful pretreatment screening of potential candidates for infliximab treatment in order to obtain optimal results.