We read with great interest the study by Ardizzone and colleagues (Gut 2006;55:47–53) and the excellent review of Sands (Gut 2006;55:437–41) commenting on the efficacy and side effects of azathioprine (AZA) in the therapy of ulcerative colitis. Ardizzone et al observed in their investigator blinded study, which included patients with steroid dependent ulcerative colitis, more mild to moderate adverse events in azathioprine than in mesalamine (5-ASA) treated patients (26% v 6%; p = 0.046). However, only two of 36 patients on AZA were withdrawn from the study because of adverse events. We would like to comment on the side effects of AZA, which we observed in a double blind, double dummy, randomised, prospective, multicentre study on the efficacy and safety of AZA (2.0–2.5 mg/kg/day) and 5-ASA (4 g/day) for prevention of postoperative endoscopic recurrence in Crohn’s disease.

Seventy nine patients (AZA, 42; 5-ASA, 37) were randomised within two weeks after surgery. TPMT genotyping was performed at baseline in order to exclude subjects with homozygous TPMT deficiency. However, the study was stopped prematurely because an interim analysis revealed that the hypothesis of superiority of AZA versus 5-ASA could not be tested with the planned sample size. In 37 patients (AZA, 18; 5-ASA, 19) who completed the study according to the protocol (treatment for one year), the primary study end point (treatment failure: severe endoscopic relapse, withdrawal due to clinical relapse or to adverse drug reaction) was evaluated.

Treatment failure was found to be equally high in each group (AZA, 9 of 18; 5-ASA, 9 of 19; p = 1.00, two sided Fisher’s exact test). Six of 18 patients on AZA and two of 19 patients on 5-ASA therapy were withdrawn because of adverse drug reactions (33% v 11%; p = 0.12, two sided Fisher’s exact test); reasons were leucopenia/anaemia (AZA, 1; 5-ASA, 1), elevated liver enzymes, arthralgia/myalgia, vomiting, abdominal pain, macroscopic fecal excretion of study medication (AZA, 1 each), and pancreatitis (5-ASA, 1). Clinical or severe endoscopic relapse was observed in three of 18 patients on AZA therapy and in seven of 19 patients on 5-ASA therapy (17% v 37%; p = 0.27, two sided Fisher’s exact test). Considering all 79 patients, adverse events were reported in approximately 70% of patients in each group (AZA, 29 of 42; 5-ASA, 26 of 37). Furthermore, in three of 42 “non-completers” an intolerable adverse event led to withdrawal (AZA, ileus; 5-ASA, cholecystitis, ankylosing spondylitis). Two further trials investigating the efficacy and side effects of AZA to prevent postoperative relapse of Crohn’s disease have been published recently.^1,2

In an open label study by Ardizzone and colleagues,¹ adverse events were observed more frequently (39% v 25%) in patients receiving AZA (2 mg/kg/day) than in those receiving 5-ASA (3 g/day). Fifteen of 69 patients in the AZA group and six of 69 patients in the 5-ASA group were withdrawn because of adverse events (22% v 9%; p = 0.04); reasons for withdrawal were leucopenia/thrombocytopenia (AZA, 7; 5-ASA, 0), elevated liver enzymes (AZA, 4; 5-ASA, 1), pancreatitis (AZA, 3; 5-ASA, 0), epigastric intolerance (AZT, 1; 5-ASA, 2), and increased serum creatinine (AZT, 0; 5-ASA, 3). In a double blind placebo controlled trial by Hanauer and colleagues,² nine of 47 (19%) patients receiving a relatively low dose of 6-mercaptopurine (6-MP 50 mg/day), six of 44 (14%) patients receiving 5-ASA (3 g/day), and four of 40 (10%) patients on placebo were withdrawn from the study because of adverse events, respectively; reasons for withdrawal were diarrhoea (6-MP, 2, 5-ASA, 2), leucopenia (6-MP, 2; 5-ASA, 0), alopecia (6-MP, 2; 5-ASA, 0), elevated liver enzymes (6-MP, 0; 5-ASA, 1), flatus, gastrointestinal bleeding, phlebitis (6-MP, 1 each), and allergic reaction, bowel obstruction, and arthralgia (5-ASA, 1 each).

In summary, we could not provide evidence for the superiority of AZA over 5-ASA in our prospective clinical trial. In contrast with the trials described above, we observed a higher rate of adverse drug reactions leading to withdrawal from the study in the AZA group. Placebo controlled trials are needed urgently to address the question of best postoperative immunosuppressive management.³ However, our observations indicate the difficulties that may arise in future trials for reaching an adequate statistical power to provide a valid answer to this question.