Safety and efficacy of azathioprine in the maintenance of ciclosporin-induced remission of ulcerative colitis


Correspondence to: Dr Actis Department of Digestive Disease and Nutrition, Ospedale Molinette, Torino 10126, Italy. E-mail:



It has been shown that azathioprine prolongs the response to ciclosporin of steroid-refractory ulcerative colitis, but no specific data are available concerning its toxicity in this indication.

Aim and methods:

The charts of 21 patients with steroid-refractory ulcerative colitis who received azathioprine overlapping with a successful ciclosporin course were reviewed for the onset of toxicity. The controls consisted of 48 initial responders to steroids who received azathioprine for steroid-dependence or resistance/toxicity.


Two of the 21 patients were withdrawn because of hypersensitivity to azathioprine. The remaining 19 were treated for a median of 18 months together with a median daily steroid dose of 35 mg (10–75 mg) to be tapered off. Toxicity (31%) included leukopenia alone (two cases), cholestasis alone (one case), cholestasis and increased amylase (one case), increased amylase alone (one case), and cutaneous infection (one case). The frequency of withdrawal was 21%. The mean daily steroid doses were reduced from 38 mg to 3.8 mg in the study cohort, and from 25 mg to 8 mg in the controls, among whom toxicity (27%) included four cases each of leukopenia and increased amylase, two cases each of alteration of liver enzymes and infection, and one case of gastric intolerance. Ten of the 48 controls (20%) were withdrawn from the study.


Azathioprine is as effective and safe in the maintenance of the response of patients with steroid-refractory ulcerative colitis to ciclosporin as it is in the treatment of those who respond to steroids.


The purine analogues, azathioprine and its metabolite 6-mercaptopurine, have been used since the 1960s. Although they were shown as long ago as 20 years to be effective in maintaining the remission of Crohn’s disease, the fear of leukopenia and malignancy has persuaded clinicians to refrain from their use in inflammatory bowel disease, or led to an incoherent array of protocols.1, 2 Accordingly, there is scanty information on the efficacy of azathioprine/6-mercaptopurine in ulcerative colitis, with only one report in the last 20 years showing that withdrawal of azathioprine favours disease relapse.3 Nevertheless, interest in the use of azathioprine was recently renewed by the need to avoid colectomy in the treatment of steroid-resistant ulcerative colitis, and to augment and prolong the time-limited pharmacological benefit of ciclosporin with a long-acting immunosuppressive agent alternative to steroids.4 On the basis of the evidence, achieved by others and ourselves, that immediate responders to ciclosporin maintain a significantly longer remission if their initial regimen is overlapped with azathioprine, the demonstration that this combination does not entail any synergistic toxicity has gained priority.5, 6

To address this issue, we reviewed the charts of our ciclosporin-responsive ulcerative colitis patients who received azathioprine as part of their maintenance regime. The data were compared with the follow-up outcomes of a similar cohort of patients who received azathioprine after conventional treatments, and with the literature data.


This retrospective chart review included a total of 69 patients with ulcerative colitis who received azathioprine as the main maintenance treatment. Twenty-one patients (14 male, seven female, aged 19–58 years) had experienced a steroid-resistant episode responding to ciclosporin within a disease duration of 5.9 years (0.5–17):7, 8 12 of them had universal or subtotal colitis, and nine left-sided disease. Four of the 21 also had extra-intestinal inflammatory manifestations including arthritis, sacroileitis, uveitis, finger clubbing, and haemolytic anaemia. One (who initially showed signs of acromegaly) was diagnosed and treated for a pituitary adenoma during the follow-up. Two patients discontinued azathioprine after the first dose because of hypersensitivity and were excluded from the intention-to-treat analysis, which was therefore based on the results in 19 patients.9 Four of them, enrolled earlier in the study, received intravenous ciclosporin followed by oral conventional ciclosporin for 6 months; the remaining patients, enrolled later, were treated with the microemulsion formulation (NEORAL) for 3 months.7 Thirteen of them received azathioprine to maintain the remission previously achieved with ciclosporin, the two drugs being overlapped for 8 weeks in nine cases; a further three started azathioprine treatment at the time of the first relapse, following the discontinuation of ciclosporin. The remaining three received concurrent NEORAL and azathioprine for 3 months. Overall, azathioprine was administered for a median of 18 months (range 2–72 months). The initial daily dose was 100 mg in 10 patients, 150 mg in eight patients, and 75 mg in one patient; this was adjusted to the target ratio of 2 mg/kg body weight. The concurrent medications included mesalamine (usually at a dose of 2.4 g daily) and oral prednisone at a dose of up to 75 mg daily (median 35 mg), which was gradually tapered off. No prophylactic antibiotic therapy was used.

The control cohort included 48 patients (31 male, 17 female, aged 19–73 years; 31 with universal/subtotal colitis, 17 with left-sided disease; disease duration 5.1 years) who had received a daily median steroid dose of 25 mg (5–50 mg) for 14 months, together with a standard dose of mesalamine. The rationale underlying the use of azathioprine was steroid dependence in 50% of the cases, and resistance/toxicity in the other 50%. These patients received azathioprine for 18 months (range 1–54 months), usually at the daily dose of 2 mg/kg. Toxicity was defined on the basis of standard clinical and/or biochemical criteria: ALT > 2 × nl, WBC < 4000 and amylase > 2 × nl were considered abnormal. The decision as to whether to reduce or discontinue the treatment was based on the physician’s judgement at any of the monthly check-up visits. Table 1 summarizes the characteristics of the two patient groups.

Table 1.   Salient characteristics of the two patient groups Thumbnail image of

Statistical analysis

The colectomy rates and the frequencies of toxicity in the two treatment groups were analysed using the χ2-test. The power of the study was calculated on the basis of these frequencies as discussed in the Results section.10


Table 1 shows that the two patient groups had comparable features, including treatment with mesalamine, which has recently been shown to influence the metabolism of azathioprine/6-mercaptopurine.11

Of the 19 patients in the ciclosporin group, 16 received azathioprine for longer than 3 months (the expected lag-time of the action of the drug) and were available for the analysis of the effects of azathioprine on steroid-tapering schedules. The results (Table 2) showed that the mean daily steroid dose of 38 mg before the beginning of azathioprine had been reduced to 3.8 mg by the end of the observation period; the percentage of patients being treated with > 10 mg of steroids decreased from 87% to 6%, and the percentage receiving ≤ 10 mg increased from 12% to 93%. Ten of the 16 patients (62%) were totally weaned from steroids.

Table 2.   Percentage distribution of the dosages of steroids and mean steroid doses before and after the use of azathioprine Thumbnail image of

In terms of surgical outcomes, five of the 19 patients receiving azathioprine (26%) underwent colectomy during the follow-up, as opposed to nine out of 11 subjects (81%) described elsewhere who did not receive azathioprine as a maintenance treatment (P=0.011).7

Six patients (31%) experienced a total of seven toxic episodes (Table 3): cholestatic hepatocellular damage, symptom-less pancreatitis, and leukopenia occurred twice, yielding a frequency of 10%. During the second year of azathioprine treatment, one patient (5%) developed two symmetrical subcutaneous abscesses of the buttocks close to the edge of a continuously worn orthopaedic bust. This was the latest event to appear; all of the others appeared within the first 6 months. The toxicity was judged to be severe enough to require treatment discontinuation in four of the six cases (21% of the 19 patients); of the remaining two cases, one resolved spontaneously and the other responded to an azathioprine dose adjustment.

Table 3.   Characteristics of the adverse events that occurred during the azathioprine treatment Thumbnail image of

Thirty-one of the 48 patients in the control cohort (64%) showed a reduced steroid need (17 were eventually weaned); the mean steroid dose at the end of the follow-up period was 8 mg/day. Signs of toxicity were noted in 13 out of 48 (27%), with four cases each of leukopenia and symptom-less pancreatitis, two cases each of abnormal liver function test results and infection, and one case of gastric intolerance. The treatment was discontinued in 10 patients (20%). These figures were not statistically different from those in the study group (P=0.946).

Given the number of patients that were included, this study has a statistical power of 80% (with alpha-error of 0.05, two-tailed) to detect differences of 30% or greater. The power to observe differences of 4% (31% vs. 27%) is therefore lower. This is also reflected by the wide 95% confidence interval (from – 19.9% to 28.9%).


A subset of ulcerative colitis patients require ciclosporin to achieve remission but we and others have found that ciclosporin is of limited value in maintaining this remission and that the timely use of the cytotoxic drug azathioprine can make a crucial difference in prolonging the response.12, 13 However, the poor general condition of patients, residually high steroid doses, and the synergistic potential of ciclosporin and azathioprine increasing the likelihood of liver damage and infection, have led to concern about the safety of azathioprine in this indication. The aim of this retrospective intention-to-treat chart review was to investigate this increasingly important issue.

Azathioprine-containing schedules reduced the need for steroids in our cohort of ciclosporin-treated patients, and those receiving the drug as maintenance therapy underwent colectomy significantly less frequently than those who did not. Furthermore, as we have reported elsewhere in relation to a similar series of patients, azathioprine is also effective in reducing the probability of a relapse requiring steroid therapy.6

The present study revealed an overall frequency of adverse events of 31%, with 21% frequency of treatment withdrawals. The frequency of leukopenia, cholestasis, and symptom-less pancreatitis was 10%; we also recorded one episode of subcutaneous infection (5%) but this was probably due to mechanical factors rather than a specific effect of azathioprine. These findings were not statistically different from those in our azathioprine-treated patients who did not receive ciclosporin; in this control cohort, the frequency of toxicity was 27% and that of withdrawal was 20%. The safety and efficacy of azathioprine in inflammatory bowel disease has been investigated in three studies published in recent years. Two of them were designed as retrospective chart reviews in analogy with the current one.14–16 In the first, leukopenia was detected less frequently (3.8%), perhaps as a result of the choice to define a lower WBC threshold as leukopenia. The second study recorded azathioprine-related side-effects as frequently as we did (32%), but also included three cases of colonic neoplasms in patients whose disease duration far exceeded our figures. The third study has recommended the discontinuation of azathioprine after the fourth year of treatment, to minimize the risks of a prolonged immune suppression: the results of the GETAID study should hopefully clarify this issue in the second half of 2001.17 More recently, a large-scale survey of the toxicity of azathioprine in paediatric patients with Crohn’s disease found an overall toxicity rate of 28%, with cholestasis being the most frequent damage (13%).18 Our study group data are therefore similar to those published in the literature.

The observed side-effects did not occur randomly (Table 2), but were related to the dose to which the patients were exposed and the known mechanisms of action of azathioprine.9 The most severe forms of dose-dependent toxicity (bone-marrow depression and infection) were restricted to the highest dose of 150 mg daily, whereas amylase changes and hepatitis (believed to be allergic reactions) occurred at lower dosages.

In conclusion, this retrospective analysis found that the frequency and the type of adverse events associated with the use of therapeutic doses of azathioprine given after or simultaneously with ciclosporin to maintain the remission of refractory ulcerative colitis were within the expected range, and did not differ from the findings in a comparable group of patients achieving remission on conventional treatments. Further confirmation to such findings will require prospective studies powered to detect the small frequency difference observed in the current analysis. The safety of azathioprine is likely to become a compelling issue, with the expanding use of anticytokine treatment strategies for inflammatory bowel disease. Like ciclosporin, anticytokine drugs seem to have a quick onset of action but their relatively short-lived efficacy means that they are not effective in maintaining remission. Although repeated injections have been proposed to boost efficacy, the use of azathioprine/6-mercaptopurine also provides an additional treatment benefit.19, 20

Lastly, we note that two of our patients were not treated because they were allergic to azathioprine, an event that has an estimated 2% frequency.9 In light of the crucial synergy that seems to exist between azathioprine/6-mercaptopurine and fast-acting immunosuppressors, it is advisable to test patients with severe recalcitrant inflammatory bowel disease for allergy to azathioprine/6-mercaptopurine before choosing between immunosuppressors and surgical options.