The use of azathioprine or 6-mercaptopurine for maintaining remission in Crohn’s disease patients who are steroid dependent or resistant is unequivocally supported by evidence from randomised controlled clinical trials. The same, however, cannot be said for the use of immunomodulator therapy in ulcerative colitis (UC). Trials are scanty, small in size, conflicting in results, and clinical practice is dominated by support from low quality evidence from open series reports. In addition, outcome measures used in different trials vary considerably, and the tools used to assess clinical disease activity are numerous and diverse.

In the first randomised controlled trial, conducted way back in 1974 by Jewell and Truelove,¹ a two by three stratification was used. Inpatients or outpatients with active UC were stratified into first attack, short history (less than five years), and long history (more than five years). The acute episode was treated with corticosteroids, either 20 mg oral prednisolone plus steroid enemas for outpatients or 40 mg prednisolone 21-phosphate with rectal hydrocortisone for inpatients. Azathioprine was added immediately at a dose of 2.5 mg/kg. In the first 40 patients, the azathioprine dose was reduced after three months to 1.5–2.0 mg/kg whereas in the next 40 patients the dose was maintained at 2.5 mg/kg throughout the trial period of one year. It was not surprising that azathioprine was of no value in induction of remission as the end point was at one month after commencement of azathioprine. As maintenance therapy, azathioprine lacked value in patients being treated for the first attack of UC. Although there was a trend towards some benefit in patients with established UC who had relapsed, this was not statistically significant. As the numbers were small, type II error was quite possible. Nevertheless, it was concluded that azathioprine was less valuable as a maintenance therapy than sulphasalazine.

Subsequently, several other studies have been performed which, unfortunately, have not necessarily clarified the situation. In a small, double blind, randomised, controlled trial in 20 steroid naïve active UC patients for three months, azathioprine 2.5 mg/kg was, however, as effective as sulphasalazine 65 mg/kg in controlling disease.² In a randomised controlled trial comparing azathioprine 2.0–2.5 mg/kg (24 patients) with placebo (20 patients) added to conventional corticosteroid therapy, clinical disease activity scores in the two groups showed no significant difference at three or six months although azathioprine did exert a steroid sparing effect.³ A small open randomised study from India on 25 active UC patients treated with 1 mg/kg steroids, azathioprine, or sulphasalazine maintenance was associated with similar relapse rates over an 18 month therapy period.⁴ Therefore, these studies did not provide entirely convincing evidence of benefit of maintenance therapy with azathioprine in UC, although none of the studies recruited steroid dependent or resistant UC patients, a more relevant clinical indication. In the first double blind randomised controlled trial in steroid dependent UC patients, treatment with azathioprine 1.5 mg/kg (n = 16) was compared with placebo (n = 14) for six months. No symptomatic or proctoscopic differences were observed between the two groups although steroid dose was lower in the azathioprine group at study termination compared with the placebo group.⁵ The dose of azathioprine in this study was low compared with the currently accepted optimal dose and type II error cannot be excluded.

Things began to clarify with the publication of a pivotal study that established the use of azathioprine in UC. This trial took the form of a double blind placebo controlled trial of withdrawal or continuation of azathioprine and was conducted in the UK.⁶ Patients with UC on azathioprine for at least six months and in remission for at least two months were recruited. The one year relapse rate was 36% for patients continuing azathioprine (n = 33), significantly less than 59% for those on placebo (n = 34). Mean azathioprine dose was 100 mg/day and most patients were also on aminosalicylates. This study established that, in patients maintained on azathioprine for a mean duration of approximately just over one and half years, withdrawal of azathioprine leads to increased relapses compared with continuation of the drug. This study therefore is relevant to a selected group of patients who have poorly controlled UC, who have a good response to azathioprine, and who tolerate it well. This study prompted the more widespread use of azathioprine in poorly controlled UC. However, although this study provides indirect evidence of the benefit of azathioprine maintenance in UC, it does not provide direct evidence of the benefit of azathioprine in inducing and maintaining poorly controlled UC patients in remission.

Retrospective or open label analysis has suggested response rates of 84%⁷ and 63%⁸ as well as remission rates of 65%⁹ and 69%¹⁰ from single tertiary care centres although the definitions of response and remission have been inconsistent. In the largest retrospective series analysis from Oxford encompassing 30 years of experience, overall remission rate was 58% in UC patients (n = 346), and duration of azathioprine treatment did not affect relapse rates after discontinuation of therapy.¹¹ Retrospective series analysis has also confirmed decreased steroid requirement and clinical relapses in the three years after commencement of azathioprine compared with two years prior to azathioprine.¹²

Two recent studies have provided further evidence in favour of the use of azathioprine in UC. In a Spanish study published last year, 34 patients with UC receiving prednisolone were randomised on a 2:2:1 basis to 1.5 mg/kg of 6-mercaptopurine, 15 mg/week of methotrexate, or 3 gm/day of 5-aminosalicylic acid.¹³ All patients who achieved remission over a 30 week treatment period continued into the maintenance phase of the study for 76 weeks. Induction of remission was significantly superior in the azathioprine group compared with the 5-aminosalicylic acid group (78.6% v 25%). Maintenance of remission in those who achieved remission was 63.6% in the azathioprine group compared with none in the 5-aminosalicylic acid group. Methotrexate was inferior to azathioprine in both induction and maintenance of remission.

In a further study from Milan reported in this issue of Gut by Ardizzone and colleagues,¹⁴ 72 patients with active steroid dependent UC were randomised (investigator-blind) to azathioprine 2 mg/kg/day or 5-aminosalicylic acid 3.2 g/day for a six month period (see page 47). Clinical and endoscopic remission with discontinuation of steroids was achieved in 53% of azathioprine treated patients compared with 21% of 5-aminosalicylic acid treated patients (intention to treat analysis: odds ratio 4.78 (95% confidence interval 1.57–14.5). Definition of steroid dependence was standardised and all patients were on the same dose of prednisolone 40 mg/day at study entry. This study provides the best evidence for the efficacy of azathioprine in steroid dependent patients. In addition, the steroid sparing effect of azathioprine in UC was confirmed.

In UC, use of immunomodulator therapy has not been backed up by good randomised controlled trial evidence (table 1). The controlled trial evidence in favour of using methotrexate is poor.^13,15 Retrospective series and audit experience of using methotrexate in UC suggests evidence of moderate efficacy.^16,17 The Milan trial reported in this issue of Gut¹⁴ provides support for the use of azathioprine in steroid dependent UC patients. However, in severely ill UC patients refractory to steroids, azathioprine is likely to be too slow to be of benefit in preventing colectomy. In this scenario, ciclosporin may be used to spare colectomy, but a recent Cochrane Database systematic review concluded that there is limited evidence that ciclosporin is more effective than standard therapy alone for severe UC.¹⁸ The rapid response is the main attraction for the short term use of ciclosporin but long term benefits are unclear and the potential toxicity profile formidable. Addition of azathioprine to ciclosporin may reduce the propensity for early relapses, at the cost of increasing toxicity,^19,20 but over seven years 58% of such patients may come to colectomy.²¹ The early experience with infliximab is very promising with respect to avoiding colectomy.²²

The Milan controlled study has provided much needed evidence in support of the use of azathioprine in steroid dependent UC patients and showed that this strategy is better than using 3.2 g/day of 5-aminosalicylic acid. This is important as previous evidence supporting such use from randomised controlled trials was limited. Whether using higher doses of 5-aminosalicylic acid in such patients might increase efficacy significantly is unclear at present but would appear unlikely. The evidence supporting the efficacy of methotrexate in UC either for inducing or maintaining remission also remains unconvincing. In steroid resistant patients the options are more limited, especially in severe disease, as azathioprine is too slow in its onset of effect. Ciclosporin, especially in a 2 mg/kg dose to limit toxicity, may be an option,²³ although infliximab may turn out to be a more attractive therapy.²² Both of these therapies may require concurrent use of azathioprine or 6-mercaptopurine. Leucocyte apheresis is emerging as a further option in need of randomised controlled trial evidence. On current evidence, bolstered by the Milan results, the place of azathioprine in the UC management algorithm is probably similar to that in Crohn’s disease, although the quality of evidence in the latter is still better. Colectomy is curative of the disease in the former, although often associated with somewhat frequent bowel movements, pouchitis, reduction of fertility in females, and (infrequently) may result in a permanent stoma.

With the rapid emergence of new therapies in inflammatory bowel disease, it is heartening to see evidence for conventional therapies being consolidated, as designing correct management algorithms will depend entirely on the quality of available data. The old adage of “out with the old and in with the new!” should therefore be replaced with “consider the new but don’t forget the old (therapies)”.