Immunosuppressive drugs in ulcerative colitis: twisting facts to suit theories?

Abstract

Immunosuppressive drugs have become a mainstay of therapy for the inflammatory bowel diseases. Although robust evidence exists in support of the use of these drugs in Crohn's disease, a close evaluation of the available data in ulcerative colitis reveals a much weaker evidence base. In particular, randomised controlled trials of azathioprine, the most commonly used immunosuppressive agent, do not provide rich evidence of efficacy whereas observational cohorts suggest this agent is effective, particularly in patients with relapsing disease who require corticosteroids. Ciclosporin is also effective in the most refractory cases but its efficacy needs to be carefully weighed against the possibility of rare but life threatening complications. Although the evidence base in support of immunosuppressive drugs in ulcerative colitis is not as strong as in Crohn's disease, these agents clearly have a role in the treatment of this disease.

“I have no data yet. It is a capital mistake to theorise before one has data. Insensibly one begins to twist facts to suit theories, instead of theories to suit facts.” Sir Arthur Conan Doyle, A Scandal in Bohemia, 1891.

Starting with the first reported use of mercaptopurine for ulcerative colitis by Bean in 1962,¹ immunosuppressive drugs gradually have become an integral part of the treatment of ulcerative colitis. A survey of the membership of the British Society of Gastroenterology taken 10 years ago found that 93% of gastroenterology consultants had used azathioprine for ulcerative colitis, 86% in the previous year.² The theoretical basis of this widespread adoption of immune suppression for ulcerative colitis includes the well established knowledge that ulcerative colitis, like Crohn's disease, is a state of uncontrolled activation of mucosal immunity. Furthermore, given the well documented success of immune suppression in Crohn's disease, it is easy to believe that the same approach should work in ulcerative colitis. If this line of reasoning is correct, immunosuppressive drugs should prove to be excellent treatment for ulcerative colitis. And yet, careful review of the best available data regarding immune suppression—particularly the most widely adopted agents, azathioprine and mercaptopurine—reveals that the evidence base in support of immune suppression in ulcerative colitis is considerably weaker than in Crohn's disease. In treating ulcerative colitis with immunosuppressive drugs are we twisting facts to suit theories?

“The evidence base in support of immune suppression in ulcerative colitis is considerably weaker than in Crohn's disease”

First, how should we judge immunosuppressive therapies? With all that we have learned over the past five decades of the negative consequences of corticosteroids, few gastroenterologists will recall the profoundly positive impact of the landmark study of Truelove and Witts.³ This study, which unequivocally demonstrated the efficacy of corticosteroid therapy for ulcerative colitis, revolutionised the treatment of acute flares. Adoption of corticosteroids for severe flares of ulcerative colitis likely led to a reduction in the case fatality rate. In addition to demonstrating the remarkable short term efficacy of corticosteroids in severe flares of the disease, this randomised controlled trial also set the standard for the type of objective evidence one desires as a basis for altering practice.

The short term success of corticosteroids however has brought other needs. Beyond the well recognised adverse effects of corticosteroids, it is now understood that the very need for corticosteroids portends an unfavourable prognosis. In the Olmsted County population based cohort, 34% of ulcerative colitis patients who required corticosteroids had a one year prognosis that included corticosteroid dependence in 22% and colectomy in 29%.⁴ Therefore, studies of immunosuppressive therapy have correctly focused attention on those patients with more recalcitrant disease, particularly those who have required corticosteroids. Accordingly, immunosuppressive drugs have been judged by their ability to establish remission, prevent relapse, and limit exposure to corticosteroids.

“When it comes to the inflammatory bowel diseases however not all immunosuppressive drugs are alike”

When it comes to the inflammatory bowel diseases however not all immunosuppressive drugs are alike. Methotrexate, clearly proven to be efficacious in Crohn's disease, has never been demonstrated to be effective in a randomised controlled trial in ulcerative colitis. Some uncontrolled series do suggest that methotrexate may be of benefit in ulcerative colitis.⁵ However, a double blind, randomised multicentre trial of oral methotrexate in 67 patients with ulcerative colitis failed to demonstrate therapeutic benefit.⁶ It is possible that this negative study was the result of inconsistent bioavailability via the oral route of administration, or inadequate dose. One may keep an open, but sceptical, mind about the possibility that rigorous exploration of dose and route of administration might one day demonstrate the efficacy of methotrexate in ulcerative colitis. However, widespread adoption in clinical practice cannot be justified until efficacy is clearly demonstrated. Likewise, mycophenolate mofetil, effective in the setting of organ transplantation, was found to be inferior to azathioprine in a one year pilot study in ulcerative colitis.⁷

But what are we to make of the data on azathioprine itself? A summary of the seven published blinded placebo controlled trials of azathioprine in ulcerative colitis (see table 1) highlights some of the difficulties in interpreting these data.⁸,⁹,¹⁰,¹¹,¹²,¹³,¹⁴ No study was larger than 80 patients, and one included as few as 30 subjects. The likelihood of type II error in these studies is high, but the marked heterogeneity of populations, azathioprine dose, and end points precludes formal meta‐analysis. All but one study was performed at a single centre—two of the studies coming from one centre in India—raising the issue of generalisability. Study designs were also widely divergent, and seen through the lens of contemporary clinical trials in the inflammatory bowel diseases leave much to be desired. The studies of Rosenberg and colleagues⁹ and Kirk and Lennard‐Jones¹⁰ focused on the ability to taper corticosteroids, and adjusted corticosteroid doses based on clinical activity. From the current perspective, these studies solve the therapeutic equation for the wrong variable, and demonstrate that corticosteroid dose can be reduced while maintaining a steady state of disease activity. However, the advent of biological therapy for the inflammatory bowel diseases has raised the therapeutic bar, and holds the promise of stringent definitions of remission accompanied by discontinuation of corticosteroids as realistic and simultaneous therapeutic goals. Indeed, such end points had already been achieved using immunosuppressive drugs in Crohn's disease.¹⁵,¹⁶

Table 1 Summary of blinded, controlled trials of azathioprine in ulcerative colitis.

Ref	No/design/ duration	Description of population	Study drug regimen	Concomitant treatment	Primary analysis	Secondary analyses
Jewell, 19748	80/single centre, double blind/1 y	Outpatients (n = 61) Inpatients (n = 19) “First attack” (n = 31) “Relapse” (n = 49) Mild (n = 31), Moderate (n = 40) Severe (n = 7)	AZA 2.5 mg/kg (n = 40; dose reduced to 1.5−2.0 mg/kg in 2nd 40 patients randomised) v placebo (n = 40)	Outpatients: prednisolone 20 mg po QID + prednisolone retention enemas nightly for 4 weeks, taper over 2 weeks Inpatients: prednisolone‐21 phosphate 40 mg IV QD, tetracycline 1 g QD, hydrocortisone hemisuccinate sodium 100 mg BID rectal drip, followed by prednisolone 40 mg po QD with taper over 4 weeks	Number of relapses over 1 y (categorised as “0,” “1–2,” “3 or failed”), p = 0.18 for difference between AZA and placebo	Subgroup analysis of number of relapses over 1 y confined to subjects admitted in relapse, p = 0.055 for difference between azathioprine and placebo
Rosenberg 19759	30/single centre, double blind/6 months	Outpatients with disease extending beyond rectum and requiring at least 10 mg prednisone daily for control of symptoms at least 12 weeks prior to enrolment	AZA 1.5 mg/kg (n = 16) v placebo (n = 14)	Prednisone dose adjusted up or down every 2 to 3 weeks depending on whether symptoms were “satisfactorily controlled”	Prednisone dose at 6 months: 5.7 mg/day for AZA v 17.2 mg/day for placebo, p = 0.05	Number of stools per day, “state of health,” proctoscopy findings, grading of rectal biopsy findings, improvement on barium enema from baseline to end of study not different between treatment groups
Kirk 198210	44/single centre, double blind/6 months	Patients with “active chronic ulcerative colitis which was apparently unresponsive to corticosteroid treatment”	AZA 2–2.5 mg/kg (n = 24) v placebo (n = 20)	Prednisolone adjusted according to “clinical activity”	Per protocol analysis of “mean dose of prednisolone necessary to control the disease” at 6 months: 2.3 mg/day for AZA group v 13.6 mg/day for placebo group, p<0.001 (Does not include 15 subjects who were withdrawn for adverse effects or lack of compliance: 9 from AZA group, 6 from placebo group). Colectomy needed in 3 placebo subjects.	Change in Powell‐Tuck score at 6 months from baseline −3.9 for AZA (p<0.001) and −1.6 for placebo (NS) among patients completing study.
Hawthorne 199211	79/multicentre, double blind/1 y	Patients on AZA at least 6 months and in remission ⩾2 months (n = 67) or with chronic stable disease (defined as low grade symptoms or symptom control with doses of prednisolone 10 mg/day or less) and on AZA for at least 6 months (n = 12)	Continued AZA (n = 38; median dose 100 mg) v placebo (n = 41). (Randomisation stratified by “remission” v “chronic stable disease” groups)	5‐aminosalicylates maintained at stable doses (67 patients). Corticosteroid enemas (2 patients), oral prednisolone (4 patients) in the “chronic stable disease” group	Rate of relapse over 1 y (defined as worsening symptoms recognised by patient as active disease with a sigmoidoscopic appearance of grade 1 or more, or grade 2 or 3 appearance at routine sigmoidoscopy regardless of symptoms): 36% AZA group v 59% placebo group for the subjects randomised in “remission,” p = 0.04 “No benefit found” in the group with chronic stable disease.	Effect estimate unchanged in Cox proportional hazards model adjusting for effects of sex, age, duration of remission before trial entry, and treatment in the randomised in “remission” group.
Sood 200012	50/single centre, single blind/1 y	Asian Indian patients with severe disease who relapsed within 2 months of corticosteroid withdrawal after a course of hydrocortisone 100 mg IV every 6 h, parenteral ciprofloxacin and metronidazole for 7 days, followed by oral prednisolone 1 mg/kg tapered over 12 to 16 weeks and sulfasalazine 6–8 g/day.	AZA 2 mg/kg (n = 25) v placebo (n = 25)	Sulfasalazine 6–8 g/day, oral prednisolone 1 mg/kg/day tapered over 12–16 weeks	Proportion in remission at 1 y: AZA group 56% v placebo group 40% (NS)	Proportion entering remission after 4 months was significantly higher in the AZA group. Relapse rate by 1 y: 12% AZA v 24% placebo (NS)
Sood 200213	35/single centre, double blind/1 y	Asian Indian patients with newly diagnosed, severe disease (Seo index >220)	AZA 2.5 mg/kg (n = 17) v placebo (n = 18)	Hydrocortisone 100 mg IV every 8 h for 5 days, followed by prednisolone 1 mg/kg by mouth daily, tapered by 10 mg every 10 days to 20 mg/day, then by 5 mg every 10 days; sulfasalazine 6 g/day	Proportion relapsing (defined as worsening of symptoms recognised by patient as active disease with sigmoidoscopic appearance of active colitis) by 1 y, after achieving complete remission (defined as clinical improvement with absence of symptoms of active disease + Baron sigmoidoscopic grade 0–1 + normal histology + Seo index ⩽150): AZA group 23.5% relapsing v placebo group 55.6% relapsing, p = 0.05	Higher mean activity index in placebo group compared to AZA group over 1 y (p<0.001)
Ardizzone 200614	72/single centre, investigator blind/6 months	Steroid dependent patients needing ⩾10 mg/day prednisolone and unable to wean off steroid without relapse at least twice in the previous 6 months, with Powell‐Tuck index ⩾8 and Baron index ⩾2	AZA 2 mg/kg (n = 36) v oral 5‐aminosalicylate 3.2 g/day (n = 36)	Prednisolone 40 mg/day for 2 weeks, then 30 mg/day for 2 weeks, then 20 mg/day for 4 weeks, then taper by 5 mg weekly until withdrawal or relapse	Success, defined as induction of clinical remission (Powell‐Tuck index of 0) and endoscopic remission (Baron index of 0 or 1): AZA group 53% v 5‐aminosalicylate group 19% (p = 0.006)	Mild to moderate adverse events more common in AZA (26%) v 5‐aminosalicylate (6%) (p = 0.046)

UC, ulcerative colitis; AZA azathioprine; IV, intravenous; po, orally; BID, twice daily; QID, four times daily; QD, once daily.

In light of these results, the outcomes for azathioprine in ulcerative colitis until recently were all the more disappointing. The study by Hawthorne and colleagues bears some similarity in design to recent maintenance studies of infliximab¹⁷,¹⁸ in its selection of patients responsive to the drug under study, and then randomising such patients to continued treatment or placebo while following for relapse. Such a withdrawal design obviously magnifies a treatment effect in that it selects those patients who in the first instance respond to the intervention of interest. However, the study by Sood and colleagues¹² from 2000—until recently the only one with a design straightforward enough to provide an estimate of the treatment effect directly attributable to azathioprine for severe disease relapsing after corticosteroid taper—suggests that the absolute benefit of azathioprine over placebo for remission approximates 15% at one year.¹² In short, the randomised controlled trials of azathioprine in ulcerative colitis all hint of some therapeutic effect, but the data suggest this effect may not be robust.

More recently, the study by Ardizzone and colleagues¹⁴ demonstrated in a straightforward manner the clearcut efficacy of azathioprine in treating patients with steroid dependent ulcerative colitis. This investigator blinded study showed that at six months, 53% of patients on azathioprine were well off steroids compared with only 19% of patients assigned to oral 5‐aminosalicylate 3.2 g/day. Although mild to moderate adverse events were more frequent in azathioprine treated patients (26% v 6%; p = 0.046), only two of 36 patients on azathioprine were withdrawn because of adverse events.

“These real world data provide compelling evidence of the utility of immunosuppressive therapy in ulcerative colitis, specifically of azathioprine”

Heretofore, it would appear that the paucity of data available from blinded randomised controlled trials, in comparison with Crohn's disease, was misleading. In fact, the preponderance of evidence in support of the thiopurine immunosuppressive drugs comes from observational cohorts in retrospective series. Published series range in size from 16 to 346 patients.¹⁹,²⁰,²¹,²²,²³,²⁴,²⁵,²⁶ By far the best among these is the 30 year cohort assembled at the Oxford IBD clinic between 1968 and 1999.²⁶ In this series, the overall remission rate in the 346 patients with ulcerative colitis who were treated with azathioprine was 58%, but increased to 87% among patients on therapy for more than six months. In fact, these remission rates were higher than those observed among patients with Crohn's disease in the same series, with an overall remission rate of 45%, and a 64% remission rate for Crohn's patients on therapy for more than six months. It is also notable that the mean dose at which remission was achieved was 1.65 mg/kg, somewhat lower than the 2–2.5 mg/kg often advocated as being effective. The maintenance effect appeared to be robust, with the proportion of patients in remission at five years being 62%, applying a strict definition of relapse, and 81% allowing for brief relapse with a short corticosteroid taper and recovery of remission. Finally, median time to relapse on stopping azathioprine was 18 months,²⁶ suggesting a return to the expected natural history on withdrawal of the drug. These real world data provide compelling evidence of the utility of immunosuppressive therapy in ulcerative colitis, specifically of azathioprine, and go far beyond what we might ever expect to learn from a randomised controlled trial regarding the durability of response.

The efficacy of ciclosporin can also be brought to bear on the question of the utility of immunosuppressive drugs in ulcerative colitis. The study of Lichtiger et al unequivocally demonstrated a short term response in 75% of patients treated with intravenous ciclosporin who have severe ulcerative colitis refractory to intravenous corticosteroids.²⁷ Case series from other centres have confirmed efficacy, although estimates of short term response have varied from 56% to 91%.²⁸,²⁹,³⁰,³¹ The largest among these series suggests that 55% of patients treated with intravenous ciclosporin will avoid colectomy by three years.³⁰ Studies further suggest that transition to azathioprine provides improved long term avoidance of colectomy.²⁸,²⁹,³⁰,³¹,³² Despite these data, there has never been widespread adoption of ciclosporin by gastroenterologists outside of large referral centres largely due to concern over the drug's toxicity profile and uneasiness over the risk of serious potentially fatal complications, including opportunistic infection.³⁰ Lower doses of intravenous ciclosporin, in the range of 2 mg/kg/day rather than the 4 mg/kg/day used in the Lichtiger study, have been touted as equally effective and potentially less toxic.³³ However, the lesser toxicity of these lower doses has not been explored in larger cohorts.

“There has never been widespread adoption of ciclosporin by gastroenterologists outside of large referral centres”

Ultimately, the safety profile of immunosuppressive drugs must be weighed against their direct and indirect effects on long term outcomes, such as functional status, the need for colectomy, incidence of colorectal cancer, and death from complications of the disease. As the stakes rise, and both efficacy and safety concerns increase, the calculus becomes more complex. Appropriate selection of patients for whom the benefits outweigh the risks of treatment is essential for the proper use of these drugs. The data at hand would suggest that the need for steroids, particularly for patients in relapse, is an appropriate threshold for the use of immunosuppressive drugs.

While the evidence for immunosuppressive drugs in ulcerative colitis remains relatively sparse, there is growing strength. It would be incorrect to say that the data are insufficient to support the prominent role these agents have taken in the treatment of ulcerative colitis. In short, the facts in this case suit the theories. Future studies will need to focus on the role of these agents as alternatives to or in combination with biological response modifiers such as infliximab.

Conflict of interest: declared (the declaration can be viewed on the Gut website at http://www.gutjnl.com/supplemental).