Low-dose azathioprine or mercaptopurine in combination with allopurinol can bypass many adverse drug reactions in patients with inflammatory bowel disease


Dr A. Ansari or T. H. J. Florin, Department of Gastroenterology, East Surrey Hospital, Redhill, UK.
E-mail: azharansari@hotmail.com or t.florin@uq.edu.au


Aliment Pharmacol Ther31, 640–647


Background  The thiopurine drugs, azathioprine and mercaptopurine (MP), are established treatments for IBD. However, therapeutic failure caused by adverse drug reactions occurs frequently.

Aim  To study combination of allopurinol with reduced-dose thiopurine in an attempt to avoid adverse drug reactions in the treatment of IBD.

Methods  Patients with drug reactions to full-dose thiopurines were recruited for combination therapy in two IBD centres in this retrospective study. Dosing was guided by measuring thiopurine methyltransferase (for UK patients) or thioguanine nucleotides and methyl-6MP (Australian patients). Response was monitored by clinical activity indices.

Results  Of 41 patients, 25 had non-hepatic and 16 had hepatitic reactions. Clinical remission was achieved in 32 patients (78%) with a median follow-up of 41 weeks (range 0.5–400). Patients who did not respond to combination therapy tended to fail early with the same adverse reaction. The relative risk of having an adverse reaction with methyl-6MP in the top interquartile range was 2.7 (1.3–28) times that with methyl-6MP in the lower three quartiles (95% confidence interval).

Conclusion  The combined experience from our centres is the largest reported experience of this combination therapy strategy in IBD, and the first to provide evidence for benefit in thiopurine and allopurinol co-therapy to avoid non-hepatitic adverse drug reactions.


The thiopurines, azathioprine (AZA) and mercaptopurine (MP), are well-accepted treatments for inducing and maintaining remission in moderate-to-severe inflammatory bowel disease (IBD), but they have a slow onset of action. Furthermore, therapeutic failure caused by poor response or adverse drug reactions (ADRs) occurs in 15–40%.1–3 Once a decision is made to employ thiopurines, a clinician must usually support an acutely unwell patient for at least a further period of 12–16 weeks with steroids, calcineurin inhibitors or anti-TNFα therapy before he can determine a response. This period will be extended if there is a sub-optimal response requiring dose escalation or alternative treatments, whilst ADRs will also necessitate other treatments. Failure of response to AZA/MP may oblige the physician to prescribe less-established immunosuppressive drugs [methotrexate, thioguanine (tioguanine) or mycophenolate], more expensive parenteral or oral calcineurin inhibitors, or biologics, or surgery. Therefore, avoidance of poor response or side effects will minimize morbidity from steroid exposure, requirement for other acute treatments or continuing disease activity in some patients.

Allopurinol was originally designed for use as a thiopurine co-therapy to reduce the first pass metabolism of MP by inhibiting small intestinal xanthine oxidase (XOD), resulting in improved bioavailability.4 However, early trials with allopurinol co-therapy proved more toxic than efficacious because of the narrow therapeutic index of thiopurines. Allopurinol then took on its more established role in the treatment of gout.

Recently, a therapeutic strategy of combining allopurinol with reduced-dose thiopurine, initially used in renal transplantation,5 has been used in IBD for poor thiopurine responders, where it has been shown to improve disease outcome,6–8 reduce biochemical hepatotoxicity6, 8 and provide safe long-term immunosuppression.7, 8 The precise mechanism by which thiopurine-allopurinol co-therapy works still remains speculative with many questions remaining unanswered. In the only in vivo human study on the subject to date, we have raised doubts about the influence of first-pass metabolism by XOD on low-dose thiopurines.9 The rule of thumb when using the thiopurine-allopurinol combination is to reduce the thiopurine dose to about 25–33% of the otherwise intended dose.5 Increasing experience with the low-dose thiopurine–allopurinol combination has shown that patients can be rescued from poor response or biochemical hepatitis. However, knowledge is lacking about its use to overcome other nonhepatitic thiopurine ADR.

There is now substantial evidence pointing to unwanted side-effects of AZA/MP being associated with methylated metabolites, with preferential methylation and high thiopurine methyltransferase (TPMT) activity predicting biochemical hepatotoxicity.10, 11 We hypothesized that minimizing methylated products by using a reduced-dose AZA/MP-allopurinol combination therapy, should avoid ADR in addition to that of hepatitis, whilst achieving sufficient TGN concentrations to effect a treatment response.1 In this study, we report our experience with combination of reduced-dose AZA/MP-allopurinol in consecutive patients with both primary hepatitic and non-hepatitic thiopurine-related ADR from two IBD centres.



A chart or database review was performed on all patients prescribed combination allopurinol with low-dose thiopurine therapy in two different centres in England and Australia for ADR to either AZA or MP or both between 2001 (England) or 2005 (Australia), and 1st August 2009. We used the strategy first proposed by Chocair et al.5 when using the thiopurine-allopurinol combination, which was to reduce the thiopurine dose to between 25% and 33% of the otherwise intended monotherapy dose. Concomitant allopurinol was dosed at 200 mg/day in England or 100 mg/day with thiopurine in Australia. Patients deemed by the investigators to have an allergy-related ADR, such as pancreatitis, were not offered the combination therapy.

In England, TPMT activity, which was measured by mass spectrometry,12 was used to inform upfront optimal thiopurine dosing (see Table 1 for dosing schedule). TPMT activity was measured once only before commencement of thiopurine therapy in the patients known to author ARA. The patients attended London hospitals (Guy’s and St Thomas’ Hospital, University Hospital Lewisham, Chelsea and Westminster Hospital and East Surrey Hospital, England). Patients having an ADR to AZA or MP thiopurine monotherapy, received a test dose for no more than one week of the same low-dose thiopurine monotherapy to exclude an allergy-related ADR and, if tolerated, were commenced on combination allopurinol and a low dose of the same thiopurine therapy.

Table 1.   Dosing regimen based on TPMT assay in London. Median TPMT was 32 pmol/h/mg/Hb12
TPMTConventional AZA DoseAZA & allopurinol co-therapy
  1. Mercaptopurine is approximately 50% of the molecular weight of AZA and was used at 50% of the appropriate AZA dose.

10–251 mg/kg0.25–0.33 mg/kg
26–352 mg/kg0.5–0.7 mg/kg
>352.5–3 mg/kg0.5–0.7 mg/kg

In the Australian centre, the patients known to investigator THJF, attended the Mater Health Services (MHS) in Brisbane, Australia, which maintains a web-based prospective IBD database, which includes clinical activity scores. In his practice, therapeutic drug monitoring and measurement of TPMT activity was not routine. Hence, MP was commenced at 1–1.5 mg/kg/day or AZA at 2 mg/kg/day, without prior therapeutic drug monitoring or TPMT testing. In patients thought to have an allergy-related ADR, the drug was temporarily stopped before re-challenge at a low dose of the same thiopurine. Patients on low-dose AZA with a recommencement of a non-pancreatitis ADR were switched to low-dose MP. Patients with a prompt ADR with recommencement of lower dose MP were not offered the combination therapy. Erythrocyte thioguanine nucleotides (TGNs) and methyl-6MP (me6MP),13 which were measured by Queensland Pathology Services using the method of Dervieux,14 were used to inform the optimal dosing in those patients who were not responding to an initial (lower) dose of thiopurine, and in those patients who had ADR. Drug metabolites were re-measured at 4 weeks in patients who were prescribed the combination therapy.

Ethics, monitoring and compliance with treatment

The results presented here arose from our normal clinical practice and as such, ethical permission was neither sought nor required. All patients in both countries received detailed verbal information about thiopurine therapy. In particular, patients were told that low-dose thiopurine-allopurinol co-therapy was commonly used in renal transplantation5 and from 2005 they were told that it had found a place as a rescue regime in IBD for poor thiopurine responders who developed drug-induced hepatitis.6 Adherence to treatment was confirmed verbally at each visit in both IBD centres and by the TGN assay at week 4 at the MHS IBD centre in Australia. Regular full blood counts and liver tests were performed before and after starting AZA or MP. Patients’ clinical response was judged by calculating the Harvey Bradshaw Index (HBI)15 for Crohn’s disease, and the Simple Clinical Colitis Activity Index (SCCAI)16 for ulcerative colitis. Clinical remission was defined at both centres by HBI ≤ 4 or SCCAI ≤ 3 and the absence of corticosteroid therapy for more than 12 weeks.


Two-way contingency table analysis was performed on ADR-me6MP data. A paired t-test was used to compare data before and after the combination intervention. Between centre data on patient ages and duration of treatments were compared non-parametrically using SPSS 17.01 (Sydney, Australia) for MAC.


Forty-one consecutive patients are reported in this retrospective two-centre experience of reduced dose thiopurine-allopurinol co-therapy (Table 2, Table 3). The ratio of male to female patients was 21:20. Median age (range) of patients, which was 34 (16–80) years, was not different between centres. Twenty-nine of the 41 patients were tried on one only of AZA (N = 20) or MP (N = 9) before attempting a reduced dose of the same thiopurine-allopurinol co-therapy. Twelve of the patients from the Brisbane centre were tried on both AZA and MP monotherapy before attempting reduced dose thiopurine-allopurinol co-therapy.

Table 2.   London Inflammatory Bowel Disease Centre
Patient coded numberDiagnosisSexAge (years)ALTTPMTDuration of AZA/MP monotherapy (weeks)Sentinel events on AZA/MP monotherapyDuration of AZA/MP & ALLO (weeks)Outcome on AZA/MP & ALLO combination Rx
  1. TP, thiopurine; ALLO, allopurinol; CR, clinical remission; PR, partial response (PR = reduced HBI or SCCAI scores but HBI > 4, SCCAI > 3 with TP-ALLO, or requiring prednisolone).

 1UC E3S2M5610432Myalgia fatigue6Headaches/stopped TP
 2UC E3S2F6823394Nausea4Nausea/stopped TP
 3UC E2S2M3313376Nausea fatigue 50CR
 4UC E3S3F3641517Myalgia fatigue42CR then relapse with poor Rx compliance
 5UC E3S3M5332481Nausea37CR
 6UC E3S3M2830432Arthralgia nausea41CR
 7UC E1S2F7723361Nausea28PR
 9CD L2B1M2629318Nausea38CR
10CD L2B1M2631332Myalgia nausea41CR
11CD L3B1F4316283Fatigue49CR
12CD L2B1F4725452Nausea29CR
13CD L3B2M3322372Myalgia fatigue43CR
14UC E1S2M5820408Nausea24CR
15UC E3S3M78145372Hepatitis5Hepatitis (mixed pattern)/TP stopped
16CD L3B2F48230340.5Hepatitis44CR
17UC E2S2F3489356Hepatitis400CR
18UC E2S2M3379416Hepatitis nausea8GI intolerance/stopped TP
19CD L3B3M50130443Hepatitis284CR
20UC E3S3M4684546Hepatitis160CR
21CD L3B2F37146382Hepatitis104CR
22UC E1S3M6991423Hepatitis224CR
23UC E1S2F80164232Hepatitis nausea128CR
24UC E2S3M45182215Hepatitis nausea216CR
25UC E3S2F21196206Hepatitis20CR
26UC E3S3F4278356Hepatitis41CR
Table 3.   Brisbane Inflammatory Bowel Disease Centre
Patient coded numberDiagnosisSexAge (years)ALTMe6MPDuration of AZA/MP monotherapy (weeks)Sentinel events on AZA/MP monotherapyDuration of AZA/MP & ALLO (weeks)Outcome on AZA/MP & ALLO combination Rx
  1. TP, thiopurine; ALLO, allopurinol; CR, clinical remission.

27CDPerianalM389230054Nausea fatigue89Mild nausea/CR
28CD L2B1M25211450012Fatigue65CR
29UC E2S3F1618480016Nausea fatigue48CR
30CD L2B1M22111471Pleuritic chest pain0.5Chest pain/stopped TP
31CD L2B1M2822124003Fatigue myalgia205CR
32CD L1B1F1920319362Fatigue76CR
33UC E2S3F57142902416Fatigue myalgia33CR
34UC E2S2F2319429512Nausea20CR
35CD L2B2F30205899125Fatigue56CR
36UC E3S3F2417196717Nausea3Nausea/stopped TP
37CD L1B3F58302055034Fatigue45CR
38UC E3S1M235717746Hepatitis35CR
39CD L3B1F3071219006Hepatitis33CR
40CD L1B1F21541975215Hepatitis nausea40CR

Other patients were not offered reduced dose thiopurine-allopurinol co-therapy for a number of reasons. Twelve patients who were prescribed both AZA and MP, were not offered the combination therapy: five patients in the Brisbane centre intolerant of AZA, were tolerant of MP, and 7 AZA-treated patients, also from the Brisbane centre, who were subsequently intolerant when switched to lower dose MP, were diagnosed with non-pancreatitis allergic ADR. Three patients with allergic pancreatitis (one AZA, two MP) were not offered thiopurine switching or reduced dose thiopurine-allopurinol combination therapy.


With thiopurine monotherapy, non-hepatitic primary non-allergic ADRs (N = 25), which were predominantly nausea and/or fatigue or myalgia, were more common than primary hepatitic ADRs (N = 16) at both centres. Hepatitis defined as alanine transferase (ALT) greater than two times the upper limit of normal in London or 1.5 times the upper limit of normal in Brisbane was associated with mainly nausea, but was often asymptomatic. Elevated me6MP, measured in Brisbane, was more frequent than expected by chance in patients with ADRs. The relative risk of having an ADR with me6MP in the top interquartile range (me6MP >4720) was 2.7 (1.3–28) times that of an ADR with me6MP in the lower three quartiles (95% confidence interval).

Combination therapy

Patients were followed on low-dose thiopurine-allopurinol co-therapy until a censored event either being clinical assessment of the co-therapy on the 1st August 2009 or failure with the co-therapy at an earlier time point. In the Brisbane centre, mean TGN on monotherapy was 370 (135–802) and on combination therapy 563 (310–1230) pmol/8 × 108 erythrocytes (P = 0.006); mean me6MP on monotherapy was 11604 (147–29024) and on combination therapy 696 (109–3503) pmol/8 × 108 erythrocytes (P < 0.001); (the metabolites were not measured on combination therapy in patient #30). There were no cases of clinically significant leukopenia or lymphopenia using our prescribing strategy of upfront TPMT assay in London, regular blood counts, on-demand thiopurine drug monitoring in Brisbane, and 25–33% dose reduction of thiopurines with allopurinol co-therapy.5 There were no ADR in this series of 41 patients, which could be ascribed to allopurinol.

Clinical remission (CR) was achieved in 18/25 non-hepatitic ADRs (72%) and 14/16 primary hepatitic ADRs (88%) at time of censoring. Liver biochemistry improved in all hepatitic patients in the London centre and in three of four hepatitic patients when measured at 4 weeks in the Brisbane group. Mean ALT in the 41 patients fell from 59 to 23 (P < 0.001) and in the 16 hepatitic patients from 117 to 43 IU (P < 0.001). Overall, 32 of this selected group of 41 patients (78%) were in CR at time of censoring data. Median duration of this follow-up (range) was 41 (0.5–400) weeks. Patients who did not respond to the combination therapy tended to fail early, median 4.5 (0.5–8) weeks, and with the same ADR.


We report an experience with low-dose thiopurine-allopurinol co-therapy in 41 IBD patients who failed thiopurine monotherapy because of thiopurine-related non-allergic ADR. They were followed up for a median 41 weeks. The combined experience from our centres is, to the best of our knowledge, the largest reported experience of the combination therapy. Smaller studies have reported on the combination therapy for poor responders to thiopurine monotherapy or for patients with biochemical hepatitis ascribed to the monotherapy.6–8, 17, 18 Our report is the first to provide evidence that there is benefit in AZA/MP and allopurinol co-therapy to avoid non-hepatitic ADRs.

Adverse drug reactions – more than half non-hepatitic – were circumvented in nearly four-fifths of patients in this experience. The non-hepatitic ADRs treated in the study included myalgia, nausea and fatigue symptoms. Our post-combination therapy data showed a consistent improvement in ALT, which was demonstrated previously.7, 8 The ADR descriptors were not exclusive with some patients reporting more than one symptom. In general, our protocols avoided enrolling patients with allergic ADR, like pancreatitis, because treatment with reduced dose thiopurines avoided the ADRs. (Patient #30 had, in retrospect an allergic ADR to the thiopurine therapy, as the reduced dose thiopurine-allopurinol combination also brought on his symptom of chest pain). Our patient group did not experience clinically significant leukopenia or lymphopenia using our prescribing strategies for thiopurine monotherapy and, for where there was an ADR, combination allopurinol-thiopurine therapy. However, higher me6MP has been found to correlate with leukopenia. This ADR appears to occur at a later time point than the other thiopurine ADRs.19 The higher TGN, mean 563 (v 370) pmol/8 × 108 erythrocytes, with much reduced me6MP on combination therapy without leukopenia in our study, is consistent with the reported toxic effect of me6MP on white cells. It also leads us to suggest that with therapeutic drug monitoring of combination allopurinol with low-dose thiopurine therapy, the target TGN range should be higher, but of course, it is not a substitute for judicious monitoring of white blood cell counts.

Allergy-related ADR do not have a standard method of treatment other than the usual withdrawal of drug therapy, and as such their handling was different between the two centres. Patients were treated according to local practice. The two centres also used different approaches to guide safe and efficacious therapy with thiopurines and used different doses of allopurinol. Furthermore, the IBD centres did not preferentially employ AZA and MP according to a protocol, but MP monotherapy was tried and failed in 19/24 AZA-related ADR cases reported in this study. Although it has been shown to be helpful,19 this interchange strategy between AZA and MP to circumvent side effects or lack of response to thiopurines20, 21 appeared less useful than the combination therapy in our uncontrolled study. Seven patients were also excluded from the combination therapy because of non-pancreatitis allergic ADR to both AZA and MP. While the AZA-MP interchange strategy has been reported in patients with AZA-related pancreatitis,22, 23 it was not attempted in the three patients with allergic pancreatitis in this study.

Previously described predictive markers of poor response to thiopurines include low TGNs24, 25 and high TPMT activity.1, 26 Relling first proposed the idea that high TPMT activity and methylated metabolites of MP may cause hepatotoxicity,27 although toxicity by me6MP metabolites was already suspected from early trials of methylmercaptopurine riboside, which caused significant hepatotoxicity as a treatment for acute leukaemia.28–30 Later, high me6MP levels were correlated to biochemical hepatotoxicity in the treatment of IBD.10 As far as we are aware, the present study is the only study to report an association between high me6MP and non-hepatitic side effects.

Although our experience is consistent with the previous studies that associate high TPMT and me6MP with toxicity and poor response, we stress that the ADR may not all be ascribable to me6MP. The benefit of allopurinol may also relate to lower dose of AZA/MP or the anti-oxidant action of allopurinol.13 Furthermore, from a clinical viewpoint, the correlations as indicated by our own TPMT and metabolite results are not so strong as to recommend exclusive reliance on these parameters to inform dosing and combination therapy because the sensitivity and specificity are poor.

Allopurinol was designed to improve the therapeutic index of MP, an effect which was demonstrated in mice, but not in humans who suffered a proportional increase in toxicity, but no improvement in the desired therapeutic index.31, 32 Almost five decades after it was synthesized, allopurinol appears to be fulfilling its inventors’ goal of improving the response to thiopurines, albeit in disorders other than leukaemia.4, 33 Our group has used both the lowest allopurinol dose (25 mg) in renal transplant patients,5 the intermediate dose of 100 mg/day6, 7 and 200 mg/d8 in combination with thiopurines. Although allopurinol’s mode of action in this co-therapy strategy and its optimal effective dose are yet to be determined, this two centre case series report from two hemispheres provides evidence to suggest that co-therapy may salvage a very significant number of patients with a range of non-allergic ADR to AZA/MP. It provides further support of efficacy of this co-therapy in patients with AZA/MP-related hepatitis or nonresponse. It is the first time that this strategy has been reported as a means to circumvent non-hepatitic AZA/MP-related ADRs. Our combined experience leads us to suggest that prior TPMT activity testing can help physicians correctly dose AZA or MP with or without allopurinol, and me6MP metabolite measurements correlate with ADRs. While neither can be relied upon exclusively at this time as an investigation to deploy allopurinol from the outset or to avoid ADRs or poor response, the use of the combination therapy and these pharmacokinetic measurements merits further study in controlled trials.


Declaration of personal interests: THJF has served on various IBD Advisory Boards in the last 2 years. Declaration of funding interests: THJF is a recipient of an NHMRC Practitioner Fellowship. The Australian web-based IBD database has been supported by equal grants-in-aid from Ferring Pharmaceuticals Pty Ltd, Orphan Australia Pty Ltd, and Pharmatel Fresenius Kabi Pty Ltd.