Prospective study of the effects of concomitant medications on thiopurine metabolism in inflammatory bowel disease


Dr M. Daperno, S.C. Gastroenterologia, A.O. Ordine Mauriziano, Largo Turati 62, 10128 Torino, Italy.


Aliment Pharmacol Ther30, 843–853


Background  Thiopurines are increasingly used in the treatment of inflammatory bowel disease (IBD), being the most common immunosuppressive therapy; however, potentially harmful interactions between thiopurines and other drugs (especially 5-aminosalicylic acid, 5-ASA) were described.

Aim  To explore potential interactions between thiopurines and concomitant medications.

Methods  A total of 183 consecutive IBD patients were enrolled. Clinical characteristics and concomitant medications were recorded. Thiopurine metabolism was analysed with thiopurine S-methyl transferase (TPMT) genetic variants and enzyme activity assays. Comparisons were carried out with stratification of patients according to clinical characteristics and active treatments.

Results  Based on TPMT genetics, 95% IBD patients were wild-type homozygous, the remaining being heterozygous. Median TPMT activity was 24.9 U/Hgb g (IQR 20.7–29.5). No difference in TPMT activity was noted according to 5-ASA exposure. IBD patients on thiopurines had higher TPMT activity levels, but no dose-effect was evident. No difference in TPMT activity was observed in 41 (63%) patients co-treated with 5-ASA. In patients on active thiopurines also, 6-TGN and 6-MMP levels were evaluated and no significant difference was observed based on co-medication. TPMT activity was independently associated only with thiopurines dose (= 0.016).

Conclusions  Our data suggest the absence of significant interactions between thiopurines and 5-ASA.


Thiopurines (azathioprine and mercaptopurine) are extensively used in clinical practice and their use in inflammatory bowel disease (IBD) has increased over the last decades.1, 2 The effect on clinical outcomes is still a matter of debate, but current guidelines and everyday routine clinical practice include azathioprine or mercaptopurine among the most commonly used drugs, especially because of their steroid-sparing effect.3–6 Moreover, their use, together with infliximab, has been advocated and they are therefore an essential therapeutic weapon.7

Metabolism of azathioprine is quite well-known and it includes, after a non-enzymatic step leading to mercaptopurine formation, several enzymatic transformations leading to formation of active metabolites, of toxic intermediate molecules and finally to complete catabolism of the drug. One of the most important steps in thiopurine metabolism is under thiopurine S-methyl-transferase (TPMT, EC control. TPMT is a cytosolic enzyme which catalyses the S-methylation of aromatic and heterocyclic sulphydril compounds.8 This enzyme represents one of the most striking examples of the potential of pharmacogenetics to contribute to personalize drug therapy with 89–94% of individuals having high TPMT activity, 6–11% having intermediate activity and approximately 0.3% with extremely low or absent TPMT activity.9 Family studies have shown that TPMT activity is inherited as an autosomal codominant trait, with at least 21 TPMT genetic polymorphisms identified, which may be associated with decreased levels of TPMT enzyme activity and/or thiopurine drug-induced toxicity.10 Many population studies were carried out to identify the predominant variant alleles and, on the basis of these researches, TPMT*3A and *3C are the predominant variant alleles, with *2 contributing to a lesser extent. These three alleles account for over 95% of cases of inherited TPMT deficiency in Caucasian subjects.11

Five-aminosalicylic acid (5-ASA) is very commonly used in inflammatory bowel disease. Its role is essential in ulcerative colitis treatment, both in induction and in maintenance of remission;6 moreover, despite its limited efficacy, 5-ASA is used also in Crohn’s disease, especially for prevention of post-surgical relapse.5 Finally, another therapeutic goal, which justifies a wide use of 5-ASA, is the potential chemopreventive role in IBD colitis.12 A recent population-based study reported an overall prevalence of oral 5-ASA use in up to 75% IBD patients in Northern Italy.13

Some Literature data showed that 5-ASA may lower TPMT activity, thus leading to increased bioavailability of upstream metabolites, with potential toxic effects and/or potential therapeutic gain.14–16In vitro experiments reported substantial inhibition of TPMT activity or effects on thiopurine active metabolites by sulphasalazine or 5-ASA.17–19 Consistent with these in vitro data, a few clinical reports suggested that there may be a significant interaction between 5-ASA and TPMT also in vivo and such an interaction may be used to maximize thiopurine therapeutic window.20–22

The aims of the study were to analyse in a consecutive series of IBD patients:

  • (i) overall TPMT activity and
  • (ii) the effects of thiopurine, 5-ASA and other concomitant IBD medications on TPMT activity and on metabolites levels

to evaluate relevant interactions clinically.

Patients and methods

Patient population

A consecutive series of patients affected by inflammatory bowel disease attending a single tertiary referral centre IBD outpatient clinic were proposed to take part in a study approved by the local Ethical Committee (ref. n. 5838). Diagnoses were established according to commonly accepted criteria23 at least 6 months prior to study enrolment.

In patients agreeing to participate to the study, after signing a specific informed consent, peripheral blood samples were taken (2 EDTA 9 mL vials) and stored at room temperature until processed within 3–6 h.

Patient history, present symptoms and active treatments were recorded by the treating physicians and a detailed clinical history was collected on a dedicated case report form (CRF). All patients were classified regarding disease behaviour and location/extension according to Montreal classification;24 previous use of steroids and possible steroid-dependency/refractoriness, present or past use of thiopurine analogues, past adverse reaction to such drugs and present drug use were also recorded. Moreover, a diary was administered for CDAI/CAI25–27 calculation, depending on Crohn’s disease or ulcerative colitis/indeterminate colitis diagnosis.

A consecutive cohort of healthy subjects were recruited by the Blood Bank among blood donors (only gender and age were recorded), after excluding using a questionnaire, possible IBD or familiar history of IBD; healthy subjects who signed the informed consent were used as controls. All control subjects were routinely screened for infectious and transmissible diseases according to commonly accepted rules.

TPMT activity assay

Erythrocyte lysate preparation was similar to that previously described28 and the lysates were kept at −80 °C until analysis. The erythrocyte lysates were analysed for TPMT activity by the assay described by29 with minor modifications. Samples were assayed in duplicate.

Thiopurine S-methyl transferase activity was determined by measuring the formation rate of 6-methylmercaptopurine (6-MMP) from MP and using S-adenosyl-L-methionine (SAM) as methyl donor. One unit of enzyme activity represents the formation of 1 nmol of 6-MMP per hour of incubation and erythrocyte TPMT activity was expressed as units per gram of haemoglobin (U/Hgb g).

Briefly, the enzymatic reaction was stopped after 60 min at 37 °C by adding 25 μL perchloric acid. After vortex stirring and centrifugation, the amount of 6-MMP formed during incubation was measured by High Pressure Liquid Chromatography (HPLC): 30 μL supernatant was injected into the chromatographic system.29

The mean inter-days coefficient of variation of our internal quality controls ranged from 3.4% (low concentration) to 2.6% (high concentration) and the mean intra-day coefficient of variation ranged from 3.2% to 2.3%.

6-TGN and 6-MMP dosage

In patients on thiopurines treatment, analysis of thiopurine metabolites was carried out. The simultaneous determination in erythrocytes of 6-tioguanine nucleotides (6-TGN) and 6-methylmercaptopurine (6-MMP) was performed by the HPLC assay described elsewhere30 with minor modifications. Plasma was decanted and the leucocytes and the upper layer of erythrocytes were removed. Haematocrit and erythrocyte counts were obtained from each sample. Results were expressed as pmol per 800 millions red blood cells (pmol/8 × 108 RBC). Briefly, after deproteinization of the samples by perchloric acid with dithiothreitol, the nucleotides were hydrolysed to their own bases by heating the sample for 45 min at 100 °C. After cooling, a 100-μL aliquot was injected into the column. All assays were run in duplicate.

DNA Extraction, genotyping assays and sequence analysis

DNA was isolated from whole blood using the FlexiGene DNA kit (Qiagen, Milano, Italy). Genotyping assays were carried out by polymerase chain reaction (PCR)-restriction fragment length polymorphism/allele-specific PCR as previously described.31, 32 All 943 subjects were genotyped for the most common polymorphisms in Caucasian population, TPMT*2, *3A, *3B and *3C and alleles without any of the assayed mutations were assumed to be the TPMT wild type gene (TPMT*1).9 The resulting PCR products and digested fragments were separated and detected in ethidium bromide-containing 2% agarose gel. Sample with known sequence variation in exon 5, kindly provided by Prof. William E. Evans, was used as control for TPMT*2 mutation. Sequence analysis was performed to confirm the presence of two mutations, G460A and A719G, in our heterozygous and homozygous mutant samples. Samples were sent to Biotech Company (MWG, Ebersberg, Germany) and the sequence of the TPMT exons 7 and 10 was performed.


Clinical and biological data were entered on a standard MS Excel spreadsheet and data analysis was performed using medcalc statistical software (version 10.0.2, Mariakerke, Belgium).

Descriptive statistics were used to summarize data, categorical variables were reported with frequencies, while for continuous variables, medians and interquartile ranges were used.

Subgroup analyses were conducted with comparison between groups; for normally distributed variables, chi-square and Student’s t-tests were used, when distribution was expected to be non-normal, Mann–Whitney or Kruskal–Wallis test was used for continuous variables, as appropriate.

To test independent contribution of different variables to a dichotomous variable, stepwise logistic regression including all covariates with a P value equal to or greater than 0.10 at univariate analysis was applied. For continuous variables, stepwise multiple regression including all covariates with a P value equal to or greater than 0.10 at univariate analysis was applied.

A P value <0.05 was considered significant.

To outline better the relevance of the observed differences, for dichotomous variables, also Odds Ratio with 95% confidence intervals is reported (OR, 95%CI), where appropriate.

As TPMT activity may be affected by TPMT mutated genotype, all analyses regarding TPMT activity and active metabolites were carried out on populations with wild-type TPMT genotype to exclude a confounding effect of genotype and to maximize reliability of data on potential interactions.

Sample size

Sample size was simulated to differentiate accurately subgroups of patients with sharp differences in TPMT activity levels. We based our estimate on mean TPMT among healthy adult subjects in the same region,33, 34 of 22 U/Hgb g, and we considered as clinically relevant differences of 3 U/Hgb g (being an arbitrary cut-off value for intermediate activity set in that study at about 19 U/Hgb g). Therefore, considering hypothetical standard deviation of 7 U/Hgb g, and accepting alpha error <0.10 and beta error <0.05, the minimal sample size was 118 cases 3 (about 60 cases for each subgroup).


Descriptive analysis: population, TPMT genetics and activity

Detailed characteristics of the IBD population are reported in Table 1. Overall, 183 IBD patients were enrolled in the study, 88 (48%) affected by Crohn’s disease, 91 (50%) by ulcerative colitis and 4 (2%) by indeterminate colitis; control subjects enrolled were 533.

Table 1.   Descriptive report of characteristics of patient population overall and according to diagnostic category
Clinical characteristicCrohn’sUlcerative colitisIndeterminateTotal
  1. n, number; NA, not applicable; L1/L2/L3/L4, disease location according to Montreal Classification21; E1/E2/E3, extension of disease according to Montreal Classification21; IQR, interquartile range; EIMs, extraintestinal manifestations; BMI, body mass index, kg/m2.

  2. * Activity index: CDAI is considered for Crohn’s disease patients, and CAI for ulcerative colitis/indeterminate colitis patients; 5-ASA, 5-aminosalycilic acid.

Diagnosis; n (%)88 (48)91 (50)4 (2)183
Gender: Female; n (%)49 (56)38 (42)2 (50)89 (49)
Location L1/L2/L3/L4; n (%)16/42/23/7 (18/48/26/8)NANANA
Extension E1/E2/E3; n (%)NA32/18/41 (35/20/45)2/2/0 (50/50/0)34/20/41 (36/21/43)
Age at diagnosis: years median (IQR)31 (24–45)38 (27–48)46 (39–60)35 (25–46)
Disease duration; years median (IQR)9.4 (5.1–17.5)11.4 (6.4–15.4)6.3 (4.7–8.9)10 (5.8–16.4)
Concomitant EIMs; n (%)32 (36)24 (26)1 (25)57 (31)
Perianal disease; n (%)24 (27)1 (1)0 (0)25 (14)
BMI; median (IQR)22.3 (20.5–25)24.2 (21.3–27.4)24.2 (21.6–26.9)23.7 (21–25.9)
Previous surgery; n (%)44 (50)6 (7)0 (0)50 (27)
Infliximab ever; n (%)22 (25)15 (16)1 (25)38 (21)
Thiopurine ever; n (%)40 (45)54 (59)1 (25)95 (52)
Active thiopurine; n (%)23 (26)44 (48)1 (25)68 (37)
Active 5-ASA; n (%)60 (67)69 (76)4 (100)133 (72)
Concomitant 5-ASA & thiopurines; n (%)11 (13)29 (32)1 (25)41 (22)
Activity index*; median (IQR)161 (79–250)0 (0–3)1 (0–2)NA
Active disease; n (%)38 (43)27 (30)1 (25)66 (36)

Regarding TPMT genetics, the occurrence of mutated haplotypes was observed in 34 controls (6.4%, only heterozygous haplotypes) and 9 cases (4.9%, only heterozygous haplotypes) respectively (= 0.591), leading to allelic frequency of 3.2% and 2.5% respectively; no homozygous mutant subject was observed.

No statistically significant difference in TPMT genetics was noted among IBD patients on the basis of different diagnoses or clinical characteristics.

Thiopurine S-methyl transferase activity was significantly lower in controls (21.42 U/Hgb g, IQR 18.19–25.91) than in cases (24.42 U/Hgb g, IQR 20.34–28.46; P < 0.0001). When only subjects displaying wild-type TPMT genotype were considered, the difference in TPMT activity was still statistically significant: 21.88 U/Hgb g, IQR 18.67–26.20 among 499 controls and 24.90 U/Hgb g, IQR 20.69–29.45 among 174 cases respectively (P < 0.0001). If only cases with no active thiopurines treatment were considered, difference compared to controls was still significant, although to a lesser extent (median TPMT activity 23.19, IQR 19.81–27.29, P = 0.028).

Regarding subjects excluded by main analysis because they were heterozygous for TPMT gene mutations, median TPMT activity was overall 14.66 U/Hgb g (IQR 12.81–17.74), with no significant difference comparing 9 cases and 34 controls (median activity being 13.35 U/Hgb g and 14.68 U/Hgb g, P = 0.095). As expected, median TPMT activity was significantly lower both in heterozygous cases and in heterozygous controls compared with remaining cases and controls (P < 0.0001 for both comparisons). As for thiopurine metabolites, only 3/9 TPMT heterozygous cases were on active thiopurine treatment at the time of analysis: median 6-TGN levels were 250.94 pmol/8 × 108 RBC (range 55.02–260.82) and median 6-MMP levels were 48.43 pmol/8 × 108 RBC (range 0–744.4), differences were not statistically significant (P = 0.3638 and P = 0.1315, respectively) when compared with remaining thiopurine-treated cases.

Univariate analysis: 5-ASA-TPMT interactions

When univariate analysis for TPMT activity and concomitant medications was carried out, only IBD cases were considered and those heterozygous for TPMT genotype were excluded.

On the basis of 5-ASA exposure, we identified 127 cases concomitantly on oral 5-ASA and 47 off oral 5-ASA respectively. TPMT activity was not significantly different in the two subgroups: median activity was 24.34 (IQR 20.39–28.36) and 25.4 (IQR 21.41–30.75, P = 0.215) U/Hgb g respectively. There were no significant differences between patients affected by Crohn’s disease or ulcerative/indeterminate colitis (P = 0.331 and 0.632, respectively). No 5-ASA dose-TPMT activity relationship was noted when comparing TPMT activity of patients not exposed to 5-ASA with those exposed to <2 g/day, 2–3 g/day or >3 g/day of 5-ASA (P = 0.348).

Univariate analysis: azathioprine-TPMT interactions

According to azathioprine exposure, on the other hand, we identified 65 patients on active thiopurine treatment and 109 nontreated patients (26 of whom had been treated in the past). Median TPMT activity was significantly higher in cases on active treatment (26.75 U/Hgb g; IQR 22.25–31.98) compared with patients not on azathioprine (23.19 U/Hgb g; IQR 19.81–27.29; P = 0.0022). There was no difference comparing Crohn’s patients on or off azathioprine (P = 0.433), while the difference was statistically significant in UC cases (P = 0.0002). No significant difference was noted comparing TPMT activity in subgroups of patients exposed to increasing doses of azathioprine (<1.5, 1.5 to 2 or >2 mg/kg b.w./day; P = 0.966). There was no difference in TPMT activity comparing the 33 patients with azathioprine-related adverse events and those without adverse events (P = 0.674). No significant difference was found in the four cases with leukopaenia (P = 0.850).

Among patients on active azathioprine treatment, 41 (63%) were on concomitant treatment with oral 5-ASA, while the remaining 24 were not exposed to combined treatment with 5-ASA and azathioprine. Also, in this subgroup of patients, no significant difference was observed comparing Crohn’s (P = 0.056) and ulcerative colitis patients (P = 0.771) concomitantly treated or not with 5-ASA. Graphic representation of TPMT levels is given in Figure 1 (a–i) and detailed results of univariate analysis for TPMT activity are reported in Table 2. Median azathioprine daily dose was overall 1.76 mg/kg of body weight (b.w.); median dose in patients exposed or not to 5-ASA was 1.72 (IQR 1.43–1.95) and 1.96 (IQR 1.60–2.07) mg/kg b.w. respectively; this difference was not statistically significant (P = 0.081).

Figure 1.

 (a–i) Box and whisker graphic representation of TPMT levels in different subgroups. (a) Crohn’s disease compared with ulcerative or indeterminate colitis (P = 0.255). (b) male compared to female gender (P = 0.521). (c) body mass index (BMI, kg/m2) classes (P = 0.727). (d) disease activity, remission compared with active disease (P = 0.358). (e) C-reactive protein (CRP, mg/l) levels (P = 0.911). (f) 5-ASA exposure (g/day, P = 0.348). (g) azathioprine exposure (mg/kg b.w./day, P = 0.022; for comparisons excluding no azathioprine treatment, P = 0.908). (h) disease duration (P = 0.026). (i) duration of azathioprine treatment in months (P = 0.016; for comparisons excluding no azathioprine treatment, P = 0.853).

Table 2.   Median TPMT activity levels (expressed in U/Hgb g) in different subgroups of patients; for simplification of results and to avoid biases in interpretation of results, only TPMT wild type homozygous are considered. Subgroups are presented as comparator followed by compared subgroup (in brackets)
Subgroups compared NMedian TPMTIQR P
  1. N, number of cases in the subgroup; TPMT, thiopurine S-methyl transferase activity, U/Hgb g; IQR, interquartile range; P, P value; vs., versus compared subgroup; NA, not applicable; IC, indeterminate colitis; BMI, body mass index; CRP, C-reactive protein; 5-ASA, 5-amino salicylic acid; AZA+/5-ASA+, azathioprine and 5-ASA active treatment; AZA+/5-ASA−, azathioprine active treatment and no 5-ASA treatment; infliximab+, active infliximab treatment; infliximab−: no active infliximab treatment; steroid+, active steroid treatment; steroid−, no active steroid treatment.

Cases (overall)17424.920.69–29.45NA
Crohn’s (vs. ulcerative colitis/IC)8325.2621.27–30.170.255
Female (vs. male)8623.7720.61–29.640.521
Actual age ≤40 years (vs. >40)5925.1921.01–29.050.522
BMI <20 kg/m2 (vs. 20–27 vs. >27)3124.9121.86–27.860.727
Active disease (vs. remission)6425.3321.21–30.540.358
CRP > 5 mg/l (vs. ≤ 5)3123.5621.20–29.600.911
 <2 g/day1427.3820.34–28.88
 2–3 g/day9323.619.99–27.17
 >3 g/day2025.1922.74–31.86
Azathioprine (active treatment), AZA
 <1.5 mg/kg b.w./day1526.6621.16–31.04
 1.5–2 mg/kg b.w./day3526.8424.22–30.53
 >2 mg/kg b.w./day1527.7420.34–33.63
AZA+/5-ASA+ (vs. AZA+/5-ASA−)4125.6321.43–32.130.221
AZA+/Crohn’s (vs. AZA+/ulcerative colitis/IC)2325.6322.37–30.820.698
Adverse effects from AZA (vs. other)3123.620.68–28.750.674
Leukopaenia due to AZA (vs. other)426.8721.28–28.630.850
Infliximab+ (vs. infliximab−)3526.9422.67–32.290.049
Steroid+ (vs. steroid−)1927.2318.51–32.800.416
Disease duration:
 <5 years3425.7722.17–31.960.026
 5–15 years8724.9121.38–30.31
 >15 years5322.4418.94–26.65
Duration azathioprine treatment:
 <6 months1226.1922.88–30.820.908
 6–36 months3026.2921.87–31.73
 >36 months2326.9421.67–33.56

Moreover, the analysis of thiopurine metabolites was available in patients on active azathioprine treatment. Median 6-TGN levels (evaluated in 63 cases) were 149.53 pmol/8 × 108 RBC (IQR 84.32–220.29) and there was no statistically significant difference according to diagnostic categories (Crohn’s disease vs. ulcerative colitis), to disease activity and to exposure to co-medication with steroids or 5-ASA. Median 6-MMP levels (analysed in the same group of 63 cases) were 577.74 pmol/8 × 108 RBC (IQR 186.62–1,544.46), with lower values in Crohn’s disease compared with ulcerative colitis, P = 0.014, while they did not vary significantly on the basis of disease activity or whether or not there was a concomitant steroid or 5-ASA treatment.

Regarding the co-medication with 5-ASA, no significant difference was observed in terms of 6-TGN or 6-MMP levels comparing patients on 5-ASA with cases not on 5-ASA (P = 0.496 and 0.922 respectively for 6-TGN and 6-MMP); no difference in 6-TGN or 6-MMP was noted in subgroups treated with different doses of 5-ASA (P = 0.197 and 0.594 respectively). Univariate analysis results for thiopurine metabolites are reported in detail in Table 3.

Table 3.   Median levels of 6-tioguanine nucleotides (6-TGN) and 6-methyl mercaptopurine (6-MMP), expressed as pMol/8 × 108 RBC, in different subgroups of patients; for simplification of results and to avoid biases in interpretation of results, only TPMT wild type homozygous are considered. Subgroups are presented as comparator followed by compared subgroup (in brackets)
Subgroups comparedNMedian 6-TGNIQRPMedian 6-MMPIQRP
  1. N, number of cases in the subgroup; IQR, interquartile range; P, P value; NA, not applicable; vs., versus compared subgroup; IC, indeterminate colitis; AZA+/5-ASA+, azathioprine and 5-ASA active treatment; AZA+/5-ASA−, azathioprine active treatment and no 5-ASA treatment; infliximab +, active infliximab treatment; infliximab−, no active infliximab treatment; steroid+, active steroid treatment; steroid−, no active steroid treatment.

Cases (overall)65149.5384.32–220.29NA577.44186.62–1544.46NA
Crohn’s (vs. ulcerative colitis/IC)23143.5872.22–201.030.288368.91100.31–831.860.014
AZA+/5-ASA+ (vs. AZA+/5-ASA−)38157.3286.19–242.930.496615.01148.40–1724.320.922
Active (vs. remission)20171.08111.83–248.600.301556.21205.01–1418.820.965
Infliximab + (vs. infliximab −)25123.9870.42–197.050.177599.61149.94–1270.760.491
Steroid + (vs. steroid −)6159.3734.88–306.050.888823.44577.44–1829.040.413

Median value of the ratio between 6-TGN levels and azathioprine daily dose per kg body weight was 89.64 (IQR 58.80–124.19), differences of this ratio between patients exposed or not to 5-ASA were not statistically significant (P = 0.327, median 90.39 and 88.94 respectively). Median value of the 6-MMP levels/azathioprine daily dose per kg ratio was 455.22 (IQR 116.02–780.11), differences of this ratio between patients exposed or not to 5-ASA were not statistically significant (P = 0.9519, median 482.81 and 345.82 respectively).

Potential interactions between thiopurines and infliximab were reported;34 therefore, a specific analysis was carried out to test potential interactions. In this series, 25/63 (40%) cases were concomitantly treated with infliximab and AZA; however, no significant difference was observed either in TPMT activity (P = 0.612) or in 6-TGN (P = 0.177) or 6-MMP levels (0.491). Considering all the cases, 35 patients being actively on infliximab had slightly but significantly higher median TPMT activity (26.94 U/Hgb g, IQR 22.67–32.29) compared with the remaining cases not exposed to infliximab (24.34, IQR 20.39–28.28; P = 0.049).

Although it was not part of the goals of this study, as it has been reported that 6-TGN values greater than 230–260 pmol/8 × 108 RBC are associated with better chances to maintain clinical remission,35 we analysed 6-TGN levels and no significant correlation was observed with disease remission or for the mentioned cut-off values (P = 0.684 and 0.549 respectively) or for the ROC (receiver operator characteristics) analysis-generated cut-off (calculated on the population of the present study) of 150 pmol/8 × 108 RBC (P = 0.422).

Multivariate analysis

When exploring independent determinants of TPMT activity and metabolite levels by means of multiple stepwise regression, TPMT activity was significantly and independently associated with increasing doses of azathioprine/kg of body weight (P = 0.016), but not with any other clinical characteristics considered, including use and dose of mesalazine (mesalamine). Regarding 6-TGN levels, no clinical characteristic was significantly and independently associated with 6-TGN levels, while for 6-MMP levels, only the diagnosis of Crohn’s disease as compared to UC/IC diagnosis was significantly and independently associated with 6-MMP levels (P = 0.023). No other clinical or therapeutic variable considered was retained in the multiple regression model.


Patients affected by IBD are likely to undergo combination treatments with several drugs, with possible harmful interactions36 and at present, possible or definite relevant interactions were reported for antibiotics, steroids, metothrexate, 5-ASA/sulphasalazine36 and biologics.34

Thiopurines (azathioprine, AZA, and mercaptopurine, MP) and 5-ASA, among them, are the most widely used drugs for the treatment of inflammatory bowel disease.

Although thiopurines are being used since 1970s,37–40 their use in clinical practice has increased substantially over the last few decades.1, 2 Moreover, co-prescription of 5-ASA in patients on thiopurines is not uncommon (probably also because of the potential chemopreventive effect of 5-ASA12) and in the reported series, up to 60% of patients on azathioprine were also treated with 5-ASA.

Thiopurines are quite well tolerated; however, severe leucopenia, hepatitis, pancreatitis and other kinds of intolerance may occur in up to 15–20% of cases, as reported in different series.41–43

Thiopurine S-methyl transferase null activity is a relevant cause of severe, potentially lethal, leucopenia and it was shown that patients with homozygous mutated TPMT gene (and also with heterozygous or wild type haplotype) are at high risk of severe and early leucopenia.9, 10, 44, 45

Since then, several studies have been published on the importance of TPMT genotyping and phenotyping in inflammatory bowel disease, demonstrating that careful genetic analysis may help prevent severe leukopenia. As the expected frequency of TPMT homozygous mutated haplotype is expected to be as high as 1 in 300 subjects in Caucasians, it has been proposed that systematic TPMT genotyping (or enzymology) prior to initiating treatment is advisable and cost-effective.46–48 The Food and Drug Administration (FDA) changed the azathioprine package insert in 2004, which now states: ‘It is recommended that consideration be given to either genotype or phenotype patients for TPMT prior to the initiation of therapy....’; the FDA, however, deferred from making testing compulsory.49 In the ECCO therapeutic guidelines, a note is given on the potential usefulness of pharmacogenetic testing for TPMT or other genes, but no recommendation is given.5, 6

Fairly recently, moreover, some pre-clinical data suggested a remarkable interaction between mesalazine or salazopyrine and thiopurine metabolism.18 Different authors suggested either not to co-treat patients with the two drugs because of harmful interactions14, 15, 50, 51 or to exploit the interaction to increase the therapeutic efficacy of thiopurines.16, 19 However, a possible significant interaction between thiopurines and mesalazine would have a great impact on the management of IBD in terms of dosage of thiopurines, efficacy and adverse events.

The largest cohort study published to date on IBD patients50 analysed the effect of co-medications (5-ASA and thiopurines) on adverse effects, with testing only for TPMT activity. In that study, significantly more adverse events were reported in patients undergoing dual therapy compared with patients on ‘simple’ thiopurine treatment. However, only the association with adverse effects was noted and no data on metabolite levels were available. The aim of our study was to observe differences in a surrogate end-point, such as TPMT levels compared with the overt adverse outcomes of thiopurine treatment (toxicity, mortality, dose-adjustments because of minor toxicity …). Larger interventional prospective studies aimed at addressing this point specifically may further clarify the issue.

Our group recently performed a study exploring TPMT genotype and phenotype in healthy populations, comparing adult and paediatric population.33 Our data suggest that in the studied North-Western Italian population, the occurrence of mutated TPMT haplotypes is lower than expected and that previously reported in Caucasians (about 1:1100 subject carries homozygous mutations or combined haplotypes), thus reducing the cost-effectiveness of a possible screening base purely on genetic analysis of TPMT. This observation warrants further confirmation in larger studies in different populations. Possibly, the interaction between salycilates and TMPT may have a greater impact and therefore a greater clinical significance in patients with lower TPMT activity, but this point needs further confirmation in prospective interventional studies.

Moreover, subjects with no major mutation in the TPMT gene may show significantly decreased TPMT activity. Therefore, it is probably safer to offer a combined genetic-phenotypic screening.

The major aim of the present study was to analyse how active treatments act on TPMT activity and if any of the most commonly used drugs displays significant effects (inhibitory or stimulating) on TPMT, which may result in useful/harmful effects during thiopurine treatment.

First, we recorded a slight but significant difference in TPMT activity comparing IBD cases with controls (which were enrolled mainly for the genetic substudy). However, when discussing the clinical significance of lower TPMT activity observed in controls compared with IBD cases, it must be kept in mind that on the basis of controls data, the cutoff value to identify the 10% low activity tail of the population distribution is TPMT activity <15 U/Hgb g, and the corresponding high activity tail is identified by TPMT activity values >32 U/Hgb g; therefore, although statistically significant, there is no clinical meaning for higher TPMT activity of IBD patients: it still runs well within the normality range. Moreover, part of the difference disappears when patients on active thiopurines are excluded, as a logical consequence of an inductive effect of thiopurines on TPMT activity.

When we analysed the effect of 5-ASA on TPMT activity, no significant difference, especially no dose-effect, was noted. This finding is somewhat different from other small prospective interventional studies21, 22 in which adding 5-ASA in patients on stable thiopurine doses led to a significant increase in 6TG levels; it may be argued that larger confirmatory studies are needed and that the present study possibly enrolled patients tolerating thiopurine-5-ASA coadministration, as patients with intolerance/adverse effects were no longer on concomitant treatment, or eventually that after an early interference, there is adaptation that leads to minor interference between the two drugs. This second hypothesis may be supported by other studies,52 which showed no major interaction between 5-ASA and thiopurines in the long-term at 1 year, but needs further evaluation.

Our study population included both patients on and off thiopurines and both these subgroups included patients on or off 5-ASA, but neither in the former nor in the latter cases we recorded any significant difference in terms of TPMT activity.

Although sub-analyses for steroid and infliximab were carried out on smaller subgroups, they equally did not highlight any major clinically relevant interactions between these drugs and TPMT activity or thiopurine metabolite levels.

One major criticism may be that the study population was relatively small, as the subgroup analysis led to comparison of fewer patients than those included in the original complete cohort. However, it must be kept in mind that although analyses on thiopurine metabolites are on limited number of patients, those on TPMT activity compare groups of greater than 50 cases, and this may lead to detect accurately, differences of 3–4 U/Hgb g, which may be considered to be of very limited clinical significance, if any.

In conclusion, our data support the absence of major interactions between thiopurines and other concomitant medications and especially with 5-ASA. Although these results require further prospective and independent validation, it may be inferred that at present, 5-ASA co-prescription in patients on thiopurines is safe enough to be continued in everyday clinical practice. The advantages are probably limited to colonic cancer chemoprophylaxis, as no clear major additive effects on clinical effectiveness were highlighted in the Literature, but they are anyway relevant to IBD patient management.


Declaration of personal interests: None. The authors are indebted to all the patients for their participation in the study and to the regional patients' association (AMICI Piemonte) for dissemination of news on the research programme. Declaration of funding interests: This study was funded by Fondazione IBD Onlus, with a grant by Fondazione Compagnia di San Paolo (grant number 2002.2003).