Adverse events leading to modification of therapy in a large cohort of patients with inflammatory bowel disease


Dr U. Hindorf, Department of Clinical Sciences, Division of Gastroenterology, Faculty of Medicine, Lund University, SE-22185 Lund, Sweden.




Adverse events leading to discontinuation or dose reduction of thiopurine therapy occur in 9–28% of patients with inflammatory bowel disease.


To evaluate the influence of thiopurine methyltransferase status and thiopurine metabolites in a large patient population for the risk of developing adverse event.


Three hundred and sixty-four patients with inflammatory bowel disease and present or previous thiopurine therapy were identified from a local database.


The adverse event observed in 124 patients (34%) were more common in adults than children (40% vs. 15%; P < 0.001) and in low to intermediate (≤9.0 U/mL red blood cell) than normal thiopurine methyltransferase activity (P = 0.02). Myelotoxicity developed later than other types of adverse event. An increased frequency of adverse event was observed in patients with tioguanine (thioguanine) nucleotide above 400 or methylated thioinosine monophosphate above 11 450 pmol/8 × 108 red blood cell. A shift to mercaptopurine was successful in 48% of azathioprine-intolerant patients and in all cases of azathioprine-induced myalgia or arthralgia.


A pre-treatment determination of thiopurine methyltransferase status might be appropriate as patients with low to intermediate thiopurine methyltransferase activity are more prone to develop an adverse event; determination of metabolite levels can be useful in the case of an adverse event. Mercaptopurine therapy should be considered in azathioprine-intolerant patients.


The thiopurine drugs azathioprine (AZA) and mercaptopurine (MP) are well established in the treatment of inflammatory bowel disease (IBD) and have proven to be effective in both inducing and maintaining remission of Crohn's disease (CD) and ulcerative colitis (UC).1–4 Adverse reactions to AZA or MP occur in 9–28% of patients and often necessitate dose reduction or discontinuation of the administered drug.5–7 Individual variations in drug metabolism is of importance for differences in tolerance to thiopurines, which undergo extensive metabolic transformation resulting in several active and inactive metabolites (Figure 1). The tioguanine nucleotide (TGN) metabolites act as purine antagonists and induce cytotoxicity and immunosuppression by inhibition of RNA, DNA and protein synthesis. These cytotoxic properties are, at least in part, due to the direct incorporation of TGN into DNA.8 Recently, it has been suggested that the immunosuppressive effects of thiopurines is due to competitive binding of TGN triphosphate to the normally guanine triphosphate-binding protein Rac1. This results in suppression of Rac1 activation and induction of apoptosis. This mechanism is thought to control elimination of activated T lymphocytes.9

Figure 1.

 The thiopurine metabolism. MP, mercaptopurine; AZA, azathioprine; XO, xanthine oxidase; TPMT, thiopurine methyltransferase; HPRT, hypoxanthine guanine phosphoribosyl transferase; GMPS, guanosine monophosphate synthetase; IMPDH, inosine monophosphate dehydrogenase; TUA, thiouric acid; MMP, methylmercaptopurine; TIMP, thioinosine monophosphate; meTIMP, methyl-TIMP; TXMP, thioxanthine monophosphate; TGMP, tioguanine monophosphate; TGDP, tioguanine diphosphate; TGTP, tioguanine triphosphate; dTGTP, deoxy-TGTP; PRPP, phosphoribosyl pyrophosphate; PRA, phosphoribosylamine; IMP, inosine monophosphate. The metabolites TGMP, TGDP, dTGDP, TGTP and dTGTP are collectively named tioguanine nucleotides or TGN and are marked with a dark grey triangle.

The methylation of MP and thioinosine monophosphate (TIMP) is catalysed by thiopurine methyltransferase (TPMT), which leads to the production of methylthioinosine monophosphate (meTIMP). This metabolite is found in concentrations that far exceed the TGN concentrations10, 11 and is a potent inhibitor of purine de novo synthesis (PDNS) in vitro.12, 13 High meTIMP concentrations have, furthermore, been proposed to increase the risk of hepatotoxicity.14–16

Genetic polymorphisms in the TPMT gene (TPMT *2 to *18) are associated with decreased TPMT activity.17, 18 Numerous studies have shown that patients with very low enzyme activity are at risk for severe, and sometimes fatal, myelotoxicity,19–21 and patients who are TPMT heterozygous for one allele with low activity have an intermediate risk for myelotoxicity22, 23 because of high TGN metabolite concentrations.

Whether other types of adverse event (AE) in patients with IBD can be attributed to the TPMT polymorphism have been addressed only in a few studies. In two studies, IBD patients with intermediate TPMT activity were shown to have an increased risk of AZA toxicity overall,24, 25 while these results could not be confirmed in another study.5 AZA-related gastrointestinal AE has been shown to be independent of the TPMT polymorphism.26

The aim of this retrospective study was to investigate whether AZA-related AE can be explained by variations in TPMT enzyme activity and the thiopurine metabolites TGN and meTIMP.


Patients with IBD in whom metabolite and/or TPMT measurements had been performed were identified from a database that included all measurements during a 7-year period (1997–2003). For this period the database contains overall 2065 metabolite and/or TPMT samples. Medical records from all sampled patients at four hospitals (Departments of Gastroenterology at University Hospital, Linköping; University Hospital, Lund; Blekinge Hospital, Karlskrona and Department of Pediatric Gastroenterology, Astrid Lindgren Children's Hospital, Stockholm) were reviewed by three experienced gastroenterologists (UH, SA, HH). Patients with present or previous treatment with thiopurine drugs were included, while patients who had not yet started therapy at the time point for laboratory analysis were not included. In patients without AE and multiple TPMT and/or metabolite measurements the last test was used in the statistical analysis of data. In patients with AE we only used metabolite measurements that were obtained at the time point for AE. In the group with AE we included patients in whom the presence of an AE led to discontinuation or dose reduction of AZA treatment. The AE were grouped into five main types of toxicity: (i) myelotoxicity which included all types of haematological toxicity (anaemia, leucopoenia, neutropenia, thrombocytopenia), (ii) gastrointestinal intolerance which included abdominal pain, nausea, vomiting and diarrhoea, (iii) hepatotoxicity which included all types of hepatotoxic reactions, (iv) allergic/systemic reactions which included arthralgia, myalgia, fever, rash, general malaise and other non-specific reactions and (v) pancreatitis.

Anaemia was defined as a haemoglobin level <120 g/L, leucopoenia as a white blood cell count (WBC) <3.0 × 109/L, neutropenia as an absolute neutrofile count (ANC) <1.5 × 109/L, thrombocytopenia as a platelet count <100 × 109/L, hepatotoxicity as an aspartate aminotransferase (AST) or alanine aminitransferase (ALT) more than five times the upper normal limit or an alkaline phosphatase (ALP) more than three times the upper normal limit, and pancreatitis as severe abdominal pain accompanied with a rise in serum amylase levels.

Thiopurine methyltransferase activity was determined as previously described.27 Briefly, we measured the formation of 6-methylmercaptopurine from MP (6-mercaptopurine) with radiolabelled S-adenosyl-l-methionine used as the methyl donor. Product formation was measured by a liquid scintillation counter. One unit of enzyme activity represents the formation of 11 nmol of 6-methylmercaptopurine per millilitre of red blood cells (RBC) per hour of incubation. The interassay and intra-assay coefficients of variation are 5% and 3% respectively.

Tioguanine nucleotide and meTIMP were determined as described previously.27 Blood collected in ethylenediaminetetraacetic acid (EDTA) tubes were centrifuged and RBCs were washed and diluted in saline to a final concentration of 8 × 108 cells per 200 μL prior to storage at −70 °C. TGN and meTIMP were then determined by reverse-phase high-performance liquid chromatography (HPLC) at 330 nm as purine bases after acid hydrolysis and an extraction procedure. The limit of quantification for TGN was 20 pmol/8 × 108 RBC and for meTIMP 300 pmol/8 × 108 RBC. At these levels, the interassay coefficients of variation are 12% and 17% respectively.

Normal TPMT activity was defined as equal or above 9.0 U/mL RBC, intermediate TPMT activity as 2.5–8.9 U/mL RBC and low TPMT activity below 2.5 U/mL RBC. Using these cut-off levels for TPMT activity, we have previously found a close relationship between TPMT genotype and phenotype.28 TPMT genotype was determined by a pyrosequencing method as described previously.18 The patients were genotyped for the following nucleotide substitutions: 238G>C, 460G>A, 719A>G, 292G>T, intron IX/exon X splice site (G>A), 146T>C, 539A>T, 681T>G, 644G>A and 430G>C (TPMT *2, *3A, *3B, *3C, *3D, *4, *5, *6, *7, *8 and *10). Genotyping for +1A>G (TPMT *14) and for IVS7-1G>A (TPMT *15) was performed as described previously.18 However, for TPMT *15, polymerase chain reaction (PCR) primers for exon VIII was used.28

Statistical evaluation

Statistical analysis was performed with the Statistical Package for Social Sciences 11.5 for Windows (SPSS, Inc., Chicago, IL, USA). In calculations including drug doses, we converted the MP dose into an equivalent AZA dose with a conversion factor of 2.08.29 Mann–Whitney U-test was used to evaluate differences between independent groups and Spearman rank order correlation coefficient was applied to test for correlation between parameters. Differences between multiple independent groups were evaluated using the Kruskal–Wallis test. Results are expressed as medians with interquartile (q1–q3) ranges throughout.

Ethical considerations

The protocol was approved by the respective local ethics committees.


A total of 364 patients with ongoing or prior thiopurine therapy were identified (Table 1). Adverse events that led to discontinuation of therapy or dose reduction were evident in 124 (34%) subjects. The remaining 240 patients had been treated with thiopurines for a median of 1.5 (0.5–3.0) years without experiencing any AE. The emergence of AE led to discontinuation of thiopurine therapy in 114 patients (31%) and to dose reduction with a median of 50 (50–81) mg AZA in 10 patients. After discontinuation the same thiopurine drug (AZA) was successfully reintroduced in seven cases, while another thiopurine drug were introduced in 54 patients. Fifty-two patients with AE on AZA where changed to MP and this shift was successful in 25 patients, while one was shifted from MP to AZA and one from AZA to 6-tioguanine (6-thioguanine). A re-challenge or shift was for various reasons not performed in 53 patients. Overall, modification of thiopurine therapy was successful in 42 of 71 patients (59%) and resulted in significantly lower thiopurine doses compared with AE [1.1 (0.8–1.5) vs. 1.8 (1.2–2.5) mg AZA/kg body weight (BW); P < 0.001]. Details of AE and the medical decisions are presented in Table 2.

Table 1.   Patient characteristics
 Total (n = 364)AE (n = 124)No AE (n = 240)P-value
  1. * Information regarding 5-ASA use was not available for six patients.

  2. AE, adverse event; RBC, red blood cell; CD, Crohn's disease; UC, ulcerative colitis; TPMT, thiopurine methyltransferase.

Disease (CD/UC)249/11577/47172/680.06
Gender (male/female)192/17258/66134/1060.10
Age (years)†28 (18–42)36 (24–51)24 (17–37)<0.001
Surgery (yes/no)94/27038/8656/1840.13
Prior anti-TNFα therapy (yes/no)25/3397/11718/2220.51
Corticosteroids (yes/no)156/20875/4981/159<0.001
5-ASA (yes/no)207/151*59/65148/860.02
TPMT (U/mL RBC)12.7 (10.9–14.5)12.5 (10.2–14.4)12.8 (11.2–14.5)0.14
Disease extension
 CD – terminal ileitis18810 
 CD – colitis1102783 
 CD – ileocolitis712645 
 CD – extensive disease491732 
 CD – total24977172 
 UC – pancolitis722745 
 UC – extensive colitis1459 
 UC – distal colitis281414 
 UC – proctitis211 
 UC – total1154768 
Indication for thiopurine therapy
 Frequent relapses351025 
 Steroid dependence1506882 
 Chronic active disease942767 
 Fistulizing disease35728 
 Remission maintenance31625 
 Postoperative prophylaxis651 
 Severe initial disease1028 
Table 2.   Details of AEs and medical decision in different types of AE
Medical decisionAdverse eventTotal
  1. * The myelotoxicity group consist of pancytopenia (n = 7), leucopoenia and neutropenia (n = 20), and isolated thrombocytopenia (n = 1).

  2. AE, adverse event.

Dose reduction8110010
Discontinuation and reintroduction330107
Change to other immunosuppression520719354

Patients who experienced AE were treated with lower thiopurine doses at the time point for AE compared to those without AE [1.5 (1.0–2.2) vs. 1.9 (1.5–2.3) mg AZA/kg BW; P < 0.001].

Adverse events were more common in adults (>18 years) and occurred in 113 of 285 cases (40%) compared with 12 of 79 cases (15%) in children (≤18 years; P < 0.001), but there were no difference in either thiopurine dose [1.8 (1.4–2.3) vs. 1.8 (1.4–2.1) mg AZA/kg BW; P = 0.46] or TPMT activity [12.7 (10.9–14.2) vs. 12.8 (10.8–14.8) U/mL RBC; P = 0.83]. Among children the AE consisted of myelotoxicity (n = 3), gastrointestinal reactions (n = 3), hepatotoxicity (n = 1), allergic/systemic reactions (n = 3) and pancreatitis (n = 2).

There were no significant differences in either gender or disease type between patients with and without AE. Patients with concomitant steroid use were more likely to experience AE and patients with concomitant 5-ASA use were less likely to experience AE (for details see Table 1).

AZA-intolerant patients shifted to MP

Fifty-two patients initially treated with AZA were changed to MP because of intolerance, which was successful in 25 of 52 cases (48%). Twenty-seven patients developed AE on MP treatment after 28 (8–74) days, while the 25 patients who tolerated MP had been treated for 792 (390–1130) days when their medical records were reviewed. There were no significant differences between the group which continued and the group which discontinued MP treatment in either time on AZA treatment [58 (32–227) vs. 39 (24–251) days; P = 0.29], AZA dose [1.5 (1.2–1.9) vs. 1.3 (0.8–1.8) mg/kg BW; P = 0.21], MP dose [0.9 (0.7–1.2) vs. 0.9 (0.7–1.2) mg/kg BW; P = 0.63], age [30 (20–50) vs. 41 (25–57) years; P = 0.20], disease type (CD/UC: 15/10 vs. 19/8; P = 0.44), or TPMT activity [12.1 (8.5–15.3) vs. 12.4 (10.4–15.1); P = 0.83]. All seven patients who reacted with myalgia or arthralgia on AZA tolerated a shift to MP, while only two of nine (22%) with nausea and four of 12 (33%) with abdominal pain did.

Time to adverse event

Eighty-seven of the 124 patients (71%) experienced their AE during the initial 3 months of therapy. Myelotoxicity developed after a median of 104 (54–490) days, which was significantly later than other types of AE; hepatotoxicity [45 (21–70) days; P = 0.001], gastrointestinal intolerance [32 (19–76) days; P < 0.001], allergic/systemic reactions [40 (20–188) days; P =0.004] and pancreatitis [21 (17–28) days; P < 0.001]. When we compared the time to AE between the other types of toxicity the only difference was that pancreatitis developed significantly earlier than hepatotoxicity (P = 0.02; Figure 2).

Figure 2.

 The majority of adverse event (AE; 71%) occurred during the initial 3 months of thiopurine therapy. AE that occurred later consisted mainly of myelotoxicity (15 of 35 cases). AE that occurred after more than 1 year of thiopurine therapy consisted in seven of 11 cases of myelotoxicity.

TPMT status and toxicity

Low TPMT activity was observed in six patients, while 45 had an intermediate TPMT activity and the remaining 313 had a normal TPMT activity. In 52 patients [TPMT deficiency (n = 6); heterozygote TPMT genotype (n = 6); wild-type TPMT genotype (n = 40)] both TPMT genotype and TPMT phenotype were determined; TPMT *3A/*3A (n = 1); TPMT *3A/*3C (n = 3); TPMT *3A/*14 (n = 1); TPMT *3A/*15 (n = 1); TPMT *1/*3A (n = 6); TPMT *1/*1 (n = 40). There was a significant correlation between genotype and phenotype (rs = 0.73; P < 0.001) with minimal overlap (one patient with a wild-type genotype and a TPMT activity of 8.9 U/mL RBC and one patient with a heterozygote genotype and a TPMT activity of 9.0 U/mL RBC). Adverse events occurred in 83% of patients with low TPMT activity and in 42% of patients with intermediate TPMT activity compared to 32% of patients with normal TPMT activity (P = 0.02). Adverse events were significantly more common among patients with low to intermediate TPMT activity (≤9.0 U/mL RBC) compared to those with normal TPMT activity (P = 0.03). However, there was no difference in TPMT activity between patients with myelotoxicity, gastrointestinal intolerance, hepatotoxicity, allergic/systemic AE, pancreatitis and patients without AE (P = 0.39; Figure 3 and Table 3).

Figure 3.

 Thiopurine methyltransferase (TPMT) activity in patients with different types of adverse event (AE) and patients without AE. There was no significant difference in TPMT activity between the groups (P = 0.39; Kruskal–Wallis test). Each box shows the median and the interquartile range values; horizontal lines show the total range within a category. bsl00043 represents outliers and * represents extremes.

Table 3.   AZA dose and TPMT status in different types of AE
 Adverse eventP-value
  1. * The first episode of toxicity is noted and one patient in whom pancytopenia occurred on MP was initially treated with AZA for 30 days and developed abdominal pain which made the patient stop the medication. At the time point when she was treated with AZA her TPMT deficiency was not known.

  2. AE, adverse event; RBC, red blood cell; AZA, azathioprine; TPMT, thiopurine methyltransferase.

AZA dose (mg)150 (100–175)100 (50–150)100 (52–100)100 (62–114)100 (75–125)0.01
AZA dose (mg/kg)2.3 (1.5–2.5)1.5 (0.8–2.2)1.4 (1.0–1.7)1.4 (0.9–1.8)1.6 (1.1–1.7)0.02
TPMT activity (U/mL RBC)12.7 (10.4–14.3)11.5 (9.3–15.2)13.1 (11.7–14.3)11.9 (9.8–13.6)12.7 (11.5–15.2)0.60
TPMT phenotype (normal/ intermediate/low)22/3/322/5/1*15/4/028/5/113/2/0
Number of cases2828193415Total: 124

The time to AE was 44 (21–118) days in the normal TPMT activity group, 77 (23–214) days in the intermediate TPMT activity group and 53 (40–77) days in the low TPMT activity group (P = 0.45). In the normal TPMT group, patients who tolerated AZA were treated with significantly higher doses compared with patients who developed AE [1.9 (1.6–2.3) vs. 1.5 (1.0–2.2); P < 0.001], while there were no differences in the thiopurine dose between patients with and without AE in the intermediate [1.3 (1.0–1.9) vs. 1.7 (1.3–2.3); P =0.28], or low [1.0 (0.5–2.3) vs. 0.27 (0.27–0.27); P =0.33] TPMT groups. However, among patients who tolerated AZA therapy (n = 240) there were significant differences in thiopurine dose over the three TPMT activity groups (Figure 4).

Figure 4.

 Azathioprine (AZA) dose in relation to adverse events (AE) in the three different thiopurine methyltransferase (TPMT) groups. There were significant differences in both absolute AZA dosage [normal: 125 (100–150); intermediate: 75 (50–109); low: 25 (25–25); P < 0.001; a] and AZA dosage per kg body weight [BW; normal: 1.9 (1.6–2.3); intermediate: 1.3 (1.0–1.9); low: 0.27 (0.27–0.27) mg/kg BW; P < 0.001; b] in patients without AE. In patients with AEs there were no significant differences in dosage between the three TPMT activity groups. Each box shows the median and the interquartile range values; horizontal lines show the total range within a category.

Thiopurine metabolites and toxicity

Thiopurine metabolites were measured in 50 patients with AE at the time point for toxicity as well as in 216 patients who did not experience any toxicity. When we compared these two groups there were no significant differences in either TGN [204 (118–399) vs. 170 (110–279) pmol/8 × 108 RBC; P = 0.14] or meTIMP [1750 (665–4050) vs. 1500 (800–3000) pmol/8 × 108 RBC; P = 0.42] levels. However, when we considered the different types of toxicity, we noticed that in the myelotoxicity group (n = 20) the meTIMP [2650 (1100–10 800) vs. 1500 (800–3000) pmol/8 × 108 RBC; P = 0.03] and in the gastrointestinal intolerance group (n = 9) the TGN [332 (189–459) vs. 170 (110–279) pmol/8 × 108 RBC; P = 0.02] metabolites were significantly higher compared to the group without AE (n = 216). When we compared the hepatotoxicity (n = 6), allergic/systemic (n = 9) and pancreatitis (n = 6) groups with patients but without AE we found no significant differences.

Twenty-nine patients (11%) had TGN levels above 400 pmol/8 × 108 RBC and 41% of these experienced AE, which was significantly higher than among patients with TGN concentrations below this level. In a similar fashion, the frequency of AE was significantly higher among the 12 patients (5%) with meTIMP levels above 11 450 pmol/8 × 108 RBC (for details, see Figure 5).

Figure 5.

 (a and b) Tioguanine nucleotide (TGN) metabolite levels divided into three strata; 0–250 pmol/8 × 108 red blood cell (RBC) represent subtherapeutic levels, 250–400 pmol/8 × 108 RBC represent therapeutic levels and >400 pmol/8 × 108 RBC represent potentially toxic levels. Twenty-nine patients (11%) had levels above 400 pmol/8 × 108 RBC and 12 patients (41%) developed adverse event (AE). It was a higher frequency of AE in the group with TGN levels above 400 pmol/8 × 108 RBC compared with the other two groups (a). There are little information about methylthioinosine monophosphate (meTIMP) levels and toxicity. We divided meTIMP levels into normal (<5700 pmol/8 × 108 RBC) and then divided further on the basis of proposed cut-off levels for hepatotoxicity (5700 pmol/8 × 108 RBC)14 and myelotoxicity (11 450 pmol/8 × 108 RBC).25 Only 12 patients have meTIMP levels above 11 450 pmol/8 × 108 RBC, but in this group seven (58%) developed AE (b). The frequency of AE in the different groups have been compared by using the Mann–Whitney U-test.


The thiopurine drugs AZA and MP are effective in both inducing and maintaining remission in IBD, but their use are limited by AEs that may lead to discontinuation of therapy.5–7 Adverse events during thiopurine therapy have been attributed to different mechanisms, but during the last years the potential influences of TPMT and ITPA polymorphisms have gained interest.26, 30

This is the largest study so far evaluating the relationship between thiopurine metabolism and AE in thiopurine-treated patients with IBD. Even though patients sampled probably infer a selection bias, as the frequency of AE leading to discontinuation or dose reduction of thiopurine therapy (34%) is higher than previously reported,5–7 this study mirrors general practice both in adults and children with IBD. Furthermore, all samples were analysed in a single laboratory and the medical records reviewed by a few experienced gastroenterologists.

In the present study, TPMT phenotype was used to evaluate the potential relationship between the TPMT polymorphism and thiopurine-related AEs. Although the majority of TPMT activity measurements were obtained after the start of treatment, we believe that this fact does not influence our results as we recently shown that there is no significant TPMT induction in the vast majority of patients.25 There was also an almost complete concordance between TPMT phenotype and genotype in the subset of 52 patients in whom TPMT genotype were determined. Furthermore, thiopurine metabolites were measured at the time point of toxicity in 50 of 124 cases, allowing an evaluation of pharmacokinetic mechanisms behind different types of thiopurine-related AE.

The AZA doses in patients who developed AE were clearly lower than in patients that tolerated thiopurine therapy. This finding is explained by the fact that many cases of AE occurred early after initiation of therapy and before target doses had been reached, a phenomenon known from previous studies.24 When we looked at the AZA doses in different types of AE, we noticed that the median doses were significantly higher in the group with myelotoxicity. This is most likely explained by the fact that this type of AE occurred later in time compared with other types of AE, making it possible for more patients in this group to reach target doses.

The vast majority of AE (71%) occurred during the initial 3 months of therapy. Thirteen of 15 cases of pancreatitis occurred within the first month of therapy, well in line with earlier observations.31, 32 Other types of gastrointestinal intolerance (abdominal pain and nausea) as well as allergic/systemic AE and hepatotoxicity were most frequent during the initial 3 months of therapy, while myelotoxicity occurred significantly later than the other types of toxicity. A substantial part (25%) of the myelotoxic events developed after more than 1 year of therapy, which is in accordance with a previous study.33 The reasons for this late development is unclear, but might be due to a number of factors including dose escalation, addition of new medications, viral infections, variations in disease activity that could influence the intestinal absorption, as well as a variable patient compliance over time. This observation illustrates the need for a continued monitoring of blood counts throughout the duration of the thiopurine therapy. However, myelotoxicity developed in four TPMT-deficient patients within the initial 2inline image months of therapy and although complete blood counts were checked weekly, the development of myelotoxicity was fast and occurred between tests obtained a week apart. Even though thiopurine therapy was discontinued immediately when bone marrow suppression became evident, all patients developed severe pancytopenia during the next few days. This was probably due to the fact that TGN levels already had reached extremely toxic levels, as illustrated by the patient who was sampled at the initial phase of pancytopenia with a TGN level of 4097 pmol/8 × 108 RBC. In our experience, the only way to avoid this possibly fatal toxicity is by a pre-treatment TPMT testing (for details, see Table 4). One TPMT-deficient patient in this study was treated with a low dose (0.32 mg AZA/kg BW), as the TPMT status was known prior to initiating the treatment. He developed flu-like symptoms with myalgia and fever and treatment had to be stopped after 78 days. The last patient with TPMT deficiency had been treated with a moderate AZA dose (0.71 mg/kg BW) for 7 years without signs of toxicity before TPMT status was checked and a compound heterozygote genotype (TPMT *3A/*3C) with TPMT activity of 0.2 U/mL RBC was observed. Repeated metabolite measurements revealed TGN levels in the range between 748 and 1144 pmol/8 × 108 RBC. Subsequently, the dose was reduced to 0.26 mg/kg BW and TGN levels had dropped to 342 pmol/8 × 108 RBC at the latest measurement while the patient remained in disease remission.34

Table 4.   Details of four patients with TPMT deficiency and pancytopenia
NumberTPMT genotypeTPMT activityAZA dose (mg)AZA dose (mg/kg BW)Time to AE (days)Hb (g/L)WBC (109/L)Trc (109/L)Comments
  1. † Treated with MP in a dose of 75 mg equivalent to 156 mg AZA.

  2. AE, adverse event; AZA, azathioprine; CD, Crohn's disease; UC, ulcerative colitis; TPMT, thiopurine methyltransferase; BW, body weight.

1*3A/*3A0.51001.776601.4133Female, 28 years, UC
2*3A/*3C0.5751.050731.345Female, 22 years, CD
3*3A/*3C1.1500.752940.881Male, 54 years, CD
4*3A/*142.3156†2.927741.017Female, 20 years, UC

When there is an indication for thiopurine treatment in IBD patients, the clinical practice in Sweden has been to initiate treatment with AZA, while MP has mainly been used in AZA-intolerant patients. In this study, 52 patients intolerant to AZA were shifted to MP and 48% of them tolerated MP. This observation is in line with previous observations in smaller groups of patients, where 45–73% of patients intolerant to AZA where able to tolerate MP.35–37

It has been postulated that some early idiosyncratic reactions, such as digestive intolerance, pancreatitis, cholestasis and flue-like symptoms during AZA therapy might be due to the nitro-imidazole compound found in AZA, which is released to produce MP.38 Interestingly, all seven patients that reacted with myalgia/arthralgia on AZA-tolerated MP without any AE. However, not more than 22% of patients with nausea and 33% of patients with abdominal pain were able to tolerate MP. This finding is in contrast to the results of Domenech et al. were 73% of patients with digestive intolerance benefited from this shift in thiopurine medication.37 We noted that one patient who had an AZA-induced pancreatitis-tolerated MP, a finding also described by others.39 Thiopurine-induced pancreatitis is generally regarded as an idiosyncratic reaction in which a re-challenge with another or the same thiopurine drug will inevitably result in a new episode of pancreatitis.32 However, as some patients seem to tolerate a re-challenge with another thiopurine drug, this might be considered in patients with a strong indication for thiopurine therapy despite previously thiopurine-induced pancreatitis.

It has been reported that concomitant use of sulfasalazine or 5-ASA medication can influence the thiopurine therapy by inhibition of the TPMT enzyme.40, 41 In the present study, a significantly smaller proportion of patients with than without (48% vs. 63%) AE were treated with 5-ASA indicating that the concomitant use of these medications does not increase the risk for AE. We also observed that a greater proportion of patients with than without (60% vs. 34%) AE were treated with steroids. This is not an unexpected finding as the majority of patients experienced their AE during the initial 3 months of thiopurine therapy, when steroids often are used as a ‘bridging therapy’.

It is well recognized that TPMT-deficient patients will develop myelotoxicity when treated with standard doses of thiopurines. Among the six patients with TPMT deficiency four patients treated with moderate to high doses of AZA (0.7–2.9 mg/kg BW) developed severe pancytopenia. The association between other types of thiopurine-related AE and the TPMT polymorphism was less clear. It has, however, been shown that the overall risk for AE is increased among IBD patients with intermediate TPMT activity compared to those with a normal TPMT activity.24, 25, 42 Other studies have shown that patients that are TPMT heterozygous have an increased risk for haematological toxicity.22, 23 In our study, individuals with a low to intermediate TPMT activity (<9.0 U/mL RBC) had an increased risk of developing AE overall. However, there was an overrepresentation of patients with low TPMT activity (TPMT-deficient) in our population (6 vs. 1–2 expected). Only seven of 29 cases of myelotoxicity could be explained by a low to intermediate TPMT activity, a finding in line with earlier observations.33 Thus, the majority of the AE in patients with low to intermediate TPMT activity were non-haematological. However, there was an overrepresentation of AE in low to intermediate TPMT activity, which can at least in part be dose-related. This assumption is further illustrated by the facts that among patients who tolerated AZA there was a significant difference in the dosage between the three TPMT activity groups. Patients with intermediate TPMT activity were treated with about 50% and the patient with low TPMT activity with 10% of standard doses.

Compared to individuals without any AE, the TGN metabolites were significantly increased only in the group of patients who developed gastrointestinal intolerance, while the meTIMP metabolites were significantly increased only in the group with myelotoxicity. This latter finding is in accordance with the results of a recent prospective study from our group.25 Both high TGN (>400) and meTIMP (>11 450) metabolite levels were associated with an increased frequency of AE. Myelotoxicity constituted 50% of AE in the group with TGN levels above 400 and 57% of AE in the group with meTIMP levels above 11 450.

Increased levels of meTIMP (i.e. methylated mercaptopurine metabolites or 6-MMP) have been incriminated as a cause of hepatotoxicity in both ALL16 and among patients with IBD.14 In the latter study 6-MMP levels above 5700 pmol/8 × 108 RBC was associated with a threefold increase in the risk of developing hepatotoxicity. In the present study, metabolite measurements were obtained at the time point for hepatotoxicity in six patients with meTIMP levels between 0 and 5500 pmol/8 × 108 RBC. Furthermore, four of 19 patients with hepatotoxicity had intermediate TPMT activity in the range of 6.7–8.7 U/mL RBC and would theoretically produce low levels of meTIMP. These observations indicate that increased meTIMP levels are not the cause of hepatotoxicity in the majority of cases.

In conclusion, the risk for developing AE during thiopurine therapy was increased among patients with low to intermediate TPMT activity. Determination of a pre-treatment TPMT status might be validated, especially to avoid the serious cases of myelotoxicity seen among patients with low TPMT activity (TPMT-deficient). Measurement of thiopurine metabolites could be useful in selected cases, especially in patients with myelotoxicity or gastrointestinal intolerance. In such cases of toxicity, metabolite measurements can indicate whether a dose reduction might be appropriate. If thiopurine metabolites are measured for other reasons (i.e. lack of efficacy or suspicion of poor patient compliance) and high metabolite levels are noted, a dose reduction or alternative therapies should be considered as these patients have an increased risk of developing AE and, furthermore, might not benefit from thiopurine therapy. Finally, MP therapy should be considered in cases of AZA intolerance, especially in patients with myalgia and arthralgia. We recommend that TPMT status is checked before initiation of thiopurine treatment and that thiopurine metabolites are measured in the case of treatment failure or suspected thiopurine-related AE. The results of this retrospective study must be confirmed in larger prospective studies.


This study was financially supported by The Research Council in the Southeast of Sweden (FORSS), grants F2000-312, P2001-303, F2002-304, F2003-304. Rut och Richard Juhlin's stiftelse 2003, the Swedish Medical Society, grant 2004-685.