Prospective Evaluation of Pharmacogenomics and Metabolite Measurements upon Azathioprine Therapy in Inflammatory Bowel Disease

Abstract Up to approximately 40% to 50% of patients discontinue thiopurine therapy during the course of inflammatory bowel disease (IBD). We investigated the role of the metabolite thiopurine in IBD treatment. This was a prospective study. IBD patients receiving azathioprine (AZA) were prospectively included. Thiopurine methyltransferase (TPMT) genotypes were examined before therapy, and thiopurine metabolite levels were examined at weeks 2, 4, 8, 12, 24, and 48. In total, 132 patients were included. The frequency of leucopenia increased at 6-thioguanine nucleotide (6-TGN) levels ≥420 pmol/8 × 108 RBC (odds ratio [OR] = 7.9; 95% confidence interval (95%CI): 3.5–18.0; P < 0.001) and increased more during the initial 12 weeks of thiopurine therapy (OR = 16.0; 95%CI: 5.7–44.9; P < 0.001). The patients with 6-TGN levels ≥420 pmol/8 × 108 RBC at weeks 4, 8, and 12 had an increased likelihood of leucopenia. Clinical response increased at 6-TGN levels ≥225 pmol/8 × 108 RBC (OR = 13.5; 95% CI: 3.7–48.9; P < 0.001) in Crohn disease (CD) patients. The CD patients with 6-TGN levels ≥225 pmol/8 × 108 RBC at weeks 8, 12, and 24 had an increased likelihood of successful clinical response. TPMT∗3C had a specificity of 100%, but a sensitivity of 8% for predicting leucopenia. A 6-TGN level between 225 and 420 pmol/8 × 108 RBC could be a therapeutic window in patients receiving AZA therapy, and it could likely predict leucopenia in the initial 12 weeks of AZA therapy and a reasonable chance of successful clinical response in CD patients. The value of TPMT genotyping before thiopurine therapy is limited in Chinese patients with IBD, considering the low sensitivity of predicting leucopenia.


INTRODUCTION
T he thiopurine drugs, azathioprine (AZA) and mercaptopurine, have been widely used in the treatment of inflammatory bowel disease (IBD). They can induce and maintain remission with a response rate of 55% to 70%. However, up to approximately 40% to 50% of IBD patients discontinue thiopurine therapy during the course of the disease. 1,2 The occurrence of intolerable adverse events (AEs) is the major reason for drawback in the thiopurine therapy, and the AEs are reported in 10% to 28% of IBD patients. [2][3][4][5][6] Other reported reasons for discontinuation of thiopurines are refractoriness (16%), request by the patient (7%), and pregnancy (1%). 6 After oral absorption, AZA is converted to 6-mercaptopurine via a nonenzymatic reaction, and 6-mercaptopurine is subsequently metabolized to active metabolites, such as 6-thioguanine nucleotides (6-TGNs) and 6-methylmercaptopurine ribonucleotides (6-MMPRs), which have immune modifier activity, and inactive metabolite: 6-methylmercaptopurine . Three enzymes competitively metabolize 6-mercaptopurine: xanthine oxidase, hypoxanthine phosphoribosyl transferase, and thiopurine methyltransferase (TPMT). The enzyme TPMT plays an important role in determination the amount of cytotoxic 6-TGN. Low baseline levels of TPMT activity led to a shift in metabolism of 6-mercaptopurine away from production of the inactive metabolite 6-MMP and toward production of the ''active'' metabolite 6-TGN. 7 The genetic variation of TPMT and its influence on thiopurine metabolism have been well investigated. The literatures showed that individuals with TPMT deficiency who receive standard doses of thiopurines are at significant risk of toxicity, primarily as a result of unchecked production of 6-TGN. [8][9][10][11] However, not all AEs or metabolite patterns can be explained by TPMT deficiency. A study from Madrid showed that the authors could not confirm whether the choice of AZA or a mercaptopurine dose based on TPMT activity prevented leucopenia in patients with IBD. 12 Reports have suggested that variation in TPMT activity accounted for no >10% of overall thiopurine toxicity and approximately one-third of cases of leucopenia. 13 Considerable interest has focused on the metabolism of thiopurines as a means of identifying ways of individualizing therapy to minimize AEs and maximize clinical responses. Previous retrospective meta-analysis studies and cross-sectional studies have confirmed a strong association between 6-TGN concentrations and induction of remission. 14 These studies showed that a 6-TGN level consistently above 230 to 260 pmol/8 Â 10 8 RBC was associated with a favorable response. 14 However, not all related studies have demonstrated an association between higher 6-TGN concentrations and improvement in disease activity. [15][16][17] Meanwhile, not all leucopenia can be explained by a high 6-TGN concentration in the IBD patients. 15,[18][19][20] Given this background, the aims of this study were to evaluate the role of TPMT polymorphisms before treatment, metabolic products of AZA, and disease activity and toxicity in patients with IBD during their first 48 weeks of thiopurine therapy.

Subjects
All patients with IBD between August 2006 and August 2014 at the Gastroenterology Clinic of the First Affiliated Hospital of Zhengzhou University and Sun Yat-sen University were included in our study. The diagnoses of IBD were according to the criteria of Lennard-Jones, 21 which are based on clinical features and endoscopic, radiological, and histopathological findings.
The inclusion criteria and exclusion criteria were defined as described previously. 5

Study Design
Drug doses were initiated at 50 mg daily for AZA (Imurel1, GlaxoSmithKline, Sweden) and 25 mg daily for mercaptopurine (Puri-Nethol1, GlaxoSmithKline) in the first week, and the target doses of nearly 2.0 mg/kg daily for AZA and 1.0 mg/kg daily for mercaptopurine were reached by week 2 without alteration in the following weeks.
Patient data collection and study visits were performed as described previously. 5 TPMT genotype was measured before initiation of therapy. Blood was drawn at weeks 1, 2, 3, 4, 8, 12, 24, and 48 for future analysis of 6-TGN and 6-MMPRs.
The primary endpoint of the study was defined as the patient undergoing AZA treatment for 48 weeks or requiring discontinuation or change of therapy because of AEs or refractoriness.

Classification of Responses
To evaluate the relationship between response to steroid sparing and metabolite levels, the patients receiving AZA plus infliximab, receiving AZA for postoperative prophylaxis, and receiving AZA for endoscopic relapse after surgical resection were excluded from the analysis. A response to steroid sparing was defined as resolution of symptoms upon complete withdrawal of all corticosteroids for at least 6 months (CDAI <150, 22 or Southland Index 2). 23 Treatment failure was ascribed to cases in which the above-stated therapeutic goals were not achieved and there was need for initiation of other immunosuppressive medications or surgery.

TPMT Genotyping
Allele-specific polymerase chain reaction (PCR) and PCRrestriction fragment length polymorphisms were used to determine the frequency of TPMT mutant alleles (ie, TPMTÃ2, TPMTÃ3A, TPMTÃ3B, and TPMTÃ3C) in patients with IBD, as previously described. 5

Erythrocyte Mercaptopurine Metabolite Assay
Erythrocyte lysis buffer was prepared as previously described. 24 The supernatant was kept at 808C until analysis.
The 6-TGN concentration in the erythrocyte lysates was determined using a previously described high-performance liquid chromatography (HPLC) method. 25 The concentration of 6-TGN was normalized to pmol/8 Â 10 8 RBC. 6-MMPRs (pmol/8 Â 10 8 RBC) levels were determined by a reverse-phase HPLC assay developed by Lennard and Singleton. 26 Statistics Statistical analysis was performed using SPSS (Statistical Package for the Social Sciences) software (SPSS Inc, Chicago, IL), version 16.0. Normality was tested for all quantitative variables. Quantitative variables are described as the median and the interquartile range (IQR; q1-q3) throughout. Parametric Student t tests or nonparametric Mann-Whitney U tests, when appropriate, were used to compare quantitative variables between the 2 groups. The differences between multiple independent groups were evaluated by the Kruskal-Wallis test. The x 2 test was used to compare categorical variables. To evaluate the relationship between erythrocyte AZA metabolite concentrations and the AEs, binary logistic regression was used, and the results were expressed as the odds ratio (OR) with a 95% confidence interval (95% CI). Receiver-operating characteristic (ROC) curves were obtained to plot the sensitivity and specificity for various metabolite concentrations to predict the AEs and clinical responses. A P value <0.05 was considered significant.

Ethical Considerations
The study protocol was approved by the ethics committee of the First Affiliated Hospital of Zhengzhou University and Sun Yat-sen University. Each patient provided written informed consent before entering the trial. The study was registered in the International Standard Randomized Controlled Trial Number Register (ISRCTN), with a trial number of ISRCTN58287360.

Subject Characteristics
One hundred and seventy-two patients consented to participate in this study. All patients in this study are ethnic Chinese. Forty subjects were subsequently withdrawn as a result of the administration of AZA or mercaptopurine within 3 months before treatment initiation (n ¼ 22), poor compliance (n ¼ 7), loss of contact (n ¼ 3), and follow-up time <48 weeks (n ¼ 8). The remaining 132 subjects (76 men) had a mean age of 34 years (18-72 years). The majority of subjects (77%) had Crohn disease (CD), with the remainder having ulcerative colitis (UC) (23%). At the study initiation, 47 patients were classified as steroid-dependent, 49 as being on remission maintenance, 24 as having postoperative recurrence of clinical symptoms, and 12 as being on postoperative prophylaxis. Of the 132 patients, 31 patients (23%) were treated with AZA and 5-aminosalicylates (5-ASA) for >4 weeks, and 12 patients (9%) were treated with AZA and infliximab. Table 1 shows the demographic characteristics of the 132 patients.

Adverse Effects
Twenty-eight of the 132 patients (21%) exhibited AEs during therapy. Of the 28 patients, 26 patients (20%) exhibited leucopenia. Of the 26 patients, 19 (73%) exhibited leucopenia during the initial 3 months of thiopurine therapy, 3 patients (2%) exhibited gastric intolerance, 1 patient (0.8%) exhibited flu-like symptoms, and no patient exhibited hepatotoxicity or pancreatitis. The emergence of AEs led to discontinuation of thiopurine therapy in 16 patients (57%). After discontinuation, 11 patients were reintroduced to AZA at half the original dose, 9 patients were shifted to mercaptopurine, 3 patients were shifted to MTX, and 3 patients were shifted to infliximab. Table 2 shows the details of AEs in the 28 patients.

6-TGN and Clinical Responses
To evaluate the relationship between responses to steroid sparing and 6-TGN levels, the patients receiving AZA plus infliximab (n ¼ 12), the patients receiving AZA for postoperative prophylaxis (n ¼ 12), and the patients whose AZA therapy was discontinued in the first 24 weeks (n ¼ 21) because of AEs were excluded from the analysis.
Of the remaining 87 patients, 58 (8 with UC, 50 with CD) were classified as exhibiting a response to steroid sparing and 29 (9 with UC, 20 with CD) as treatment failure. In 17 patients with UC, the 6-TGN levels at weeks 8 and 24 were lower in UC patients with responses to steroid sparing than that in UC patients with treatment failure; the 6-TGN levels at week 12 were not significantly different between the 2 groups. However, the 6-TGN levels at weeks 8, 12, and 24 were higher in CD patients with responses to steroid sparing than in CD patients with treatment failure ( Table 5).

DISCUSSION
Using a prospective study, we evaluated the use of AZA in adult patients with IBD by assessing the association between pretreatment TPMT polymorphisms, metabolic products of AZA, and disease activity and toxicity during their first 48 weeks of thiopurine therapy.
The frequency of leucopenia in this study was higher than previously reported in a Western population, 1 -4,11,12,16-19 which could be explained by the different definition of leucopenia (WBC <3.5 Â 10 9 cells/L), co-administration of 5-ASA, the rapid dose escalation practice, and the fact that the patients were followed for a longer time. The higher percentage of leucopenia could also have been a result of racial differences. The frequency of gastric intolerance was lower than the reported frequency in Western populations. 2,11,17,18 The difference could be explained by the different definitions of gastric intolerance (the occurrence of any or a combination of the following: nausea, vomiting, and abdominal pain with normal amylase and the need to discontinue or change therapy) and racial differences.
No hepatotoxicity was reported in this study, and possible explanations include the low thiopurine dose and racial differences. Pancreatitis is thought to be an idiosyncratic reaction, and the reported frequency was 1% to 11% in Western populations. 3,4,11,17,19 However, no pancreatitis was reported in this study or in other domestic studies, 5,27,28 which could be explained by racial differences. The vast majority of AEs occurred during the initial 3 months of therapy, which is well in line with other observations. 29 A substantial number (27%) of the leucopenia events developed after the initial 3 months of therapy, which is in accordance with a previous study. 5 This observation illustrates the need for continued monitoring of blood counts throughout the duration of thiopurine therapy.
Individual variations in susceptibility to AZA have been attributed to variable intracellular concentrations of the cytotoxic 6-TGN. There is increasing evidence that 6-TGN levels above a certain threshold are associated with a higher risk of leukopenia during AZA therapy. Schütz et al 30 reported that 2 heart recipients with a high 6-TGN level (2211 and 698 pmol/8 Â 10 8 RBCs) developed severe AZAinduced myelosuppression with a fatal outcome, and 6-TGN concentrations that were 450 pmol/8 Â 10 8 RBCs were not associated with myelosuppression in the remaining 18 heart recipients. Subsequently, Gupta et al 31 and Dubinsky et al 19 reported that leukopenia was significantly associated with high 6-TGN levels in 101 children with IBD and in 92 pediatric patients with IBD. However, Cuffari et al 18 found that meTIMP, not 6-TGN levels, was associated with myelosuppression. However, we found that 6-TGN levels were only useful for predicting leucopenia in the initial 3 months of AZA therapy rather than after 3 months of AZA therapy. Another important finding of this study was that the delayed leukopenia in patients receiving AZA is not associated with accumulation of 6-TGN. These findings may influence the clinical application of metabolite measurements.
It has been postulated that some early idiosyncratic reactions, such as gastric intolerance and flu-like symptoms during AZA therapy, might be due to the nitro-imidazole compound found in AZA, which is released to produce 6-mercaptopurine. 32 In accordance with the above results, all 3 patients who reacted with gastric intolerance and the one patient who reacted with flu-like symptoms in this study had low 6-TGN and 6-MMPR levels when the AEs occurred at week 2 on AZA therapy, and a re-challenge with mercaptopurine inevitably resulted in a new episode of gastric intolerance at week 2.
Several previous studies have suggested that higher levels of 6-TGN are associated with clinical responses in IBD. 1,14,19,31  Similarly, 6-TGN levels correlated well with clinical responsiveness to therapy in our patients with CD. However, we found a lack of correlation between clinical response and 6-TGN levels in UC, which is in accordance with the study by Cuffari et al. 33 One possible explanation might be the inconsistent or negative results of AZA for the induction of remission in chronic active UC. [34][35][36] Another possible explanation might be our aggressive use of 5-ASA preparations in 15 of 17 patients with UC. Thus, our results suggested that a 6-TGN level !225 pmol/8 Â 10 8 RBC could be a useful therapeutic target in patients with CD. Similar results have been reported in other studies in which IBD patients in disease remission correlated well with 6-TGN levels >230 and 260 pmol/8 Â 10 8 RBC. 14 An important finding of this study was that 6-TGN levels at weeks 8, 12, and 24 could predict a reasonable chance of successful clinical response in patients with CD. Patient with TPMT H /TPMT L and patient with TPMT L have been demonstrated at significant risk of leucopenia when treated with standard doses of AZA or mercaptopurine. [8][9][10]11,37,38 However, the sensitivity is low in using the TPMT genotype to predict leucopenia. A study from Colombel showed that three-quarters of leucopenia were not explained by TPMT gene mutations. 13 Meanwhile, only 8% of patients with leucopenia had TPMTÃ3C gene mutations in this present study, which corroborates our previous pilot study on IBD. 5 The distribution of TPMT polymorphism is different significantly in different population. The predominant TPMT mutant allele is TPMTÃ3A (3.2-5.7%) in white populations, 39-42 followed by TPMTÃ2 (0.5%) and TPMTÃ3C (0.8%), and the most prevalent mutant allele is TPMTÃ3C (0.5%-1.5%) in Asian populations. 5,28,43 Thus, the value of TPMT genotype measurement before initiation of AZA was limited in predicting leucopenia because of low sensitivity.
Previous studies have demonstrated that co-administration of 5-ASA can result in a higher 6-TGN level in patients with IBD on AZA or mercaptopurine, 16,44,45 which was also confirmed in our study. Of the patients on AZA plus 5-ASA medications, there was an increase in the frequency of clinical leucopenia of 20% and in the 6-TGN level compared with baseline therapy with AZA alone.
It should be noted that there were several potential weaknesses in our study. It is better to use endoscopic and objective biomarkers to define response contrast to the high variability of CDAI. To avoid this weakness, the aim of this study only evaluate the relationship between response to steroid sparing and metabolite levels, the patients on AZA plus infliximab, on AZA for postoperative prophylaxis, and on AZA for endoscopic relapse after surgical resection were excluded from the analysis. Meanwhile, a response to steroid sparing was defined as resolution of symptoms upon complete withdrawal of all corticosteroids for at least 6 months.

CONCLUSIONS
In summary, AZA metabolite measurements may provide clinicians with useful tools to optimize and individualize AZA therapy in IBD. The advantages of metabolite monitoring observed in this study include predicting leucopenia in the initial 3 months of AZA therapy and predicting a reasonable chance of successful clinical response in patients with CD. Another potential value of 6-TGN testing concerns the timing of the testing. Early testing (after 2-4 weeks of therapy) could identify patients likely to benefit from a dose adjustment. Although AZA should be dosed according to body weight, previous research has documented a poor correlation between dose on a per-kilogram basis and 6-TGN levels. Thus, even in patients prescribed the recommended dose of these medications, the measurement of 6-TGN levels may provide additional useful information. Patient with 6-TGN level >420 pmol/ 8 Â 10 8 RBC at week 2 could reduce AZA dose to avoid leucopenia, and increase drug dose for a successful clinical response if patient has 6-TGN level below 225 pmol/8 Â 10 8 RBC. A full blood count may be more important after 3 months of AZA therapy considering that 6-TGN is more appropriate to predict leucopenia in the initial 3 months of AZA therapy. This study also suggests that co-administration of 5-ASA with AZA results in a high frequency of leucopenia in patients, which could be associated with an increase in 6-TGN concentrations. Considering ethnic differences, a larger prospective study in different ethnic groups is now needed to verify this observation.