- Top of page
- MATERIALS AND METHODS
Background: Small uncontrolled trials have suggested that 5-aminosalicylate (5-ASA) medications increase 6-thioguanine nucleotide (6-TGn) levels in adults with Crohn's disease (CD) on azathioprine (AZA) or 6-mercaptopurine (6-MP), presumably through the inhibition of thiopurine methyltransferase (TPMT). We tested the theory that coadministration of 5-ASA agents with AZA/6-MP results in higher 6-TGn levels in a large cohort of children and adults with CD or ulcerative colitis (UC). Methods: A retrospective cohort study identified all children and adults treated for IBD with AZA/6-MP at 2 tertiary medical centers. Patients were included if their TPMT genotype was known and 6-TGn and 6-methymercaptopurine (6-MMP) levels had been obtained after 3 months of clinical remission at a stable dose of AZA/6-MP. 6-TGn and 6-MMP levels were compared between patients taking and those not taking 5-ASA medications through the use of linear regression models to identify and adjust for potentially confounding variables. Results: Of the 126 patients included, 88 were taking 5-ASA medications. Patients on 5-ASA agents had higher mean 6-TGn levels after adjustment for confounding variables (Δ6-TGn, 47.6 ± 21.8 pmol/8 × 108 red blood cells; P = 0.03). CD and TPMT heterozygosity was independently associated with higher 6-TGn levels (P = 0.01 and P = 0.03, respectively). 5-ASA exposure was not associated with a change in 6-MMP levels. Conclusions: We found that 5-ASA therapy is associated with higher 6-TGn levels in children and adults with IBD on 6-MP/AZA. TPMT inhibition may not explain this effect because 5-ASA exposure did not affect 6-MMP levels. The observed association of CD with higher 6-TGn levels is novel and needs to be verified in prospective studies.
It is known that 6-mercaptopurine (6-MP) and its prodrug azathioprine (AZA) have proven efficacy in the maintenance of remission in patients with inflammatory bowel disease (IBD).1–3 The 6-thioguanine nucleotides (6-TGn) are believed to be the active metabolites of 6-MP by acting as purine antagonists, disrupting nucleic acid metabolism and purine synthesis.4 Several studies have shown a correlation between 6-TGn levels and clinical remission rates.5,6 Moreover, 6-TGn levels appear to have an inverse correlation with leukocyte counts, consistent with their presumed immunosuppressive effect.7
6-TGn is derived from 6-MP via a multistep enzymatic pathway initiated by hypoxanthine phosphoribosyltransferase. Two other enzymes compete with hypoxanthine phosphoribosyltransferase for the metabolic transformation of 6-MP. Xanthine oxidase catabolizes 6-MP to thiouric acid through oxidation of its purine ring. Alternatively, thiopurine methyltransferase (TPMT) S-methylates 6-MP as an initial step to its ultimate conversion to 6-methylmercaptopurine (6-MMP). Although not therapeutically active, the 6-MMP metabolite has been shown to correlate with 6-MP-associated hepatotoxicity.6 The metabolism of azathioprine and 6-MP demonstrates significant interpatient variability, in large part because of genetic polymorphisms of the alleles encoding for TPMT.8 Patients with ≥1 mutant alleles for TPMT have significantly reduced enzyme activity. The metabolism of 6-MP is diverted away from 6-MMP in favor of 6-TGn, with a concomitant increased risk of leukopenia.9
5-Aminosalicylic acid (5-ASA) and its various prodrugs (e.g., sulfasalazine, olsalazine, balsalazide) are widely used in the treatment of IBD. Several studies have demonstrated the efficacy of 5-ASA medications in the induction and maintenance of remission in ulcerative colitis (UC).10,11 Although the data are less robust, some studies suggest a role for 5-ASA medications in the induction and postoperative maintenance of remission in patients with Crohn's disease (CD).12 However, the utility of 5-ASA medications as adjuncts to azathioprine and 6-MP in the treatment of moderate to severe IBD is unclear. Currently, the decision whether to continue 5-ASA therapy after the initiation of 6-MP or azathioprine remains at the discretion of the physician (and patient).
Recent studies have revealed that the 5-ASA family of medications may alter thiopurine metabolism. Benzoic acid derivatives, including sulfasalazine and other aminosalicylates, have been shown to inhibit TPMT in vitro.13–16 However, the low rates of absorption of the 5-ASA compounds raise the question of whether this interaction is meaningful in vivo. Clinical studies investigating this potential interaction in patients with IBD are limited to a few trials with small numbers of patients. One nonrandomized trial of 31 adults with CD on 6-MP or AZA found a trend toward higher rates of leukopenia after the introduction of sulfasalazine or mesalamine but not balsalazide.17 Among its secondary endpoints, the study revealed an increase in mean whole blood 6-TGn levels in patients given sulfasalazine or mesalamine at most but not all study time points. In a second trial of 16 adults with CD on AZA, mean 6-TGn levels were observed to fall modestly after the withdrawal of 5-ASA medications. Interestingly, TPMT activity was unaffected.18
Convincing evidence supporting an in vivo interaction between 6-MP/AZA and aminosalicylates has not been presented in a large cohort of patients with IBD. Indeed, no studies to date address the possibility of such an interaction in patients with UC for whom the evidence for lifelong 5-ASA therapy is arguably stronger. Moreover, no data exist as to whether an interaction between 6-MP/AZA and 5-ASA medications occurs in a pediatric population. We sought to observe the effect of 5-ASA medications on 6-MP and AZA metabolism across a broad age spectrum in both UC and CD.
- Top of page
- MATERIALS AND METHODS
We identified 336 patients (135 children, 201 adults) treated with AZA or 6-MP for IBD at our 2 institutions between January 2003 and January 2005. Of these, 209 patients had no visit that met the specified inclusion criteria. Most excluded patients had all previous metabolite testing done in the setting of active disease, concomitant steroid or infliximab therapy, or both. Several had never had metabolite and/or TPMT genotype testing. A total of 126 patients met the criteria for inclusion in the cohort study (62 children, 64 adults). Demographic data, TPMT genotypes, and clinical disease status are summarized in Table 1. The median patient age was 22 years (range 6–79 years). Sixty-seven patients (53%) were male; 98 (78%) had a diagnosis of CD; and 28 (22%) had UC. One hundred seventeen (93%) were TPMT genotype homozygous wild-type (+/+), and 9 (7%) were TPMT heterozygotes (+/−). No patients were found to be homozygous mutant (−/−).
Table 1. Cohort Demographics
|Patients, no. (%)||126|
| Pediatric||62 (49)|
| Adult||64 (51)|
|Sex, no. (%)|| |
| Male||67 (53)|
| Female||59 (47)|
|Age, yr|| |
| Median (range)||22 (6–79)|
|Disease, no. (%)|| |
| UC||28 (22)|
| Modified Mayo score 0||24|
| Modified Mayo score 1||1|
| Modified Mayo score 2||3|
| CD||98 (78)|
| HBI 0||50|
| HBI 1||20|
| HBI 2||20|
| HBI 3||8|
|TPMT genotype, no. (%)|| |
| Homozygous normal||117 (93)|
| Heterozygous||9 (7)|
| Homozygous mutant||0 (0)|
|Metabolite data|| |
| Mean ± SD 6-TG||231.1 ± 116.0|
|Median 6-MMP (range)||1917.0 (187–21,350)|
One hundred eight patients (84%) were treated with 6-MP, with a mean dose of 1.2 ± 0.4 mg/kg. Eighteen patients (16%) were treated with AZA at a mean dose of 1.8 ± 0.6 mg/kg. After conversion of AZA doses to 6-MP equivalents, the mean adjusted 6-MP dose for the entire cohort was 1.1 ± 0.4 mg/kg.
The mean 6-TGn level in the cohort was 231.1 ± 116.0 pmol/8 × 108 red blood cells (RBCs; Table 1). The 6-MMP levels for the cohort had a nonparametric distribution. The median 6-MMP level was 1917.0 pmol/8 × 108 RBCs, with a range of 187 to 21,350 pmol/8 × 108 RBCs.
Eighty-eight patients (70%) in the cohort were taking a 5-ASA medication at the time of 6-TGn and 6-MMP measurements. Of these, 77 (88%) were prescribed mesalamine (38 Asacol, Proctor & Gamble Pharmaceuticals, Phoenix, Ariz; 39 Pentasa, Ferring-Shire Pharmaceuticals, Florence, Ky). Six patients were taking balsalazide, and the remaining 5 patients were taking sulfasalazine. Median and range doses for each 5-aminosalicylate are shown in Table 2. When converted to mesalamine equivalents, the balsalazide doses were comparable to those of patients on Asacol and Pentasa. Patients given sulfasalazine had a lower median dose and range after conversion to mesalamine equivalents. The median adjusted mesalamine dose for the entire cohort was 43.8 mg/kg (range 9.0–87.3 mg/kg).
Table 2. 5-ASA Treatment Regimens
|Medication||Patients, no.||Median Dose and Range, mg/kg||Mesalamine Equivalent, mg/kg|
|Sulfasalazine|| 5||47.6 (26.3–95.2)||18.3 (10.1–36.6)|
|Mesalamine|| || || |
| Asacol||38||53.7 (16.9–87.3)||53.7 (16.9–87.3)|
| Pentasa||39||40.4 (9.0–67.8)||40.4 (9.0–67.8)|
|Balsalazide|| 6||110.9 (37.5–76.5)||38.8 (13.1–61.7)|
The patients taking a 5-ASA medication were similar to those not taking these drugs in terms of baseline demographics, disease type, and TPMT genotype (Table 3). In addition, the mean 6-MP dose, mean AZA dose, and mean adjusted 6-MP dose were comparable between the 2 groups (Table 3).
Table 3. Comparison of Variables Between Patients Taking and Those Not Taking a 5-ASA Medication
| ||5-ASA||No 5-ASA|
|Male sex, no. (%)||44 (50)||23 (61)|
|Adult patients, no. (%)||45 (51)||19 (50)|
|Median age (range), yr||25 (6–79)||21 (6–78)|
|CD, no. (%)||65 (74)||33 (87)|
|TPMT wild-type, no. (%)||82 (93)||35 (92)|
|Patients taking 6-MP, no. (%)||76 (86)||32 (84)|
|Mean dose 6-MP, mg/kg||1.2 ± 0.4||1.2 ± 0.4|
|Patients taking AZA, no. (%)||12 (14)||6 (16)|
|Mean dose AZA, mg/kg||1.6 ± 0.4||2.2 ± 0.9|
|Mean equivalent dose 6-MP (entire cohort), mg/kg||1.1 ± 0.4||1.2 ± 0.4|
In a crude analysis, patients prescribed a 5-ASA medication had a mean 6-TGn level of 242.7 ± 115.0 pmol/8 × 108 RBCs compared with a mean 6-TGn level of 204.2 ± 115.5 pmol/8 × 108 RBCs among patients not taking a 5-ASA drug (P = 0.09). Patients taking a 5-ASA medication had a median 6-MMP level of 2248.5 pmol/8 × 108 (range 187-21,350 pmol/8 × 108) RBCs, whereas patients not taking a 5-ASA medication had a median 6-MMP level of 1781.0pmol/8 × 108 (range 216-19,087 pmol/8 × 108) RBCs. The mean ln 6-MMP of patients taking 5-ASA medications was 7.6 ± 1.3 compared with 7.4 ± 1.3 among patients not taking a 5-ASA (P = 0.41).
Bivariate linear regression was used to test potential confounders affecting the crude relationship between 5-ASA exposure and 6-MP metabolite levels. Age, sex, institution (pediatric vs adult), and drug choice (6-MP vs AZA) showed no significant association with 6-TGn levels (Table 4). Two variables (TPMT genotype and disease type) were each associated with a difference in 6-TGn levels at a value of P < 0.1. TPMT heterozygotes tended to have higher 6-TGn levels compared with homozygous normal patients. Patients with CD tended to have higher mean 6-TGn levels compared with patients with UC. The adjusted 6-MP dosage was not different in patients with CD and patients with UC to account for the different 6-TGn levels in the 2 disease groups (P = 0.64).
Table 4. Association Between Patient Variables and 6-TGn Metabolite Levels by Bivariate Linear Regression Analysis
| ||6-TGn Level, pmol/8 × 108RBCs||P|
|Age||−0.9/yr of life||0.16|
|Sex|| || |
| Male||238.7 ± 123.4||0.44|
| Female||222.5 ± 107.5|| |
|Institution|| || |
| Pediatric||246.7 ± 120.0||0.14|
| Adult||215.9 ± 110.9|| |
|Disease|| || |
| UC||190.8 ± 99.9||0.04|
| CD||242.6 ± 118.2|| |
|TPMT genotype|| || |
| Homozygous normal||225.6 ± 114.0||0.06|
| Heterozygous||302.0 ± 126.4|| |
|Antimetabolite|| || |
| 6-MP||236.0 ± 118.9||0.25|
| AZA||201.8 ± 94.4|| |
The relationship between 5-ASA exposure and 6-TGn levels was then tested by multivariate analysis with adjustment for TPMT genotype and disease type. Patients taking 5-ASA medications had a significantly higher adjusted mean 6-TGn level compared with patients not taking a 5-ASA drug (P = 0.03). After adjustment for confounding variables, 5-ASA exposure was associated with a 47.6 ± 21.8 pmol/8 × 108 RBC increase in 6-TGn levels (Table 5). In this multivariate regression model, CD and TPMT heterozygous genotype continued to be predictive of higher 6-TGn levels (P = 0.01 and P = 0.03, respectively). The overall fit (R2) of the model was 0.10. The relationship between 5-ASA exposure and 6-MMP levels remained statistically insignificant after adjustment for all potentially confounding variables.
Table 5. Factors Associated With Increased 6-TGn Levels by Multivariable Linear Regression Model
| ||Change in 6-TGn levels (mean ± SE), pmol/8 × 108RBCs|| |
|5-ASA exposure (vs no exposure)||47.6 ± 21.8||0.03|
|CD (vs UC)||63.3 ± 24.1||0.01|
|TPMT heterozygous (vs homozygous wild-type)||85.6 ± 38.6||0.03|
Adjusted mesalamine dosages were tested for correlation with 6-TGn levels to examine for a possible dose-response effect. The Pearson coefficient for a linear association between adjusted mesalamine dose and 6-TGn levels was 0.16 (P = 0.14). After adjustment for TPMT genotype and disease type, this association remained statistically insignificant (P = 0.22). Adjusted mesalamine dosages also were separated into quartiles to test for a nonlinear association between higher 5-ASA dosages and higher 6-TGn levels. ANOVA showed no significant difference between the mean 6-TGn levels in each quartile (P = 0.60).
One hundred twenty-one patients in the cohort (96.0%) had a WBC count measured at the time of the metabolite measurement. Higher 6-TGn levels correlated with lower WBC counts (Pearson coefficient, −0.19, P = 0.03) in an analysis of the entire cohort. The mean WBC count and rate of leukopenia were not significantly different between patients exposed to and those not exposed to 5-ASA medication (Table 6). Patients receiving a 5-ASA medication had a mean MCV of 90.3 ± 7.4 fL compared with 86.6 ± 7.9 fL in patients not on these drugs (P = 0.008).
Table 6. Effect of 5-ASA Exposure on Hematological and Biochemical Profile
| ||5-ASA P||No 5-ASA||P|
|Hematological parameters|| || || |
| Mean ± SD WBC (103cells/μL)||5.6 ± 1.9||5.8 ± 2.3||0.60|
| % WBC < 3.5||10.6||11.1||0.93|
| Mean ± SD MCV (fL)||90.3 ± 7.4||86.3 ± 7.9||0.008|
|Transaminase parameters|| || || |
| Median ALT (range) (U/L)||20 (4–299)||13 (3–191)||0.68|
| % ALT > 60||7.1||12.1||0.39|
One hundred seventeen patients in the cohort (92.9%) had an ALT measurement at the time of 6-MP metabolite assays. Higher lognormal 6-MMP levels correlated with higher ALT levels (Pearson coefficient, 0.34; P < 0.001). ALT levels and the rate of transaminase elevation were not different in the 2 exposure groups (Table 6).
- Top of page
- MATERIALS AND METHODS
In our cohort of 126 children and adults with UC or CD in remission, exposure to 5-ASA medications was associated with an increase in mean 6-TGn levels after adjustment for TPMT genotype and disease type. This difference was observed despite comparable mean adjusted 6-MP doses between the 2 exposure groups. These findings are consistent with previous studies that have reported higher 6-TGn levels when 5-ASA agents are combined with 6-MP or AZA in adults with CD.17,18 They are not, however, entirely consistent with the hypothesis that this effect is borne through TPMT inhibition. Instead of observing a decrease in 6-MMP levels, we observed a statistically insignificant increase in 6-MMP levels among patients undergoing 5-ASA therapies.
Previous studies have established that 5-ASA medications and their metabolites inhibit TPMT in vitro.13–16 However, this effect has not been shown to exist in vivo. Indeed, 2 prior trials of adults with CD showed a trend toward increased TPMT activity with exposure to a 5-ASA medication, although this effect did not achieve statistical significance.17,18 We were unable to test directly for this relationship in our study because most patients had not had their TPMT activities measured concurrently with their 6-MP metabolites. Our finding that 5-ASA therapy is associated with increased 6-TGn but not decreased 6-MMP levels suggests that 5-ASA medications may affect AZA and 6-MP metabolism through mechanisms other than TPMT inhibition. Clinical trials directed toward this question are warranted.
In our cohort, TPMT genotype was an independent predictor of 6-TGn levels, with heterozygous patients having higher 6-TGn levels compared with homozygous normal patients. This finding was consistent with previous studies that demonstrate higher 6-TGn levels in patients with reduced TPMT activity, as well as clinical expectations based on our understanding of 6-MP metabolism.6,23 We also observed an association between 6-TGn levels and disease type. Patients with CD had higher mean 6-TGn levels than patients with UC despite comparable 6-MP dosages. Age, sex, choice of drug (AZA vs 6-MP), and institution (pediatric vs adult) had no significant impact on the relationship between 5-ASA exposure and 6-TGn levels.
Ours is the first study to describe the association between 5-ASA and 6-TGn levels in populations other than adults with CD. We observed a comparable effect in adults and children. Thus, adult and pediatric gastroenterologists alike should be aware of this drug interaction when treating patients with IBD. Moreover, the association between 5-ASA exposure and 6-TGn levels holds for patients with UC and for patients with CD.
We did not observe a linear relationship between mesalamine dose and 6-TGn levels. Because of the large interpatient variance of 6-TGn levels, we likely were underpowered to detect a modest drug-response effect. Alternatively, it is possible that low doses of mesalamine are sufficient to affect an increase in 6-TGn levels and that higher doses have no additive impact. This question would be best addressed by a large prospective trial.
Our study presents data on a much larger cohort of patients than any previous investigation into this drug interaction. However, its retrospective design precludes us from proving causation. We adjusted for anticipated confounders (TPMT genotype and 6-MP/AZA dose) and tested for other potential confounding variables. Conceivably, differences in TPMT activity between our 2 exposure groups could have caused us to observe different 6-TGn levels. However, this possibility is unlikely given the identical distribution of TPMT genotypes, relatively large cohort size, and similar 6-MMP levels.
In this retrospective study, we were unable to verify adherence with 5-ASA therapy by pill count or pharmacy records. Previous studies have shown that adherence to mesalamine therapy is substantially <100%; thus, we probably overestimated our exposure of interest.24 However, overestimation of 5-ASA exposure would bias our finding toward the null hypothesis. Therefore, we expect that the association between 5-ASA dose and 6-TGn level may be even stronger than reported here.
The decision to give 5-ASA therapy introduced the possibility of selection bias, in that physicians may have been compelled to give dual therapy (6-MP plus 5-ASA) to sicker patients. We attempted to minimize any effect of disease severity by limiting our study to patients in remission. We know of no reason to think that 6-MP metabolism would be different in patients with previously severe IBD in remission compared with patients with milder IBD in remission.
Although the relationship between TPMT genotype and 6-TGn levels was anticipated, we were surprised to find an independent association between disease type and 6-TGn levels. To the best of our knowledge, a difference in remission 6-TGn levels between patients with UC and CD has not been previously reported. Patients with active CD may be expected to have lower 6-TGn levels for a given dosage of 6-MP on the basis of impaired small bowel absorption, but it is not clear why patients with quiescent CD would have higher 6-TGn levels. Theoretically, patients with CD may have had lower TPMT activities in our cohort. However, we observed no difference in 6-MMP levels in patients with UC compared with patients with CD, arguing against a difference in TPMT activity. It is conceivable that adherence with 6-MP therapy was better among patients with CD, although again we would expect higher 6-MMP levels. Further investigation into a possible differential metabolism of 6-MP between patients with UC and CD is required.
We observed 6-TGn levels to correlate inversely with WBC counts, corroborating previous reports of the same relationship.6,7 Similarly, 6-MMP correlated with ALT, consistent with earlier studies indicating that high 6-MMP levels are predictive of hepatotoxicity.6 MCV levels were higher among patients receiving 5-ASA therapy. Previous studies have demonstrated an association between MCV and 6-TGn levels.25,26 The higher MCV levels seen in our patients receiving 5-ASA therapy thus are likely attributable to the higher 6-TGn levels in this exposure group. Exposure to 5-ASA therapy was not associated with leukopenia or even a statistically significant change in WBC counts in our study. This finding suggests that although adding mesalamine to 6-MP therapy may increase 6-TGn levels, the likelihood of precipitating a clinically meaningful drop in leukocyte counts is low. In our multivariate regression model, we predict that the addition of mesalamine is associated with a rise in 6-TGn levels of ≈48 pmol/8 × 108 RBCs, a relatively modest amount.
Nonetheless, the described interaction between 5-ASA therapy and 6-MP or AZA should be considered by all physicians who prescribe these medications. Patients with low TPMT activity or taking high 6-MP doses may be at particular risk for high 6-TGn levels and leukopenia if they are also given 5-ASA drugs. Conversely, this interaction may be clinically advantageous in certain cases. Previous studies have indicated that higher 6-TGn levels are associated with higher rates of remission.5,6 Therefore, the addition of a 5-ASA medication may help a subset of patients who have difficulty reaching “therapeutic” 6-TGn levels, such as those with high TPMT activity. A trial of combination therapy (5-ASA plus 6-MP/AZA), if feasible, may be considered in refractory IBD patients before antimetabolite therapy is discontinued. We anticipate further research to inform us about the efficacy of combined therapy versus 6-MP/AZA alone in this population.