Concomitant 5-aminosalicylic acid treatment does not affect 6-thioguanine nucleotide levels in patients with inflammatory bowel disease on thiopurines

Introduction

For decades, immunomodulating drugs like thiopurines, e.g., azathioprine (AZA) or 6-mercaptopurine (6-MP), have been a mainstay in the long-term treatment of glucocorticoid-dependent or glucocorticoid-refractory inflammatory bowel disease (IBD). Thiopurines were introduced into clinical practice more than 60 years ago. In the 1980s, thiopurines were shown to be effective in Crohn’s disease (CD), and later also in ulcerative colitis (UC) [1,2]. AZA and 6-MP are prodrugs that undergo a complex metabolic transformation, resulting in the formation of the pharmacologically active, immunosuppressive 6-thioguanine nucleotides (6-TGN) [3]. The main metabolites of 6-TGN are active phosphorylated 6-thioguanosine triphosphate (TGTP) and its inactive precursor 6-thioguanosine diphosphate [2,4]. By acting as a purine antagonist, 6-TGN disrupts nucleic acid metabolism and also purine synthesis, which leads to cytotoxicity and immunosuppression by inhibiting DNA, RNA and protein synthesis [5-7]. Tiede et al demonstrated in 2003 that AZA and its metabolites induced apoptosis of T cells in both patients with CD and a control group, by co-stimulation of CD28. This was mediated by binding of AZA-generated TGTP to Rac1 instead of guanosine triphosphate, leading to a specific blockage of Rac1 activation [8]. Later, in 2006, the same investigators reported a suppression of T cell-APC conjugation through inhibition of Vav guanosine exchange activity on Rac proteins [9]. According to the literature, the therapeutic range of 6-TGN concentrations is 235-450 pmol/8×10⁸ erythrocytes (ECs) [1,10].

The conversion of thiopurine drugs is a multistep enzymatic process initiated by hypoxanthine phosphoribosyltransferase [11,12]. Thiopurine methyltransferase (TPMT) is a cytosolic enzyme controlling one of the most important steps in the thiopurine metabolism [12]. It catalyzes the S-methylation of AZA and 6-MP to its ultimate conversion 6-methylmercaptopurine (6-MMP) and 6-methylmercaptopurine ribonucleotide (6-MMPR) [6]. Although 6-MMP is an inactive metabolite, a correlation has been revealed between 6-MMP levels and thiopurine-associated hepatotoxicity, pancreatitis and marrow suppression [13,14].

TPMT is capable of influencing 6-TGN concentrations indirectly by shunting thiopurine drug metabolism away from 6-TGN [15]. The genetic polymorphism of the alleles encoding for TPMT results in various activities of this enzyme, leading to a significant interpatient variability of 6-TGN levels [16,17]. Most individuals are characterized by a high level of TPMT activity, though 11% have intermediate, and approximately 0.3% extremely low or even absent TPMT activity [18]. Patients who carry an intermediate or low TPMT activity phenotype are at elevated risk of myelosuppression under the standard therapeutic dose of thiopurines, due to a potential excess of 6-TGN [19-21]. However, mutation in the TPMT alleles is only responsible for myelosuppression in 27% of cases. Low leukocyte cell counts were more often caused by other factors [22].

In addition to thiopurine, 5-aminosalicylic acid (5-ASA) is frequently used as a co-treatment in IBD. There are conflicting data on whether 5-ASA may influence 6-TGN concentrations, and whether this co-treatment puts patients at greater risk of side-effects from AZA and 6-MP. Small, uncontrolled trials indicated that 5-ASA co-medication might rise 6-TGN levels. Hande et al, in 2006, reported in a retrospective study that 5-ASA therapy was associated with higher 6-TGN levels. However, a main limitation of the study was the inclusion of patients in clinical remission, without considering patients who had to stop the therapy early [5]. De Boer et al described a dose-dependent effect of 5-ASA on 6-TGN levels 1 year later. Co-medication with 2 g 5-ASA daily was associated with a statistically significant increase of 6-TGN levels by 40% (absolute 84 pmol/8x10⁸ EC), whereas 4 g of 5-ASA elevated 6-TGN levels by 70% (absolute 154 pmol/8x10⁸ EC) [23]. Lowry et al suggested that 5-ASA may reversely inhibit TPMT activity in vitro, supporting the experimental work of Szumlanski et al [24,25]. Later, De Graaf et al endorsed this concept by measuring smaller 6-MMPR levels after co-administration of 5-ASA in IBD patients [4]. However, the low absorption rate of 5-ASA raises the question whether this interaction is relevant in vivo [15]. Daperno et al drew the conclusion that 5-ASA seems to have no clinical influence on TPMT activity [26].

Therefore, the aim of the present study was to determine 6-TGN levels in ECs of IBD patients treated with AZA/6-MP, either as monotherapy or in co-treatment with 5-ASA, and whether elevated 6-TGN levels are associated with higher rates of side-effects/worse tolerability or higher therapeutic success.

Materials and methods

Statistical analysis

For comparisons between the 2 groups (5-ASA co-medication vs. no 5-ASA co-medication) in case of discrete outcomes (e.g., sex, disease location or complication [y/n)), we generally used chi-square (χ²) tests. Fisher’s exact tests were implemented when the sample size in a given category for a given group was less than 5. In case of continuous outcomes (for instance 6-TGN or 6-MMPR levels and age), Wilcoxon rank-sum tests were considered. A P-value <0.05 was considered statistically significant (2-tailed). Since 6-TGN and 6-MMPR levels are asymmetrically distributed, their natural logarithm was considered as dependent variable for the multivariate regression model. In order to design each of these 2 models, we first performed univariate regressions with each factor listed in Tables 1-5. Then we fit together all variables so that the corresponding P-value in univariate regressions was <0.2. In the presence of certain variables, others may cease to be significant. The multivariate model was then built by removing nonsignificant covariates one after the other, based on likelihood ratio tests. We then reconsidered each factor and tried to include them in the model. Finally, we checked that no factor in the model could be removed or added, based on likelihood ratio tests. For all analyses we calculated adjusted odds ratios (OR) with 95% confidence intervals (CI). As in the descriptive part above, a P-value <0.05 was considered statistically significant. For all of these analyses, Stata software for PC was used (StataCorp. 2015. Stata Statistical Software: Release 14. College Station, TX: StataCorp LP).

Table 1 Comparison of variables between patients taking and those not taking 5-ASA medication

Table 2 Comparison of disease characteristics between patients taking and those not taking a 5-ASA medication

Table 3 Comparison of 6-TGN and 6-MMPR levels in patients on monotherapy and those receiving 5-ASA co-medication

Table 4 Comparison of disease courses between patients taking and those not taking a 5-ASA medication*

Table 5 Comparison of personality and quality of life between patients taking and those not taking a 5-ASA medication

Results

A total of 144 IBD patients treated with AZA or 6-MP who met the inclusion criteria were identified in the SIBDCS database. Of this population, 34 patients were later excluded because of incomplete/missing information. Of the 110 patients finally enrolled in this study, 40 were receiving a co-medication with 5-ASA at the time of blood sampling.

Factors associated with higher 6-TGN/6-MMPR levels by multivariate linear regression

The distribution of 6-TGN and 6-MMPR levels was asymmetrical. Consequently, we used the logarithm function for the multivariate linear regression model to test the relationship between 6-TGN levels and several clinical parameters (Table 6). The model selection for 6-TGN level returned a model with 3 significant parameters. We identified 6-TGN levels to be significantly higher in non-smokers, patients without prior surgery (intestinal or for fistula) and those without signs of stress-hyperarousal (P=0.03, P=0.01, P<0.001, respectively).

Table 6 Factors associated with elevated 6-TGN levels by multivariate linear regression model

The multivariate linear regression model testing the correlation of 6-MMPR levels resulted in 2 statistically significant parameters (Table 7). We found 6-MMPR levels to be higher in patients without complications and stress-hyperarousal (P=0.02, P=0.001, respectively).

Table 7 Factors associated with elevated 6-MMPR levels by multivariate linear regression model

Discussion

The most substantial finding from this SIBDCS cohort was that co-medication with 5-ASA was not associated with differences in 6-TGN or 6-MMPR levels, as similar 6-TGN and 6-MMPR levels were measured in both groups. Previous small-scale trials described significantly higher levels of whole blood 6-TGN concentration in patients with IBD who received AZA or 6-MP with co-administration of 5-ASA [4,15,32]. However, these studies included only up to 34 patients. limiting their overall power. An expanded investigation was performed by Hande et al in 2006 [5]. They reported an association between 5-ASA and higher 6-TGN levels, but found no difference in 6-MMP levels. A critical point is that this study only included patients in clinical remission, and no patients who had to stop the therapy early.

In a large-scale trial in 2009 (n=183), Daperno et al observed that there was no significant difference in 6-TGN and 6-MMP blood concentrations associated with co-medication in patients on active thiopurine [26]. A potential influence of dual therapy on TPMT activity was not found. Although we did not measure TPMT activity in our cohort, we hypothesize that TPMT activity is not influenced by 5-ASA therapy in vivo. As this enzyme catalyzes the S-methylation of thiopurine to its ultimate conversion 6-MMP and 6-MMPR, an inhibition of TPMT activity would consequently lead to lower 6-MMP/6-MMPR blood concentrations. Remarkably, Hande et al also reported non-altered 6-MMP levels (see above) [5]. These different findings could be interpreted in the context of an adaption that leads to minor interference between thiopurines, 5-ASA and TPMT [25]. By including not only patients in clinical remission, but also in any stage of disease activity, we minimized the influence of a potential adaption.

In addition, a potentially reduced absorption rate in patients suffering from IBD should be acknowledged. This could lead to a false negative result, due to malabsorption of the administered medication. However, an altered absorption rate would mainly influence the results if the 2 compared groups had significantly different disease activity indexes (CDAI and MTWAI). However, in our study, no asymmetrical distribution of disease activity between the compared patients was found. The physician’s decision to prescribe dual therapy or not may have had an impact on the results. The symmetrical distribution of different disease activities/courses in our cohort reduces the risk of such selection bias.

From a mechanistic perspective, combination treatment with 5-ASA and AZA offers a plethora of beneficial effects. In particular, it was revealed in a recent study, utilizing intestinal organoids derived from wild-type mice, that the abovementioned co-medication offered a junctional complex modulation and restoration of epithelial barrier function in a setting of intestinal inflammation. Moreover, 5-ASA—in contrast to AZA (which demonstrated antiproliferative effects)—promoted wound healing of colonic epithelial cells [33].

Within our study, it was not possible to verify adherence to 5-ASA medication. The overall long-term adherence rate in patients with UC taking mesalamine was found to be lower than 50% [34,35]. Therefore, given the absence of such relevant adherence information, the actual administered dosage could not be analyzed meaningfully, although it was between 2.4 and 4.8 g, in accordance with ECCO Guidelines [36]. This might lead to an underestimation of the effect of dual therapy on 6-TGN levels. For future studies, a verification of pill adherence is advisable to reduce the impact of adherence bias. Nevertheless, recent evidence endorses a high dosage of 5-ASA to induce remission in patients with IBD, without monitoring administered dosage or estimating the level of 5-ASA/metabolites [37], since 5-ASA ranks at the top among comparator treatments regarding safety and tolerability [38].

The unbalanced distribution of the diagnosis of CD and UC in our selection of patients may have had a significant impact on the basic 6-TGN concentration of each group. The majority (67.5%) of patients co-treated with 5-ASA had the diagnosis of UC, while 94.3% of patients medicated with only AZA or 6-MP suffered from CD. It was suggested in a previous study that the diagnosis of CD is independently associated with elevated 6-TGN blood concentrations [5]. Further descriptive statistics of co-treated patients were performed to evaluate these concerns for each diagnosis. Using a Wilcoxon rank-sum test, we were unable to establish a significant difference in 6-TGN levels between patients with UC and those suffering from CD (P=0.39).

The shift in the distribution of diagnosis was most probably responsible for the statistically significant difference between the compared groups in 2 of the 12 investigated parameters of disease courses. Fistulas, abscesses or anal fissures occurred in only 10.0% of patients receiving dual therapy at blood collection, whereas a manifestation rate of 41.4% was observed in the non-co-medicated group. Additionally, surgery for fistula was required more often in the non-co-medicated, predominantly CD cohort.

Using a multivariate linear regression model, we tested the relationship between 6-TGN levels and different parameters. We found that the 6-TGN level of a non-smoker was on average 1.54 times greater than that of a smoker. In this respect, emerging scientific evidence supports that nicotine possesses anti-inflammatory properties for both UC and CD [39,40]. In contrast, it is postulated that 6-TGN levels correlate with therapeutic success/remission [13,41]. Further investigations into this interaction are warranted. The multivariate regression model revealed that the parameter of surgery was associated with lower TGN levels, providing further support for the abovementioned idea of Cuffari et al [42]. This theory implies that higher 6-TGN levels may be associated with a better disease course. As we did not observe altered 6-TGN concentrations or disease courses in either of the compared groups, we cannot offer a more specific statement regarding the advantages of elevated 6-TGN levels in relation to a better therapy outcome.

For the same reason, we cannot assert that higher 6-TGN levels are associated with more side-effects. However, in contrast to other studies claiming an elevated risk up to 47% for leukopenia when AZA is paired with 5-ASA [43], in our cohort no hepatitis, leukopenia, pancreatitis or flu-like symptoms were reported. It is worth noting that this finding referred to a monotherapy group with a maximum 6-TGN level of 1232 pmol/8×10⁸ EC, compared to 1069 pmol/8×10⁸ EC in the dual therapy group.

A further limitation of our study was the missing documentation regarding the different formulations of 5-ASA; in this respect, it was recently demonstrated [44] that 5-ASA and N-acetyl-5-ASA levels were significantly higher in IBD individuals receiving time-dependent mesalazine compared to those with pH-dependent mesalazine and multimatrix mesalazine, and that this was also accompanied by greater TPMT inhibition. Moreover, the rather small number of recruited patients has to be acknowledged, even though the present study included more patients than the ones of the past.

In conclusion, our data support no effect of concomitant 5-ASA treatment on 6-TGN and 6-MMPR levels. Treatment-associated side-effects or worse tolerability never occurred. Nevertheless, physicians should be careful in administrating 5-ASA co-medication, although our data warrant neither more frequent laboratory monitoring, nor dose adaptation of AZA in thiopurine patients receiving concomitant 5-ASA treatment.