Single nucleotide polymorphisms associated with methotrexate-induced nausea in juvenile idiopathic arthritis

Methotrexate (MTX) continues to have a key role in the treatment of children with juvenile idiopathic arthritis (JIA) [1‐3]. MTX treatment is commonly associated with nausea. Large inter-individual variation exists in the level of MTX-induced nausea among children with JIA. From mild discomfort to a level of MTX-induced nausea that affects health-related quality of life to such an extent that premature termination is necessary [4‐6]. Genetic factors are thought to influence the inter-individual variation in MTX-induced nausea.

MTX is mainly excreted through the kidneys, but the drug also undergoes enterohepatic circulation [7, 8]. The level of enterohepatic circulation is thought to be associated with the level of MTX-induced gastrointestinal side effects, including nausea [9]. Single nucleotide polymorphisms (SNPs) in genes encoding MTX transporter proteins may alter the function of these transporter proteins and hence affect the enterohepatic circulation of MTX and the level of MTX-induced gastrointestinal side effects [10‐13] (Fig. 1).

MTX inhibits folate pathway enzymes including methylentetrahydrofolatereductase (MTHFR) [14]. This leads to an altered folate homeostasis and specifically the reduced MTHFR activity causes accumulation of homocysteine (Fig. 2), which has been hypothesized to be associated with MTX-induced gastrointestinal adverse events [15]. Specific SNPs in the gene encoding MTHFR have been shown to cause a decreased enzyme-activity [16, 17] and studies in smaller JIA cohorts have indicated that these SNPs may be associated with MTX-induced adverse effects [18, 19].

Nausea-inducing medicine can activate the 5-hydroxytryptamine type 3 (5-HT₃)-receptor placed in the gastrointestinal tract and the medulla oblongata, which causes a signal to be transmitted to the emetic center in the central nervous system [20, 21] (Fig. 3). Studies have shown that SNPs in HTR3A and HTR3B encoding subunits of the 5-HT₃-receptor are associated with impaired receptor function and nausea in adult cancer patients [22, 23]. Hence children with SNPs affecting the function of these genes may be more prone to experience MTX-induced nausea.

The objective for this study was to investigate whether MTX-induced nausea was associated with selected SNPs in candidate genes encoding MTX transporter proteins (SLCO1B1, SLCO1B3, SLC19A1, ABCB1, ABCC2), the MTHFR enzyme (MTHFR) and the 5-HT₃-receptor (HTR3A, HTR3B), in children with JIA.

The study population was composed of children diagnosed with JIA according to the International League of Associations for Rheumatology criteria, treated with methotrexate and aged nine years or above. Children were excluded if cognitively impaired or not fluent in Danish (Fig. 4). All children were followed at our pediatric rheumatology outpatient clinics and enrolled from December 2013 until July 2016.

MTX-induced nausea was determined using two outcome measures previously described [24]: a Danish adaption of the methotrexate intolerance severity score (MISS) [6] and a nausea diary. The outcome measures were developed in electronic versions. The MISS was completed on the day of enrolment by the patients’ parents and they were instructed to focus on their child’s current state of MTX intolerance. The nausea diary was intended to be completed daily by the children for 28 consecutive days following enrolment. It consisted of categorically structured questions covering whether the child had experienced nausea and if so, its timely relation to MTX.

Phenotypes were established based on the two outcome measures. As defined by the MISS development paper [6], a child was categorised as MTX_intolerant if the MISS total score was six or above with at least one point for an anticipatory and/or associative and/or behavioral symptom; otherwise categorised as MTX_tolerant. The phenotype “MTX-nausea” was defined based on the nausea diary. It included children with a nausea diary illustrating a nausea pattern timely related to MTX administration, i.e. the child had to report a clear coherent pattern of nausea on the day of MTX administration and/or the day after; and at least have completed seven diary entries.

On the day of enrolment a blood sample was drawn for the SNP analysis. DNA was extracted from EDTA blood by salt precipitation with triton lysis buffer (Tris 10 mM, MgCl₂ 5 mM, Triton X-100 1% and Sucrose 0.32 M), nucleus lysis buffer (NaCl 75 mM and EDTA 24 mM), saturated NaCl (6 M), isopropanol and Te-buffer (Tris-HCl 10 mM, EDTA 1 mM, pH 7.5). DNA concentration and purity was measured by a Nanodrop spectrophotometer.

For the SNP analysis an amplicon with the specific SNP was created, PCR amplified and then sequenced by Sanger Sequencing. The analysis was performed at Eurofins Genomics using BigDye terminator cycle sequencing chemistry (version 3.1/3.0; Thermo Fisher Scientific), peqStar 96 HPL (PEQLAB Biotechnologie GMBH) and/or GeneTouch (Biozym Scientific GmbH) thermal cyclers, ABI3730xl capillary sequencers (Thermo Fisher Scientific) and GoTaq HotStart Green MasterMix (Promega). The quality scoring of the sequences was based on the quality value at the base-position, the peak-appearance, the overall signal intensity and quality of the sequence, the relation of forward and reverse sequence, and the identification of unspecific/artificial background signals or secondary PCR products and their subsequent signal. The quality scoring was done manually by Eurofins Genomics.

Within the specified candidate genes SNPs were selected based on clinical and genetic criteria. The SNPs should preferably have been proven clinically significant for gastrointestinal intolerance or toxicity, or proven phenotypically significant in animal studies. Further, they should have been studied in a relevant study population, have a reasonable minor allele frequency, a functional significance and preferably cause an amino acid change. The GTExportal.​org was used to see whether the relevant gene section is expressed in relevant tissue. SNPsnap Broad Institute was used to see whether SNPs were in linkage disequilibrium.

All statistical analyses were performed in STATA-13, PLINK 1.9 or haplo.stats R-package. Data on demographics and MTX treatment were analysed with Wilcoxon rank-sum test, chi-squared test/Fisher’s exact test or Student’s T-test, where applicable. The associations between SNP genotypes and phenotypes were analysed with chi-squared test/Fisher’s Exact test in STATA-13. Logistic regression (PLINK 1.9) was used to analyze the additive effects of the minor alleles. The analyses were filtered for Hardy-Weinberg equilibrium and took into account whether SNPs had more than two possible alleles present in the study population. Haplotype based association tests were performed for the two phenotypes – the method implemented in the R-package haplo.stats (https://​CRAN.​R-project.​org/​package=​haplo.​stats) and described by Schaid et al. [25] was used. The method allowed missing alleles and was developed for unrelated subjects, marker-alleles with unknown phase and ambiguous haplotypes. A progressive insertion algorithm was used to compute maximum likelihood estimates of haplotype probabilities. Additive models were chosen for all datasets.

Table 1 shows demographic data, details of the MTX treatment and the distribution of children within the phenotype subgroups. The nausea diary was completed at least one day by 100/121 (83%) of the children. The 77 children who had completed at least seven diary entries were included in the genetic analysis for this phenotype. A total of 8/77 (10%) of the children reported no nausea in all their diary entries.

All patients were treated with folic acid supplement (5 mg) the day after MTX administration. Antiemetic medicine was prescribed for 41/121 (34%) of the children at enrolment.

The selected SNPs within the candidate genes were: SLCO1B1 (rs4149056; rs4149081), SLCO1B3 (rs2117032), SLC19A1 (rs1051266), ABCC2 (rs2273697; rs3740066; rs717620), ABCB1 (rs2032582; rs1045642). No significant associations were found between the genotype distribution for these SNPs and the phenotype subgroups (Table 2), nor any significant additive effect of the minor alleles on either of the two phenotypes (MTX_intolerant or “MTX-nausea”) (Table 3 and Table 4). No significant haplotype associations with either phenotype were found (Table 5 and Table 6).

The selected SNPs were rs1801131 and rs1801133. There was a significant association between the genotype distribution for rs1801133 and the parent-assessed phenotype (the MISS) (p = 0.02); but not for the child-reported phenotype (the nausea diary) (Table 2). There was no significant additive effect of the minor alleles on either phenotype for the two SNPs (Table 3 and Table 4), or any significant haplotype associations (Table 5 and Table 6).

The selected SNPs within the candidate genes were: HTR3A (rs1062613; rs1985242; rs1176713) and HTR3B (rs1176744). No significant associations were found between the genotype distribution and the phenotype subgroups (Table 2). None of these SNPs were significantly associated with the two phenotypes (“MTX-nausea” or MTX_intolerant) with regard to an additive effect of the minor alleles (Table 3 and Table 4). No significant haplotype associations with either phenotype were found (Table 5 and Table 6).