PharmGKB summary: methotrexate pathway

Methotrexate is a folate analog that is used in the treatment of cancers (e.g. acute lymphoblastic leukemia, non-Hodgkin lymphoma, osteosarcoma, and colon cancer) and autoimmune diseases (e.g. rheumatoid arthritis, Crohn’s disease, and psoriasis). In the treatment of autoimmune diseases, methotrexate is usually administrated orally or subcutaneously, whereas in the cancer treatment, it can be given orally, intramuscularly, as intrathecal injections, or as intravenous infusions (up to 12 g/m²) [1-3]. The pharmacokinetics and pharmacodynamics of methotrexate show large interpatient variability regardless of the route of administration or disease being treated [4-6]. The goal of this study is to provide an introduction to methotrexate pharmacogenomics, showing the candidate genes in the PharmGKB methotrexate pathway (Fig. 1), important variants (Tables 1 and 2), discussing key knowledge, and pointing to more in-depth resources.

Fig. 1.

Stylized cell showing candidate genes involved in pharmacokinetics and pharmacodynamics of methotrexate. A fully interactive version is available online at http://www.pharmgkb.org/do/serve?objId=PA2039&objCls=Pathway, and includes additional views that show transport and action in different cell types.

Table 1. Well-studied genes and variants involved in methotrexate pharmacogenomics.

Gene	Variant (rs#)	Phenotype (population) (PMID)	No association (population) (PMID)
MTHFR	677C > T  (rs1801133)	C toxicity/adverse events (pediatric ALL, CML) (16870553  [7]) (16920564 [8]) T toxicity/adverse events (pediatric ALL; NHL; CML; RA,  Black; PsA; RA, White; RA, Asian; JIA, White; pediatric ALL,  Asian) (12453860 [9], 17488658 [10], 11418485 [11],  18381794 [12], 20472929 [13], 17512587 [14],  16501586 [15], 15051775 [16], 11710708 [17],  20863444 [18], 20595278 [19], 18458567 [20]) C efficacy (pediatric ALL; NHL; CML) (14647408 [21];  15781665 [22], 17488658 [10], 17512587 [14],  15385937 [23])	Toxicity (pediatric ALL; pediatric ALL, Asian; RA, Black, White;  RA, Hispanic; Psoriasis; NHL) (12915598 [24], 16439441  [25], 19016697 [26], 20514079 [27], 18368069 [28],  15713801 [29], 16462575 [30], 15781665 [22],  16463153 [31]) Efficacy (psoriasis; NHL; pediatric ALL) (19016697 [26],  16463153 [31], 19827168 [32], 16019535 [33])
MTHFR	1298A>C  (rs1801131)	A toxicity (RA, White; pediatric ALL)(16439441 [25],  18368069 [28], 17323057 [34]) C toxicity/adverse events (NHL; CML; RA; RA Hispanic; CML)  (17488658 [10], 16501586 [15], 16572443 [35],  20514079 [27], 16920564 [8]) A efficacy (RA, White; CML, White) (16572443 [35],  15569990 [36]) C efficacy (RA, Asian; pediatric ALL; CML) (20863444 [18],  14647408 [21], 15385937 [23])	Toxicity (RA, Black; RA, White, Black; Psoriasis, PsA, pediatric  ALL) (16439441 [25], 18381794 [12], 19016697 [26],  20472929 [13], 16870553 [7], 15713801 [29], 15781665  [22], 17512587 [14]) Efficacy (Psoriasis; pediatric ALL) (19016697 [26], 15781665  [22], 17512587 [14], 16019535 [33])
SLC19A1	80G>A  (rs1051266)	A toxicity/adverse events (pediatric ALL; ALL; pediatric ALL,  Asian) (12411325 [37], (17264302 [38], 17180579 [39],  20335220 [40]) G toxicity (pediatric ALL, Asian) (16462575 [30]) A efficacy (RA; RA, Asian; ALL) (17325736 [41], 18322994  [42], 20335220 [40], 19827168 [32])  G efficacy (pediatric ALL) (12411325 [37])	Toxicity (pediatric ALL, PsA) (12915598 [24], 20472929 [13],  18368069 [28], 16870553 [7], 15713801 [29])
TYMS	28 bp repeat  (rs34743033)	3R/2R toxicity (CML) (16920564 [8]) 3R/3R reduced efficiacy (pediatric ALL) (11937185 [43],  12972956 [44], 15713801 [29], 14647408 [21],  18096764 [45])	Efficacy (RA, White) (20386493 [46], 12651279 [47]) Toxicity (pediatric ALL, CML) (16870553 [7], 16501586 [15])

ALL, acute lymphoblastic leukemia; CML, chronic myelogenous leukemia; NHL, non-Hodgkin’s lymphoma; PMID, PubMed identifier; PsA psoriatic arthritis; RA, rheumatoid arthritis.

Table 2.

Additional candidate genes and variants for methotrexate pharmacogenomics

Gene	Studied variants (rs)	PMIDs
ABCB1	3435C>T (rs1045642), 1236 C>T (rs1128503)	19093297 [48], 20386493a[46], 18381794 [12], 20136356 [49]
ABCC1	Arg723Gln (rs4148356), rs3784862, Ala1337Thr (rs28364006), 4002 G>A (rs2230671), IVS 14+115C>T, IVS 18-30C>G (rs2074087)	16041243 [50], 18381794a[12], 20386493 [46], 18256692 [51]
ABCC2	1249 G>A (rs2273697), IVS23+56C>T (rs4148396)	18381794 [12], 20386493 [46]
ABCC3		11585759b[52], 19996279b[53]
ABCC4	934A>C (rs2274407), −1393T>C	19515727 [54]
ABCG2	914C>A (rs2231142), rs17731538	20386493 [46], 12208758b[55],  12874005b[56]
ADA	Val178Val (rs244076), IVS 2 C/G (rs1799880)	19902562 [57]
ADORA1	rs16851020, rs3766553, rs3766554, rs17530497, rs4423076, rs1845466, rs3753475, rs10800901, rs10920574, rs7549561, rs6427993, rs6701725, rs12123037, rs1494487, rs3898276	9214432b[58], 18256692a[51]
ADORA2A	1976T>C (rs5751876), rs5760410, rs2298383, rs3761422, rs2267076, rs2236624, rs9624472, rs4822489	18539621 [59], 19902562 [57]
ADORA3		19790066 [60], 17922624 [61]
AMPD1	34C>T (rs17602729)	16947783 [62], 19902562a[57], 20386493 [46]
ATIC	347C>G (rs2372536) rs4673993	19902562a[57], 16947783 [62], 19016697a[26], 20386493a[46], 19193698 [63]
CBS	844ins68	14647408a[21], 16013960 [64]
CCND1	870A>G (rs9344)	12972956 [44], 16870553 [7], 18096764 [45]
DHFR	35289A>G (rs1232027), −473T>C (rs1650697), 308G > A (rs1105525), rs7387	20472929 [13], 19861437 [65], 18096764 [45], 19902562 [57], 20136356a[49]
FOLH1	1561C>T	12204797a[66]
FOLR1		19493312b[67], 19093297a[48]
FPGS	rs1544105	19902562 [57], 19093297a[48]
GART		15142881b[68]
GGH	452C>T (rs11545078) −401T>C (rs3758149) 16T>C (rs1800909)	20386493a[46], 16999998 [69], 19827168 [32], 15564880 [70], 17286537 [71]
GSTM1	Null	17180579 [39], 15713801 [29]
HPRT1		12429535b[72]
HLA-G	14bp in/del	16906016 [73], 17626975 [74], 19088262a[75], 20136356a[49]
IMPDH2	+787C>T	20136356a[49]
ITPA	94C>A (rs41320251), IVS2 +21A>C (rs7270101)	16947783 [62], 17530705 [76], 19193698a[63], 20136356a[49]
MTHFD1	1958G>A (rs2236225)	14647408 [21], 15797993a[77], 17530705 [76], 20386493a[46]
MTHFR	1793G>A (rs2274976), rs2066462	20863444 [18]
MTHFS		3801490b[78]
MTR	2756A>G (rs1805087)	20386493a[46], 19159907 [79], 18322994 [42], 17855192 [80], 17611986 [81]
MTRR	66A>G (rs1801394)	20386493a[46], 15797993 [77], 18368069 [28]
NOS3	894G>T (rs1799983)	14647408 [21], 16013960 [64]
NP		12429535b[72]
PPAT		14529544b[82]
SHMT1	1420C>T (rs1979277)	20386493 [46], 18368069 [28], 15797993 [77]
SLC22A11		11327718b[83]
SLC22A6		20460822b[84]
SLC22A8		20460822b[84]
SLC46A1		19403800b[85]
SLCO1B1	rs11045879, rs4149081	19901119 [4]
SLCO1A2	Glu172Asp (rs11568563), Arg168Cys (rs11568564), Asn277DEL (rs72559749)	16702441b[86]
TGFB1	869T>C (rs1982073)	20136356a[49]
TLR4	896A>G (rs4986790)	20136356 [49]
TYMS	1494del (TTAAAG)	18381794 [12], 16130010 [87]

Annotations are available in full at http://www.pharmgkb.org/do/serve?objCls=Drug&objId=PA450428&tabType=tabGenetics#tabview=tab2.

PMID, PubMed identifier.

^aNo association.

^bIn-vitro study, (a) folate only, not methotrexate.

The interindividual variability in methotrexate pharmacokinetics can be explained partially by genetic variations in membrane transporter proteins with an affinity for methotrexate [4,89]. In the gastrointestinal tract, methotrexate is absorbed through active transport mediated by the reduced folate carrier (SLC19A1) and possibly also by the proton-coupled folate transporter SLC46A1 (HCP1, PCFT) at the apical membrane of enterocytes [90]. Furthermore, the bioavailability of methotrexate after oral dosing may be affected by ABC transporters, which can move methotrexate out of the enterocytes and back into the intestinal tract (ABCC2, ABCB1, and ABCG2) or into the blood (ABCC1 and ABCC3) [91-93]. Systemic clearance of methotrexate happens primarily through renal glomerular filtration and active secretion over the proximal tubular cells. Several renal transporter proteins have an affinity for methotrexate (SLC22A6, SLC22A8, SLC19A1, ABCG2, ABCC2, and ABCC4), and single nucleotide polymorphisms (SNPs) in ABCC2 have been associated with delayed methotrexate clearance [53,89,94-98]. Hepatic uptake of methotrexate involves the SLCO1B1 and SLCO1B3 transporters, in which SNPs have recently been found to explain up to 10% of the interpatient variability in the clearance of high-dose methotrexate [4,99-101]. Most of the methotrexate in hepatocytes reenters the blood circulation by transporter proteins (ABCC3 and ABCC4) in the basolateral membrane, and only a small portion of the methotrexate is excreted into the bile duct by the ABCC2 and ABCB1 transporters [102-105]. Inside the hepatocytes, aldehyde oxidase can convert methotrexate to 7-hydroxymethotrexate, a metabolite that is eliminated by the same route as methotrexate. At the blood–brain barrier, ABC transporters with affinities for methotrexate are located in endothelial cells, but the effect of facilitated transport of methotrexate out of the cerebrospinal fluid remains to be unclear [106].

The pharmacodynamic profile of methotrexate can, to a large extent, be explained by its interactions with enzymes in the folate pathway. The effects on methotrexate response of variations in these genes have been extensively studied in cancer treatments [107]. At extracellular methotrexate concentrations below 20 μmol/l, methotrexate enters cancer cells primarily through the reduced folate carrier (SLC19A1) [108,109], whereas efflux across the cell membrane is mediated by various ABC transporters; variations in these genes are known mechanisms of drug resistance in cancer cells [110]. Inside the cells, methotrexate is converted to active methotrexate polyglutamates (MTXPGs) by folylpolyglutamate synthetase, which adds glutamate residues to methotrexate [111]. The primary action of methotrexate is inhibition of the enzyme dihydrofolate reductase (DHFR), which converts dihydrofolate to tetrahydrofolate (THF)[112]. Tetrahydrofolate is essential for de novo purine synthesis, and in the biologically active form, 5-methyl-tetrahydrofolate, is an important cofactor in one-carbon metabolism. The effect of methotrexate depends on the function and expression of several other enzymes in the folate pathway, including methylenetetrahydrofolate dehydrogenase (MTHFD1), 5,10-methylenetetrahydrofolate reductase (MTHFR), and thymidylate synthetase (TYMS). Compared with methotrexate, the active metabolites, MTXPGs induce stronger inhibition of the target enzymes (i.e. TYMS and DHFR) and further inhibit key enzymes such as phosphoribosylglycinamide formyltransferase, phosphoribosylglycinamide synthetase, phosphoribosylaminoimidazole synthetase and 5-aminoimidazole-4-carboxamide ribonucleotide formyltransferase/IMP cyclohydrolase (ATIC) in the de novo purine synthesis pathway. MTXPG inhibition results in decreased protein and DNA methylation in addition to impaired DNA formation and repair. MTXPG levels are sustained inside the cells for a longer time than those of methotrexate; degradation of MTXPGs to methotrexate depends on the activity of the lysosomal enzyme γ-glutamyl hydrolase, which catalyzes the removal of polyglutamates [113-115]. MTXPGs have been investigated in relation to clinical outcomes in rheumatoid arthritis (RA) and juvenile idiopathic arthritis (JIA). Specifically, higher concentrations of long-chain MTXPGs have been associated with favorable outcomes in RA [116] and risk of gastrointestinal and hepatic toxicity in JIA [117].

Gene expression and genetic variation in candidate genes have been studied extensively in relation to many methotrexate response measures, including MTXPG accumulation (γ-glutamyl hydrolase, folylpolyglutamate synthetase, and SLC19A1) [108,118,119], reduction in tumor size (SLC19A1 and DHFR) [120], toxicity (MTHFR, and TYMS) [11,15], and relapse (DHFR, TYMS, MTHFR, DHFR, and SLC19A1) [29,37,40,45,54,87,121]. Conflicting results among studies suggest that the effects of genetic variation are therapy dependent and probably reflect the route of administration, dose, and duration of methotrexate treatment [37,40]. Although these studies have contributed to our understanding of the effects of Methotrexate and the molecular mechanisms involved in drug resistance, no genetic variant has yet been prospectively evaluated as a predictor of an outcome in a clinical trial.

Genome-wide studies have linked genes outside the folate pathway to the pharmacokinetics and effects of methotrexate; many of these genes have not been analyzed earlier in studies using the candidate approach. A recent study analyzed the association between MTXPG accumulation and genetic variations such as leukemic cell gene expression, somatic copy number variation, and SNPs [122]. Six genes on chromosome 18 (FHOD3, IMPA2, ME2, SLC39A6, SMAD2, and SMAD4) and one on chromosome 10 (RASSF4) were found to be associated with in-vivo intracellular accumulation of MTXPG in leukemic cells in all three categories of genetic variation. In another genome-wide study of patients with acute lymphoblastic leukemia, in-vivo response to MTX was found to be significantly associated with the expression of genes in the nucleotide pathway (e.g. TYMS), but also with genes involved in cell proliferation and apoptosis, and DNA repair and replication in the leukemic cells [123]. Finally, a genome-wide association study that assessed the link between inherited genomic variation and initial treatment response in patients with acute lymphoblastic leukemia showed 14 SNPs significantly associated with both treatment response and methotrexate clearance or MTXPG accumulation in leukemic cells [124]; early treatment response assessed by eradication of leukemic cells is strongly associated with cure rates and is therefore considered an important clinical phenotype.

No genome-wide association studies have yet been carried out in patients with rheumatoid arthritis, but inherited variations in most genes from the folate pathway have been examined in relation to methotrexate treatment response and toxicity [12,35]. However, to see a clinically relevant effect of genetic variants in the folate, purine, and pyrimidine pathways, it seems crucial to study gene–gene interactions; it has been suggested that the effects of individual SNPs are enhanced when they occur in combination with other common SNPs in these pathways [1]. Combinations of SNPs in the genes ATIC and adenosine receptor 2a have been associated with differential MTXPG concentrations in JIA [117]. The anti-inflammatory effect of methotrexate is further thought to be mediated through interaction with the adenosine biosynthesis pathway [125]. MTXPGs inhibit the enzyme ATIC, which after a cascade of events leads to the accumulation of the anti-inflammatory molecule adenosine; SNPs in genes from the adenosine biosynthesis pathway (i.e. ATIC, inosine triphosphatase, and adenosine monophosphate deaminase 1) have been found to predict the efficacy of methotrexate treatment for RA and JIA [1,62,76,126].

Regardless of disease, it seems clear that future studies should continue to examine the combined effect of variations in multiple genes to characterize the extent of genomic determinants on variation in the pharmacokinetics and pharmacodynamics of methotrexate.