Pharmacogenetics of Pemetrexed Combination Therapy in Lung Cancer: Pathway Analysis Reveals Novel Toxicity Associations

Home , Antimetabolite, Combination therapy

ORIGINAL ARTICLE Pharmacogenetics of pemetrexed combination therapy in lung cancer: pathway analysis reveals novel toxicity associations

A Corrigan1,6, JL Walker2,6, S Wickramasinghe1, MA Hernandez1, SJ Newhouse3, AA Folarin3, CM Lewis2, JD Sanderson4, J Spicer5 and AM Marinaki1

Identification of polymorphisms that influence pemetrexed tolerability could lead to individualised treatment regimens and improve quality of life. Twenty-eight polymorphisms within eleven candidate genes were genotyped using the Illumina Human Exome v1.1 BeadChip and tested for their association with the clinical outcomes of non-small cell lung cancer and mesothelioma patients receiving pemetrexed/platinum doublet chemotherapy (n ¼ 136). GGH rs11545078 was associated with a reduced incidence of grade X3 toxicity within the first four cycles of therapy (odds ratio (OR) 0.25, P ¼ 0.018), as well as reduced grade X3 haematological toxicity (OR 0.13, P ¼ 0.048). DHFR rs1650697 conferred an increased risk of grade X3 toxicity (OR 2.14, P ¼ 0.034). Furthermore, FOLR3 rs61734430 was associated with an increased likelihood of disease progression at mid-treatment radiological evaluation (OR 4.05, P ¼ 0.023). Polymorphisms within SLC19A1 (rs3788189, rs1051298 and rs914232) were associated with overall survival. This study confirms previous pharmacogenetic associations and identifies novel markers of pemetrexed toxicity.

The Pharmacogenomics Journal (2014) 14, 411–417; doi:10.1038/tpj.2014.13; published online 15 April 2014

INTRODUCTION (FPGS).16 Polyglutamation serves for both prolongation of cellular Pemetrexed is a folate antimetabolite approved for the treatment retention of pemetrexed and also increases the affinity of the 5,17 of advanced non-small cell lung cancer (NSCLC) both in drug for its cellular targets. Gamma-glutamyl hydrolase (GGH) combination with a platinum agent in the first-line setting and catalyses the removal of these glutamic acid residues from as single agent maintenance or second line therapy.1 In addition, pemetrexed, facilitating extrusion from the cell and reducing the 18 it has proven efficacy in the treatment of mesothelioma.2,3 activity of the drug. Of interest, polymorphic variants within GGH Cytotoxicity is achieved through simultaneous inhibition of three are reported to be associated with significant pemetrexed toxicity, 9,10 key enzymes, which results in the disturbance of nucleotide pools as well as altered response and survival rates (Table 1). and disruption of DNA and RNA synthesis. Pemetrexed is primarily Although high rates of pemetrexed resistance have driven a a potent inhibitor of thymidylate synthetase and to a lesser extent number of pharmacogenetic studies investigating the predictive of glycinamide ribonucleotide formyltransferase and dihydrofolate or prognostic influence of polymorphisms within the enzymatic reductase (DHFR).4,5 targets and drug metabolic pathway, few have focussed on the Several transporters are known to be involved in the intracel- impact on treatment tolerability (Table 1). Pemetrexed therapy lular uptake of pemetrexed. Reduced folate carrier 1 (RFC1, can be complicated by dose-limiting haematological and gastro- 19,20 SLC19A1) has a high affinity for pemetrexed6 and the importance intestinal toxicity, with subsequent dose reductions and treat- of this transport mechanism is illustrated by studies that demon- ment delays potentially impairing treatment efficacy. Identification strate a complex interplay between alterations in SLC19A1 gene of powerful pharmacogenetic markers would enable informed expression and sensitivity to pemetrexed cytotoxicity.7,8 Further- treatment choices, which maximise efficacy and maintain quality more, a number of polymorphic SLC19A1 variants have been of life through avoidance of side effects. shown to influence pemetrexed tolerability and survival after Response to pemetrexed therapy and the incidence of adverse therapy (Table 1).9–11 More recently, the proton-coupled folate drug reactions (ADRs) are likely to be due to a complex interplay transporter has been shown to possess high affinity for peme- between variation in genes encoding pemetrexed targets, as well trexed and has been confirmed as an important mechanism of as those involved in the metabolism of the drug. To reflect this cellular influx.12,13 A third route of cellular uptake of uncertain complex situation, Illumina Human Exome v1.1 chip genotyping significance, is mediated by folate receptors (FR) -a and -b, which was used to assess the influence of single nucleotide polymorph- are anchored to the cellular membrane, transporting folate isms (SNPs) within 11 genes involved in the pemetrexed pathway derivatives via a complex endocytotic mechanism.6,14,15 including SLC19A1, FOLR3 (FR-g), FPGS, GGH, DHFR, GART (GARFT), Upon entering the cytoplasm, pemetrexed is rapidly polygluta- MTHFR, MTHFD1, MTR, SHMT2 and ERCC2 (Table 2).21 We predicted mated in a process mediated by folylpolyglutamate synthetase that variation within these genes, involved in pemetrexed uptake

1Purine Research Laboratory, GSTS Pathology, Guy’s and St Thomas’ Hospital NHS Foundation Trust, 4th Floor, North Wing, St Thomas Hospital, Lambeth Palace Road, London, UK; 2Department of Medical and Molecular Genetics, King’s College London, 8th Floor Tower Wing, Guy’s Hospital, London, UK; 3National Institute for Health Research (NIHR) Biomedical Research Centre for Mental Health at South London and Maudsley NHS Foundation Trust and (Institute of Psychiatry) King’s College London, London, UK; 4Department of Gastroenterology, Guy’s and St Thomas’ Hospital NHS Foundation Trust, College House, St Thomas’ Hospital, London, UK and 5Division of Cancer Studies, King’s College London, Guy’s Hospital, London, UK. Correspondence: Dr AM Marinaki, Purine Research Laboratory, GSTS Pathology, Guy’s and St Thomas’ Hospital NHS Foundation Trust, 4th Floor, North Wing, St Thomas Hospital, Lambeth Palace Road, London SE1 7EH, UK. E-mail: [email protected] 6These authors contributed equally to this work. Received 21 November 2013; revised 17 January 2014; accepted 19 February 2014; published online 15 April 2014 Pharmacogenetics of pemetrexed combination therapy A Corrigan et al 412 Table 1. Previously reported pharmacogenetic associations inﬂuencing pemetrexed outcomes

Gene dbSNP ID Change Association Study participants P-value Evidence (n)

SLC19A1 rs914232 c.190–688A4G GG genotype have increased grade 48 0.05 9 X3 leukopaenia rs2838958 c.1151 þ 2117C4T TT genotype have longer OS 54 0.03 10 rs3788189 c.1294–1517A4C AA genotype have longer OS 54 0.03 10 rs1051298 3’-UTR: c.*64C4T CC genotype have longer OS 54, 45 0.03, 0.022 10,11 FPGS rs138322374 c. À 327G4A GA genotype risk of high ALT 54 0.07 10 GGH rs3780126 c.109 þ 1307C4T T allele causative of increased grade 48 0.03, 0.04 9 X3 lymphopaenia, TT genotype have shorter OS rs7010484 c.500–930G4A GA genotype better response rates and 54 0.06, 0.03 10 decreased risk grade 4 ADRs DHFR rs442767 c. À 1188C4A CC genotype increased fatigue 90 0.008 28 MTHFR rs1801133 c.665C4T; p.Ala222Val TT genotype longer PFS and OS 208 0.012, 26 0.026 rs1801131 c.1286A4C; AC/CC genotype associated with shorter OS and 65 0.001, 27 p.Glu429Ala PFS 0.019 ERCC2 rs13181 c.2251A4C; CC genotypes associated with shorter PFS and 208 0.021, 26 p.Lys751Gln OS 0.044 Abbreviations: ADRs, adverse drug reactions; ALT, alanine transaminase; DHFR, dihydrofolate reductase; ERCC2, excision repair cross-complementing rodent repair deﬁciency, complementation group 2; FPGS, folylpolyglutamate synthase; GGH, gamma-glutamyl hydrolase; MTHFR, methylenetetrahydrofolate reductase; OS, overall survival; PFS, progression-free survival; SLC19A1, solute carrier family 19 (folate transporter), member 1.

and metabolism, would affect treatment tolerability, response and Analysis Software v1.0 clustering algorithm in Genome Studio v2011.1 survival in this mixed NSCLC and mesothelioma cohort. (Illumina). This clustering data was further refined to improve rare variant genotype call rates using the zCall algorithm.22 Finally, data underwent quality control for sex mismatch, individual call rate of 497%, SNP call rate of 499%, Hardy–Weinberg Equilibrium P-value o10 À 6, cryptic relatedness MATERIALS AND METHODS and minor allele frequency 40.02. Variants with a minor allele frequency Patients and methods o0.02 are unlikely to be genuinely informative of true associations. 136 Subjects were recruited to a retrospective pharmacogenetic study at The selection of candidate genes was based on previously reported Guy’s and St Thomas’ Hospitals NHS Foundation Trust. Appropriate ethical pharmacogenetic studies (Table 1) and additionally, included genes identi- approval was obtained and written consent was provided by all fied as potentially influencing pemetrexed transport and metabolism.21 participants. Patients with cytological or histological confirmation of After quality control, of the 192 SNPs within 19 genes selected for malignant mesothelioma (n ¼ 42) or advanced NSCLC at stage XIIIB investigation, 20 SNPs within 11 genes remained for association tests (n ¼ 94) were eligible for inclusion. All patients had received at least one (Table 2). The majority of SNPs were omitted from the analysis on the basis cycle of pemetrexed therapy at a standard dose of 500 mg m À 2 in of low minor allele frequency and this is reflective of the chip design, combination with either cisplatin or carboplatin. which is intended to enable rare variant analysis. Physical examination, medical history, demographics (sex, age and A further seven intronic SNPs and one exonic SNP of interest, that were ethnicity) as well as baseline full blood count, renal and liver function tests not present on the chip, were genotyped by TaqMan assay (Applied were completed as part of a pretreatment assessment. Before each cycle of Biosystems, Warrington, UK) using a Stratagene Mx3005P RT-PCR instru- therapy, patients were evaluated for treatment tolerance and had ment (Agilent Technologies, Edinburgh, UK) (Table 2). All SNPs had a monitoring of full blood count, liver and renal function profiles. Clinical genotyping completeness of 499%. outcome data were obtained from standardised oncology outcome records completed at each clinic visit. All ADRs occurring within the first four cycles of treatment were graded Statistical analysis in accordance with the National Cancer Institute Common Terminology Of the initial 136 study participants, 132 were included in toxicity analyses Criteria for Adverse Events Version 4 (14 June 2010). Treatment-related with four patients excluded due to inadequate toxicity data. In total, 113 adverse events were considered severe if they reached or exceeded participants had adequate radiological evaluation data for inclusion in Common Terminology Criteria for Adverse Events grade 3. response analyses. Principal components analysis was carried out using Radiological evaluation was completed before the start of therapy, mid- EIGENSOFT v3.023 on all common SNPs (minor allele frequency 40.01) to treatment and post the final cycle of therapy. Radiological surveillance was account for the structure of genetic variation across races. Adjusting for completed every 2 months thereafter, unless clinical indication dictated principal components models population stratification within study otherwise. participants due to this variation. The continuous variables of the Overall survival (OS) was defined as the time from the first day of principal components contain more information about genetic variation therapy to the date of death due to any cause. Individuals lost to follow-up than stratifying by ethnicity or including ancestry as a covariate in the were censored at the date of last contact. Progression-free survival (PFS) regression analysis and consequently should reduce false positive results.24 was defined as the time from the first day of therapy to the date of disease Logistic regression analyses in PLINK v1.07 were used to explore progression as evidenced by radiological evaluation, or death by any associations between genetic variants and incidence of ADRs and response cause. Individuals lost to follow-up or alive without disease progression under an additive model.25 All logistic regression models were adjusted for were censored at the date of their last follow-up evaluation. the first four principal components. Four ADRs were analysed (gastrointestinal toxicity, neutropaenia, haematological toxicity and any toxicity at grade X3), within cycles 1–2 and cycles 1–4, giving a total of eight ADR Genotyping and quality control outcomes. DNA was extracted from EDTA whole blood using the QIAamp DNA Blood Analyses of OS and PFS were performed using Cox’s proportional Mini Kit (Qiagen, Crawley, West Sussex, UK). Illumina Human Exome v1.1 hazards model to allow adjustment for the first four principal components. DNA Analysis BeadChip Kits (Illumina UK, Saffron Walden, UK) were used to To give two groups of individuals, a dominant mode of action for the genotype 4240 000 SNPs across the human exome and chips were minor allele was assumed, testing for differences in survival between subsequently imaged using an Illumina iScan following the manufacturer’s individuals with the homozygous wildtype genotype and individuals who instructions. Genotype calling was carried out using Illumina’s GenCall Data carried at least one copy of the minor allele.

The Pharmacogenomics Journal (2014), 411 – 417 & 2014 Macmillan Publishers Limited Pharmacogenetics of pemetrexed combination therapy A Corrigan et al 413 Table 2. Polymorphic variants included in the analysis of grade X3 toxicity, response and survival

Gene dbSNP ID Base change/amino Minor allele dbSNP MAF Cohort MAF Genotype Genotyping acid change distribution platform

SLC19A1 rs1051266 c.80A4G p.His27Arg G 0.491 0.555 GG 47 Chip GA 57 AA 32 SLC19A1 rs1051298 c.*746C4T T 0.495 0.449 CC 46 TaqMan CT 58 TT 32 SLC19A1 rs3788189 c.1294–529A4C A 0.475 0.533 AA 44 TaqMan AC 57 CC 35 SLC19A1 rs914232 c.190–688A4G G 0.500 0.563 AA 31 TaqMan AG 57 GG 48 SLC19A1 rs2838958 c.1151 þ 2117C4T T 0.411 0.522 CC 34 TaqMan CT 62 TT 40 FOLR3 rs61734430 c.292C4T p.Arg98Cys T 0.023 0.048 CC 125 Chip CT 9 TT 2 FOLR3 rs1802608 c.585C4T p.His195His T 0.083 0.070 CC 117 Chip CT 19 TT 0 FPGS rs10106 c.*192T4C C 0.478 0.430 TT 44 Chip TC 67 CC 25 FPGS rs138322374 c.138 þ 28G4A A 0.018 0.048 GG 124 TaqMan GA 11 AA 1 GGH rs11545077 c.91G4A p.Ala31Thr A 0.203 0.246 GG 81 Chip GA 43 AA 12 GGH rs11545078 c.452C4T p.Thr151Ile T 0.071 0.085 CC 115 Chip CT 19 TT 2 GGH rs3780126 c.109 þ 1307C4T T 0.410 0.360 CC 53 TaqMan CT 68 TT 15 GGH rs7010484 c.500–930G4A G 0.161 0.320 AA 61 TaqMan AG 63 GG 12 DHFR rs1650697 c. À 473T4C T 0.227 0.228 CC 80 Chip CT 50 TT 6 GART rs6517178 c.2107 þ 675A4G G 0.481 0.426 AA 50 Chip AG 56 GG 30 GART rs8788 c.1261G4A p.Val421Ile G 0.128 0.202 AA 88 Chip AG 41 GG 7 MTHFR rs17367504 c.236 þ 160T4C C 0.110 0.162 TT 95 Chip TC 38 CC 3 MTHFR rs1801133 c.665C4T p.Ala222Val T 0.325 0.327 CC 61 Chip CT 61 TT 14 MTHFR rs1801131 c.1286A4C p.Glu429Ala C 0.228 0.300 AA 66 TaqMan AC 57 CC 12 MTHFR rs17375901 c.1531–80G4A A 0.030 0.055 GG 122 Chip GA 13 AA 1 MTHFR rs2274976 c.1781G4A A 0.051 0.044 GG 125 Chip p.Arg594Gln GA 10 AA 1 MTHFD1 rs1950902 c.401A4G p.Lys134Arg A 0.191 0.188 GG 88 Chip GA 45 AA 3 MTHFD1 rs10498514 c.1494 þ 476A4C C 0.058 0.029 AA 130 Chip AC 4 CC 2

& 2014 Macmillan Publishers Limited The Pharmacogenomics Journal (2014), 411 – 417 Pharmacogenetics of pemetrexed combination therapy A Corrigan et al 414 Table 2. (Continued)

Gene dbSNP ID Base change/amino Minor allele dbSNP MAF Cohort MAF Genotype Genotyping acid change distribution platform

MTHFD1 rs2295639 c.1495–171T4C C 0.050 0.029 TT 130 Chip TC 4 CC 2 MTHFD1 rs2236225 c.1958G4A A 0.346 0.386 GG 52 Chip p.Arg653Gln GA 63 AA 21 MTR rs1805087 c.2756A4G G 0.193 0.224 AA 81 Chip p.Asp919Gly AG 49 GG 6 SHMT2 rs73338162 c.149C4T p.Ser50Leu T 0.024 0.029 CC 129 Chip CT 6 TT 1 ERCC2 rs13181 c.2251A4C p.Lys751Gln C 0.237 0.287 AA 67 Chip AC 60 CC 9 Abbreviations: DHFR, dihydrofolate reductase; ERCC2, excision repair cross-complementing rodent repair deﬁciency, complementation group 2; FOLR3, folate receptor 3 (gamma); FPGS, folylpolyglutamate synthase; GART, phosphoribosylglycinamide formyltransferase, phosphoribosylglycinamide synthetase, phosphoribosylaminoimidazole synthetase; GGH, gamma-glutamyl hydrolase; MAF, minor allele frequency; MTHFD1, methylenetetrahydrofolate dehydrogenase (NADP þ dependent) 1, methenyltetrahydrofolate cyclohydrolase, formyltetrahydrofolate synthetase; MTHFR, methylenetetrahydrofolate reductase; MTR, melhionine synthetase; SHMT2, serine hydroxymethyltransferase 2 (mitochondrial); SLC19A1, solute carrier family 19 (folate transporter), member 1. Genotyping platform: Chip, genotype determined using the Illumina Human Exome v1.1 DNA analysis BeadChip; TaqMan, genotype determined using Applied Biosystems TaqMan assay.

SPSS software v21 was used for the survival analyses (SPSS, Chicago, IL, severe ADRs (Table 4). The minor T allele of GGH rs11545078 USA). No P-value corrections for multiple testing were applied as this study c.452C4T conferred a protective effect against grade 3 or 4 was primarily exploratory with a view to replicating previously reported toxicity across both cycles 1–2 (P ¼ 0.027, odds ratio (OR) 0.099) associations. To this end, Po0.05 was considered statistically significant, and cycles 1–4 (P ¼ 0.017, OR 0.25). Furthermore, the rs11545078 whereas the threshold for significance using the Bonferroni method of variant T allele was protective of severe haematological toxicity correction for multiple comparisons testing would be (1) P 0.0018 (0.05/ o across the first four cycles of therapy with each copy of the T allele 28 SNPs) or (2) Po0.00018, correcting for 28 SNPs and 10 outcome variables (eight ADRs, OS, PFS). However, these thresholds are highly conferring an OR of 0.13, P ¼ 0.050 (Table 4). conservative as there is a correlation between the SNPs and outcome A polymorphism within the promoter region of DHFR, variables. rs1650697 c.-473T4C, was predictive of severe toxicity within cycles 1–4 of therapy. Individuals with rs1650697 CT or TT genotypes had an increased incidence of grade 3 or 4 toxicity with RESULTS an OR of 2.20, P ¼ 0.027. Patient characteristics The ERCC2 variant rs13181 c.2251A4C has previously been reported to be associated with shorter PFS and OS in patients The baseline demographics and clinical characteristics of study 26 participants are shown in Table 3. This mixed cohort consisted of treated with pemetrexed and carboplatin combination therapy. 69% (n ¼ 94) advanced stage NSCLC and 31% (n ¼ 42) malignant In the present study, the minor C allele indicated protection mesothelioma patients. All participants received an initial dose of against occurrence of severe gastrointestinal toxicity within the pemetrexed at 500 mg m À 2, with the exception of two patients first four cycles of therapy, but failed to reach significance who had 20% and 25% dose reductions for the first cycle due to thresholds (P ¼ 0.051, OR 0.30). age and renal function, respectively. Cisplatin was the chosen combination agent for 63.2% of study participants and the Analysis of response remaining 36.8% received carboplatin. Mid-treatment radiological evaluation was available for 113 Across the first four cycles of therapy, 44.7% (59/132) of patients patients. Analysis of response at first scan revealed that a single experienced Common Terminology Criteria for Adverse Events polymorphic variant in FR-g (FOLR3), rs61734430 c.292C4T grade X3 treatment-related toxicity. Furthermore, participants (p.Arg98Cys), was significantly associated with an increased risk who experienced grade X3 ADRs spent a total of 169 days in of progressive disease. Individuals harbouring one or more variant hospital compared with only 21 days for those with grade p2 T alleles were at increased likelihood of disease progression at toxicity. Over the course of the first four cycles of therapy, 33% of mid-treatment scan with an OR of 3.80 (95% confidence interval patients required a dose reduction and 31% had treatment delays. 1.15–12.53, P ¼ 0.029). No other significant associations were Treatment was discontinued in 25 out of 136 patients, 10 of which found. were attributable to toxicity and 15 due to disease progression. Haematological and gastrointestinal toxicities occurred most Survival analysis frequently, affecting 25.8% and 11.4% of the study population, respectively. All 136 patients recruited to this study had sufficient data for survival evaluation. Analysis of MTHFR rs1801133 c.665C4T revealed that patients with the CC genotype had longer OS Analysis of ADRs (19.4 months) compared with those with polymorphic CT or TT Twenty-eight SNPs were tested for associations with ADRs in 132 genotypes (13.8 months). Patients carrying one or more copies of eligible study participants. Logistic regression analysis revealed the variant T allele had an increased risk of early death with a associations between variants in GGH and DHFR, and the onset of hazard ratio (HR) of 1.56, P ¼ 0.048 (Table 5).

The Pharmacogenomics Journal (2014), 411 – 417 & 2014 Macmillan Publishers Limited Pharmacogenetics of pemetrexed combination therapy A Corrigan et al 415 Three polymorphisms within the reduced folate carrier gene Table 3. Demographics and cohort characteristics of this pemetrexed- SLC19A1, which are in incomplete linkage disequilibrium, were treated cohort associated with OS (Table 5). Individuals with rs3788189 AA (n) (%) genotypes had a longer OS (20.9 months) when compared with those with AC and CC genotypes (14.2 months). Presence of one Disease type or more copies of the C allele conferred a HR of 1.71 for increased Non-small cell lung carcinoma 94 69.1 risk of shorter survival (P ¼ 0.026). Similarly, patients with Mesothelioma 42 30.9 rs1051298 CC genotypes had a median survival of 20.9 months, Total 136 whereas those with CT or TT genotypes survived for a median of 13.8 months. Cox regression analysis revealed that patients Co-drug Cisplatin 86 63.2 harbouring one or more copies of the T allele had an increased Carboplatin 50 36.8 risk of death with a HR of 1.76, P ¼ 0.016. In addition, SLC19A1 rs914232 homozygous AA genotypes were signiﬁcantly associated Race with shorter OS with a HR of 1.99, P ¼ 0.007 (Table 5), surviving for Caucasian (European ancestry) 114 83.8 a median of 12.1 months compared with patients with AG or GG African ancestry 14 10.3 genotypes who had a median survival of 17.9 months. Asian 1 0.7 None of the polymorphic variants investigated in this cohort Other 6 4.4 were signiﬁcantly associated with PFS. Not stated 1 0.7

Smoking status Never 25 18.4 DISCUSSION a Ex-smoker 45 33.1 We have found that interindividual variation in tolerability and Current smoker 56 41.2 outcome with pemetrexed-containing regimens remains a sub- Not stated 10 7.4 stantial issue. In our cohort, 7.6% of patients required discontinua- tion of treatment due to ADRs and a further 33% of patients Grade X3 ADRs (cycles 1 þ 2) Affected Unaffected % required dose reductions or treatment delays. In addition to Gastrointestinal 12 120 9.1 having a negative impact upon treatment efficacy and quality of Neutropaenia 11 121 8.3 life, there is also a substantial resource burden on the healthcare All haematological ADRs 19 113 14.4 service, with 169 days spent in hospital due to grade X3 ADRs in All ADRs 41 91 31.1 this current cohort alone. Identification of clinically useful pharmacogenetic loci remains an important and relevant research Grade X3 ADRs (cycles 1–4) goal aiming to provide a personalised approach to chemotherapy. Gastrointestinal 15 117 11.4 In a setting such as NSCLC where several drug combinations have Neutropaenia 23 109 17.4 equivalent efficacy, ascertaining which individuals are at risk of All haematological ADRs 34 98 25.8 severe ADRs or poor response before therapy would facilitate All ADRs 59 73 44.7 informed selection of treatment and possible improvement in Mid-treatment response clinical outcomes. Complete response 0 0 To date, there is limited data regarding the influence of Partial response 27 23.9 pharmacogenetic loci in pemetrexed treatment. The majority of Stable disease 68 60.2 studies have involved small cohorts, (mean study size 85 in the six Progressive disease 18 15.9 most relevant studies), and have investigated variants in a limited number of candidate genes.9–11,26–28 The present study utilised Treatment changes the Illumina Human Exome BeadChip, a technology that enables Dose reduction 43 32.6 assessment of genetic variation across the exome, allowing Delay 41 31.1 variation in candidate genes involved in the pemetrexed Overnight hospital 29 22.0 admission pathway to be interrogated. A limitation of this technique is the omission of many of the intronic variants previously found to be Survival event Time to event Event associated with pemetrexed outcomes, although these were (months) (n) genotyped using alternate methods. Two GGH polymorphisms have been reported to influence Death 15.8 92 pemetrexed tolerability.9,10 An intronic GGH variant, rs3780126 Disease progression 6.8 119 c.109 þ 1307C4T, has been correlated with increased rates of Abbreviation: ADRs, adverse drug reactions. aDefined as subjects who quit grade X3 lymphopaenia, and the homozygous TT genotype has 9 smoking 45 years before diagnosis. also been associated with shorter OS, whereas a second GGH variant c.500–930G4A rs7010484, has been linked to favourable

Table 4. SNPs showing signiﬁcant associations with grade X3 toxicity outcomes (Po0.05)

Gene dbSNP ID Minor allele Toxicity association Across cycles P-value OR 95% CI

GGH rs11545078 T All ADRs 1 þ 2 0.0270 0.099 0.013–0.770 GGH rs11545078 T All ADRs 1–4 0.0165 0.245 0.078–0.774 DHFR/MSH3 rs1650697 T All ADRs 1–4 0.0272 2.204 1.093–4.444 GGH rs11545078 T Haematological 1–4 0.0498 0.129 0.017–0.998 Abbreviations: ADRs, adverse drug reactions; CI, conﬁdence interval; DHFR/MSH3, dihydrofolate reductase/mutS homologue 3; GGH, gamma-glutamyl hydrolase; OR, odds ratio.

& 2014 Macmillan Publishers Limited The Pharmacogenomics Journal (2014), 411 – 417 Pharmacogenetics of pemetrexed combination therapy A Corrigan et al 416 Table 5. SNPs associated with OS adjusted for PCs 1–4 using Cox regression analysis

Gene dbSNP ID Variant Frequency n (%) OS months (median) Genotype Hazard ratio 95% CI P-value

MTHFR rs1801133 c.665C4T CT þ TT 1.56 1.01–2.44 0.048 CC 61 (44.9) 19.4 CT þ TT 75 (55.1) 13.8 SLC19A1 rs3788189 c.1294–529A4C AC þ CC 1.71 1.07–2.73 0.026 AA 44 (32.4) 20.9 AC þ CC 92 (67.6) 14.2 SLC19A1 rs1051298 c.*746C4T CT þ TT 1.76 1.11–2.78 0.016 CC 46 (33.8) 20.9 CT þ TT 90 (66.2) 13.8 SLC19A1 rs914232 c.190–688A4G AA 1.99 1.21–3.27 0.007 AA 31 (22.8) 12.1 AG þ GG 105 (77.2) 17.9 Abbreviations: CI, conﬁdence interval; MTHFR, methylenetetrahydrofolate reductase; OS, overall survival; SLC19A1, solute carrier family 19 (folate transporter), member 1.

response rates and a decreased incidence of grade 4 ADRs.10 A progressive disease at mid-treatment scan observed in third GGH polymorphism, c.452C4T (rs11545078) was previously rs61734430 c.292C4T variant allele carriers. These findings infer reported not to influence pemetrexed toxicity.26 However, in our an important role for FR-g in the transport of pemetrexed. cohort, variant genotypes conferred protection against severe Interestingly, FOLR3 is expressed in haematopoietic tissues,33 toxicity across all cycles, particularly haematological toxicity, but suggesting that it regulates folate transport in these tissues. In our did not significantly alter outcome. Previously reported functional study, the presence of FOLR3 rs61734430 was associated with less studies of the rs11545078 T allele using methotrexate, an anti- severe haematological toxicity, but this did not reach significance folate drug with structural similarity to pemetrexed, have (P ¼ 0.063). demonstrated decreased GGH enzyme activity due to conforma- In one of the largest studies to date, MTHFR rs1801133 variant tional changes within the catalytic cleft-tail. This decreased allele carriers are reported to have increased PFS and OS when enzyme activity led to the accumulation of polyglutamated treated with pemetrexed.26 The MTHFR rs11801133 variant T allele substrates.29 It is likely that this variant would also result in is associated with reduced enzyme activity and would be increased pemetrexed polyglutamates, hence the protective effect expected to be associated with a favourable clinical response to conferred is contrary to expectations. Interestingly, the decrease in therapy. However, in our cohort, this variant did not significantly catalytic activity appears to be limited to long chain polyglutamates, impact upon PFS. The association found between individuals with with short chain polyglutamates still effectively hydrolysed.29 As variant T genotypes and a shorter OS contradicts the above study, the affinity of pemetrexed for its enzymatic targets is greatly but was of marginal significance and may reflect the smaller altered in relation to the degree of polyglutamation, it is possible cohort size in our study. that the inhibitory profile of the drug differs in individuals with Three polymorphic SLC19A1 variants that are in incomplete variant rs11545078 genotypes. Clearly, further studies are required linkage disequilibrium with one another were associated with OS. to elucidate the underlying functional mechanism. SLC19A1 rs3788189 AA genotypes have previously been found to DHFR is one of the enzymatic targets of pemetrexed and correlate with longer OS in pemetrexed-treated NSCLC patients;10 catalyses the formation of tetrahydrofolate, which acts as a formyl this finding was replicated in the current study with the C allele donor required for de novo purine synthesis. Promoter region carrying a HR for shorter OS of 1.71. Similarly, Adjei et al.,10,11 polymorphisms in DHFR have recently been linked to severe found that rs1051298 CC genotypes were also associated with fatigue following pemetrexed treatment (Table 1).28,30 A polymor- longer OS, a finding that has recently been replicated in a study phic DHFR variant c. À 473T4C (rs1650697) located in the of Asian NSCLC patients and also reached significance in the 50-promoter region of the gene correlated with an increased risk current study. Furthermore, a third SLC19A1 polymorphism that of severe ADRs in our study cohort. The minor T allele forms part has previously been linked to increased incidence of severe of a promoter region haplotype that is reported to upregulate leukopaenia, but not OS, was associated with survival in the DHFR expression, but the presence of this variation alone does not current study. SLC19A1 rs914232 AA genotypes resulted in shorter appear to alter mRNA levels.30 The promoter region of the DHFR OS following pemetrexed therapy. These findings provide gene is poorly understood and our finding is consistent with convincing evidence for the influence of SLC19A1 polymor- rs1650697 variant genotypes being associated with decreased phisms on the survival rates of pemetrexed-treated lung cancer DHFR expression. patients, consistent with the central role of the reduced folate Resistance to pemetrexed therapy is common and has driven carrier on pemetrexed uptake. pharmacogenetic studies in this field. Indeed, in our cohort, only Finally, although retrospective, this is one of the largest studies 24% of patients experienced response to treatment and 16% had of pemetrexed pharmacogenetics to date. It is also noteworthy disease progression at the mid-treatment scan. Although the that all patients received combination therapy and, other than importance of FRs in pemetrexed transport has been considered ERCC2, we have not considered pharmacogenetic influences on negligible, current evidence indicates a more prominent role. FR-a concomitant platinum therapy. This study has provided an is highly expressed in mesothelioma, lung cancers and other important replication of previously reported pharmacogenetic tumour types, and a recent study found that overexpression of associations and the use of an exome chip for genotyping has FR-a in NSCLC tumour specimens correlated with favourable enabled the identification of novel variants influencing peme- outcomes with pemetrexed-based therapy.31,32 In our mixed trexed therapy outcomes. Future studies should test the role of NSCLC and mesothelioma cohort, we found that germline such predictive markers to personalise pemetrexed therapy and variation in FR-g (FOLR3) was associated with increased rates of improve both efficacy and tolerability.

The Pharmacogenomics Journal (2014), 411 – 417 & 2014 Macmillan Publishers Limited Pharmacogenetics of pemetrexed combination therapy A Corrigan et al 417 CONFLICT OF INTEREST 14 Chattopadhyay S, Wang Y, Zhao R, Goldman ID. Lack of impact of the loss of The authors declare no conflict of interest. constitutive folate receptor alpha expression, achieved by RNA Interference, on the activity of the new generation antifolate pemetrexed in HeLa cells. Clin Cancer Res 2004; 10: 7986–7993. ACKNOWLEDGMENTS 15 Nutt JE, Razak AR, O’Toole K, Black F, Quinn AE, Calvert AH et al. The role of folate receptor alpha (FRalpha) in the response of malignant pleural mesothelioma to We are grateful for the assistance given by Hamel Patel. We acknowledge the pemetrexed-containing chemotherapy. Br J Cancer 2010; 102: 553–560. financial support from Guy’s and St Thomas’ Charity, the Purine Metabolic Patient 16 Habeck LL, Mendelsohn LG, Shih C, Taylor EC, Colman PD, Gossett LS et al. Association (PUMPA) and from the UK Department of Health via the National Institute Substrate specificity of mammalian folylpolyglutamate synthetase for 5,10- for Health Research (NIHR) Biomedical Research Centre (BRC) award to Guy’s & dideazatetrahydrofolate analogs. Mol Pharmacol 1995; 48: 326–333. St Thomas’ NHS Foundation Trust in partnership with King’s College London and 17 Touroutoglou N, Pazdur R. Thymidylate synthase inhibitors. Clin Cancer Res 1996; King’s College Hospital NHS Foundation Trust. We also acknowledge support from 2: 227–243. the NIHR BRC for Mental Health at South London and Maudsley NHS Foundation 18 Rhee MS, Ryan TJ, Galivan J. Glutamyl hydrolase and the multitargeted antifolate Trust and the Institute of Psychiatry, King’s College London. King’s College London is LY231514. Cancer Chemothe Pharmacol 1999; 44: 427–432. an Experimental Cancer Medicine Centre. 19 Rusthoven JJ, Eisenhauer E, Butts C, Gregg R, Dancey J, Fisher B et al. Multi- targeted antifolate LY231514 as first-line chemotherapy for patients with advanced non-small-cell lung cancer: A phase II study. National Cancer Institute of REFERENCES Canada Clinical Trials Group. J Clin Oncol 1999; 17: 1194. 1 Scagliotti GV, Parikh P, von Pawel J, Biesma B, Vansteenkiste J, Manegold C et al. 20 Clarke SJ, Abratt R, Goedhals L, Boyer MJ, Millward MJ, Ackland SP. Phase II Phase III study comparing cisplatin plus gemcitabine with cisplatin plus peme- trial of pemetrexed disodium (ALIMTA, LY231514) in chemotherapy- trexed in chemotherapy-naive patients with advanced-stage non-small-cell lung naive patients with advanced non-small-cell lung cancer. Ann Oncol 2002; 13: cancer. J Clin Oncol 2008; 26: 3543–3551. 737–741. 2 Green MR. Alimta (pemetrexed disodium): a multitargeted antifolate for the 21 Whirl-Carrillo M, McDonagh EM, Hebert JM, Gong L, Sangkuhl K, Thorn CF et al. treatment of mesothelioma. Lung Cancer 2002; 38: S55–S57. Pharmacogenomics knowledge for personalized medicine. Clin Pharmacol Ther 3 Vogelzang NJ, Rusthoven JJ, Symanowski J, Denham C, Kaukel E, Ruffie P et al. 2012; 92: 414–417. Phase III study of pemetrexed in combination with cisplatin versus cisplatin 22 Goldstein JI, Crenshaw A, Carey J, Grant GB, Maguire J, Fromer M et al. zCall: a rare alone in patients with malignant pleural mesothelioma. J Clin Oncol 2003; 21: variant caller for array-based genotyping: genetics and population analysis. 2636–2644. Bioinformatics 2012; 28: 2543–2545. 4 Taylor EC, Kuhnt D, Shih C, Rinzel SM, Grindey GB, Barredo J et al. A dideazate- 23 Patterson N, Price AL, Reich D. Population structure and eigenanalysis. PLoS Genet trahydrofolate analogue lacking a chiral center at C-6, N-[4-[2-(2-amino-3,4- 2006; 2:e190. dihydro-4-oxo-7H-pyrrolo[2,3-d]pyrimidin-5-yl)ethyl]benzoyl]-L-glutamic acid, is 24 Price AL, Patterson NJ, Plenge RM, Weinblatt ME, Shadick NA, Reich D. Principal an inhibitor of thymidylate synthase. J Med Chem 1992; 35: 4450–4454. components analysis corrects for stratification in genome-wide association stu- 5 Shih C, Chen VJ, Gossett LS, Gates SB, MacKellar WC, Habeck LL et al. LY231514, a dies. Nat Genet 2006; 38: 904–909. pyrrolo[2,3-d]pyrimidine-based antifolate that inhibits multiple folate-requiring 25 Purcell S, Neale B, Todd-Brown K, Thomas L, Ferreira MA, Bender D et al. PLINK: a enzymes. Cancer research 1997; 57: 1116–1123. tool set for whole-genome association and population-based linkage analyses. 6 Westerhof GR, Schornagel JH, Kathmann I, Jackman AL, Rosowsky A, Forsch RA Am J Hum Genet 2007; 81: 559–575. et al. Carrier- and receptor-mediated transport of folate antagonists targeting 26 Tiseo M, Giovannetti E, Tibaldi C, Camerini A, Di Costanzo F, Barbieri F et al. folate-dependent enzymes: correlates of molecular-structure and biological Pharmacogenetic study of patients with advanced non-small cell lung cancer activity. Mol Pharmacol 1995; 48: 459–471. (NSCLC) treated with second-line pemetrexed or pemetrexed-carboplatin. Lung 7 Wang Y, Zhao R, Goldman ID. Decreased expression of the reduced folate carrier Cancer 2012; 78: 92–99. and folypolyglutamate synthetase is the basis for acquired resistance to the 27 Chen JS, Chao Y, Bang YJ, Roca E, Chung HC, Palazzo F et al. A phase I/II and pemetrexed antifolate (LY231514) in an L1210 murine leukemia cell line. Biochem pharmacogenomic study of pemetrexed and cisplatin in patients with unresect- Pharmacol 2003; 65: 1163–1170. able, advanced gastric carcinoma. Anticancer Drug 2010; 21: 777–784. 8 Zhao R, Zhang S, Hanscom M, Chattopadhyay S, Goldman ID. Loss of reduced 28 Jung M, Lee CH, Park HS, Lee JH, Kang YA, Kim SK et al. Pharmacogenomic folate carrier function and folate depletion result in enhanced pemetrexed inhi- assessment of outcomes of pemetrexed-treated patients with adenocarcinoma of bition of purine synthesis. Clin Cancer Res 2005; 11: 1294–1301. the lung. Yonsei Med J 2013; 54: 854–864. 9 Adjei AA, Mandrekar SJ, Dy GK, Molina JR, Gandara DR, Ziegler KL et al. Phase II 29 Cheng Q, Wu B, Kager L, Panetta JC, Zheng J, Pui CH et al. A substrate specific trial of pemetrexed plus bevacizumab for second-line therapy of patients with functional polymorphism of human gamma-glutamyl hydrolase alters catalytic advanced non-small-cell lung cancer: NCCTG and SWOG study N0426. J Clin Oncol activity and methotrexate polyglutamate accumulation in acute lymphoblastic 2010; 28: 614–619. leukaemia cells. Pharmacogenetics 2004; 14: 557–567. 10 Adjei AA, Salavaggione OE, Mandrekar SJ, Dy GK, Ziegler KL, Endo C et al. Cor- 30 Al-Shakfa F, Dulucq S, Brukner I, Milacic I, Ansari M, Beaulieu P et al. DNA variants relation between polymorphisms of the reduced folate carrier gene (SLC19A1) in region for noncoding interfering transcript of dihydrofolate reductase gene and survival after pemetrexed-based therapy in non-small cell lung cancer: a and outcome in childhood acute lymphoblastic leukemia. Clin Cancer Res 2009; North Central cancer treatment group-based exploratory study. J Thorac Oncol 15: 6931–6938. 2010; 5: 1346–1353. 31 Chattopadhyay S, Moran RG, Goldman ID. Pemetrexed: biochemical and cellular 11 Li WJ, Jiang H, Fang XJ, Ye HL, Liu MH, Liu YW et al. Polymorphisms in thymidylate pharmacology, mechanisms, and clinical applications. Mol Cancer Ther 2007; 6: synthase and reduced folate carrier (SLC19A1) genes predict survival outcome in 404–417. advanced non-small cell lung cancer patients treated with pemetrexed-based 32 Christoph DC, Asuncion BR, Hassan B, Tran C, Maltzman JD, O’Shannessy DJ et al. chemotherapy. Oncol Lett 2013; 5: 1165–1170. Significance of folate receptor alpha and thymidylate synthase protein expression 12 Qiu A, Jansen M, Sakaris A, Min SH, Chattopadhyay S, Tsai E et al. Identification of in patients with non-small-cell lung cancer treated with pemetrexed. J Thorac an intestinal folate transporter and the molecular basis for hereditary folate Oncol 2013; 8: 19–30. malabsorption. Cell 2006; 127: 917–928. 33 Shen F, Ross JF, Wang X, Ratnam M. Identification of a novel folate receptor, a 13 Zhao R, Qiu A, Tsai E, Jansen M, Akabas MH, Goldman ID. The proton-coupled truncated receptor, and receptor type beta in hematopoietic cells: cDNA folate transporter: impact on pemetrexed transport and on antifolates activities cloning, expression, immunoreactivity, and tissue specificity. Biochemistry 1994; compared with the reduced folate carrier. Mol Pharmacol 2008; 74: 854–862. 33: 1209–1215.

& 2014 Macmillan Publishers Limited The Pharmacogenomics Journal (2014), 411 – 417