Abstract Supported by a PHRC grant (#09-18-005) Background: Recent literature has suggested that germline genetic variants of drug-metabolizing enzymes or CYP19A1 (coding for aromatase) may be involved in the systemic aromatase inhibitors (AI) concentrations or the occurrence of side effects (Hertz et al. Pharmacogenomics 2017). A prospective multicentre 3-year follow-up study was carried out to investigate the relationships between pharmacogenetics (PG), pharmacokinetics (PK) and toxicity in breast cancer patients treated with adjuvant AI (n=1098) or tamoxifen (n=879). The clinical results and the tamoxifen PG/PK analyses are described elsewhere (abstracts #851544 and #850248). Methods: SNP genotyping of 95 SNPs was performed on the Biomark (Fluidigm) with Taqman assays and was available for 373, 515 and 151 patients treated with anastrozole (ANA), letrozole (LETRO) and exemestane (EXE) respectively. CYP2A6 metaboliser status (MS) (poor, intermediate or normal) was determined based on alleles function (*1, *9, *2) and number of CYP2A6 copies. Trough plasma concentrations of each drug were determined 6 months after the start of the study by UPLC-MS/MS and were available for 342, 463 and 130 patients of the ANA, LETRO and EXE arms. Patients with AI concentrations below the limit of quantification were excluded for non-compliance (9 patients for ANA, 8 patients for LETRO and 7 patients for EXE). Toxicity was measured as a binary outcome (occurrence or worsening of hot flushes, fatigue, pain, arthralgia, vaginal dryness). All genetic associations were adjusted for multiple testing. Results: ANA concentration was significantly higher in patients experiencing pain (p=0.025) and was associated with rs28365063 (UGT2B7 g.372A>G). LETRO concentrations were strongly associated with CYP2A6 metabolizer status (p=0.0001) but did not differ in patients with or without toxicity. In the EXE arm, patients with hot flushes or arthralgia had a significantly lower level of exemestane (p= 0.0002 and p=0.023 respectively) but since the metabolism of EXE leads to active 17-hydroexemestane, we can hypothesize that the lower EXE concentration is an indirect reflection of the metabolite formation. A SNP (rs2307424) in NR1I3 gene (coding for the constitutive androstane receptor CAR) was associated with EXE concentrations. CAR has been shown to regulate CYP2B6, which is involved in the formation of 6-hydroxy-methyl-exemestane (inactive metabolite). Regarding the relationships between PG and toxicity, in the ANA arm, 3 SNPs of CYP19A1 gene tended to be associated with hot flushes worsening (rs934635) and arthralgia (rs10046 and rs2304463) but did not remain significant after multiple tests correction. In the EXE arm, several SNPs in NR1I3 gene were associated with fatigue. In the LETRO arm, patients with a poor CYP2A6 MS had a higher risk of experiencing depression. Conclusions: Our study confirms the predominant role of CYP2A6 in LETRO PK. To our knowledge, this is the first study to report on the role of UGT2B7 rs28365063 in ANA and NR1I3 in EXE PK and side effects. These relationships need to be re-evaluated with the drug concentrations obtained during the 3-year follow-up. Citation Format: Thomas F, Marquet P, Pinguet F, White-Koning M, Robert J, Tafzi N, Solassol I, Despax R, Levasseur N, Ellis S, Massoubre A, Mbatchi L, Le Morvan V, Roché H, Chatelut E, Evrard A. Pharmacogenetic determinants of aromatase inhibitors pharmacokinetics and side effects: 6-month results of the adjuvant breast cancer longitudinal PHACS study (NCT01127295) [abstract]. In: Proceedings of the 2017 San Antonio Breast Cancer Symposium; 2017 Dec 5-9; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2018;78(4 Suppl):Abstract nr P3-12-07.

Abstract Supported by a PHRC grant (#09-18-005) Background: The role of CYP2D6 genetic polymorphisms and plasma levels of active metabolites of tamoxifen (TAM) on clinical response and occurrence of side effects remains controversial. We conducted a prospective, adjuvant, multicentre, 3-year follow-up study of breast cancer patients in order to evaluate the relationships between pharmacogenetics, pharmacokinetics and toxicity of TAM and its metabolites (n=879) or aromatase inhibitors (AI, n=1098). The present report focuses on the evaluation at 6 months after inclusion of 864 patients treated with 20 mg/day TAM.The clinical results and the AI PG/PK analyses are described elsewhere (abstracts #851544 and #851525). Methods: Residual plasma concentrations for tamoxifen and its 6 major metabolites (endoxifen ENDO, 4-hydroxy-tamoxifen 4-OH-TAM, N-desmethyl TAM, TAM-N-oxyde, 4'-OH-TAM and Z'ENDO) at 6 months after start of treatment were measured by UPLC-MS/MS in 789 patients. Nine patients with TAM concentrations below the limit of quantification were excluded for non-compliance. SNP genotyping of 95 selected SNPs was performed on the Biomark (Fluidigm) in a microfluidic multiplex 96 dynamic array chip with Taqman assays and was available for 857 patients. Patients were classified according to their CYP2D6 metaboliser status (MS) (PM, IM, EM and UM) based on presence of functional, decreased function or no functional alleles (*4, *6, *7, *9, *10, *17, *41) and number of CYP2D6 copies (*5 or duplication). Metabolic ratios (MR) were calculated for TAM/4-OH-TAM, TAM/N-desmethyl tamoxifen (NDT), NDT/ENDO and 4-OH-TAM/ENDO. Anti-estrogenic activity score (AAS) was calculated according to a recently proposed algorithm (De Vries Schultink et al.,Breast Cancer Res Treat. 2017).Toxicity was measured as a binary outcome (first occurrence or worsening of hot flushes, fatigue, depression, pain, arthralgia, vaginal dryness). All genetic associations were adjusted for multiple testing. Results: ENDO concentration and AAS increased significantly with CYP2D6 MS (p<0.001). The presence of a CYP3A4*22 allele was significantly associated with endoxifen concentrations; this association remained significant after adjusting for CYP2D6 MS. TAM/4-OH-TAM MR was significantly influenced by the presence of CYP3A4*22, CYP2C19*2 and *17, and CYP2D6 status. The percentage of patients having an AAS>=1798 (i.e., threshold previously associated with recurrence-free survival RFS by De Vries et al. 2017) was 6%, 50%, 84% and 91% of patients respectively classified as PM, IM, EM and UM. Side effects were not significantly associated with higher levels of TAM metabolites concentrations. After correction for multiple testing, SNPs or CYP2D6 MS were not significantly associated with occurrence or worsening of adverse events, premature treatment discontinuations or switch due to toxicity within the first 6 months. Conclusions: In this large prospective study, we quantified the impact of PG on TAM PK and AAS, previously shown to predict RFS. Although the toxicity observed after 6 months of TAM does not seem correlated with PK or PG, these relationships need to be re-evaluated during the 3-year follow-up. Citation Format: White-Koning M, Arellano C, Le Morvan V, Evrard A, Puzskiel A, Vachoux C, Dauba J, Houyau P, Poublanc M, Robert J, Boyer J-C, Roché H, Thomas F, Chatelut E. Impact of genetic polymorphisms on plasma levels of tamoxifen and its metabolites and toxicity: 6-months results of the adjuvant breast cancer longitudinal PHACS study (NCT01127295) [abstract]. In: Proceedings of the 2017 San Antonio Breast Cancer Symposium; 2017 Dec 5-9; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2018;78(4 Suppl):Abstract nr P3-12-03.

Background: This study aimed to determine whether the BRCA1 haplotype was associated with trabectedin efficacy in soft-tissue sarcoma (STS) patients.Methods: We analysed BRCA1 single-nucleotide polymorphisms (SNPs) in tumour specimens from 135 advanced STS patients enrolled in published phase 2 trials or in a compassionate-use programme of trabectedin. Forty-four advanced STS patients treated with doxorubicin and 85 patients with localised STS served as controls. The 6-month nonprogression rate and overall survival (OS) were analysed according to BRCA1 haplotype using log-rank tests.Results: A favourable BRCA1 haplotype (presence of at least one AAAG allele) was significantly associated with an improved 6-month nonprogression rate. It was the only variable significantly associated with OS. No correlations were found between outcomes for patients with localised or advanced STS treated with doxorubicin.Conclusions: The BRCA1 haplotype represents a potential DNA repair biomarker that can be used for the prediction of response to trabectedin in STS patients.

Irinotecan is a cytotoxic agent administered by IV infusion in the treatment of advanced colorectal cancer. Its anticancer activity results from its bioactivation into SN-38 metabolite, which is cleared through glucuronidation by the hepatic enzyme uridine diphosphate-glucuronosyltransferase 1A1 (UGT1A1). In the general population, there is wide inter-subject variability in UGT1A1 enzyme activity related to UGT1A1 gene polymorphisms. The French joint workgroup coming from the National Pharmacogenetic Network (RNPGx) and the Group of Clinical Oncologic Pharmacology (GPCO) herein presents an updated review dealing with efficacy and toxicity clinical studies related to UGT1A1 genetic variants. From a critical analysis of this review it clearly emerges that, for doses higher than 180 mg/m(2), hematologic and digestive irinotecan-induced toxicities could be prevented in daily clinical practice by generalizing the use of a simple pharmacogenetic test before starting treatment. The clinical relevance of this test is also discussed in terms of treatment efficacy improvement, with the possibility of increasing the irinotecan dose in patients not bearing the deleterious allele. This test involves using a blood sample to analyze the promoter region of the UGT1A1 gene (UGT1A1*28 allele). Best execution practices, laboratory costs, as well as results interpretation are described with the aim of facilitating the implementation of this analysis in clinical routine. The existence of a French laboratories network performing this test in clinical routine makes it possible to generalize UGT1A1 deficiency screening, so as to guarantee equal access to safe treatment and optimized irinorecan-based therapy for the many patients receiving irinotecan-based therapy in advanced colorectal cancer.

Using cell line panels we identified associations between single-nucleotide polymorphisms (SNPs) and chemosensitivity. To validate these findings in clinics, we genotyped a subset of patients included in a neoadjuvant breast cancer trial to explore the relationship between genotypes and clinical outcome according to treatment received and p53 status. We genotyped 384 selected SNPs in the germline DNA extracted from formalin-fixed paraffin-embedded non-invaded lymph nodes of 243 patients. The polymorphisms of five selected genes were first studied, and then all 384 SNPs were considered. Correction for multiple testing was applied. CYP1B1 polymorphism was significantly associated with pathological complete response (pCR) in patients who had received DNA-damaging agents. MDM2 , MDM4 and TP53BP1 polymorphisms were significantly associated with pCR in patients harboring a p53-positive tumor. In the complete SNP panel, there was a significant association between overall survival (OS) and a SNP of ADH1C , R272Q ( P =0.0023). By multivariate analysis, only ADH1C genotype and p53 status were significantly associated with OS.

Background: Gene polymorphisms may be associated to response to chemotherapy. Using cell line panels we identified single nucleotide polymorphisms (SNPs) in MDM2 and CYP1B1 genes associated with chemosensitivity; for MDM2 (T309G) the association was strictly dependent upon the p53 status (Br J Cancer 2009; 101: 350-6); for CYP1B1 (V432L), the association was only observed for DNA-damaging agents (Mol Cancer Ther 2010; 9: 3315-21). To validate these findings in the clinical setting, we genotyped a subset of patients of the EORTC10994 neoadjuvant trial (Lancet Oncol 2011; 12: 527-39) to explore the relationship between these SNPs and the pathological response to chemotherapy (pCR) as a function of treatment received and TP53 status as assessed by a functional assay. Methods: We genotyped uninvolved lymph nodes removed at definitive surgery in 243 patients, using a customised Illumina platform for the assessment of 384 selected SNPs. A stepwise analysis was conducted to preserve statistical power: first, only five selected polymorphisms, including MDM2 T309G and CYP1B1 V432L, were tentatively associated with pCR, PFS and OS. In the second step, all SNPs present on the chip were considered for possible associations with clinical outcome. At both steps, a Benjamini-Hochberg correction for multiple testing was applied and only corrected p values are shown below. Since the rate of pCR could vary independently by treatment arm and p53 status, interaction tests were performed by logistic regression. Results: The CYP1B1 polymorphism interacted only on pCR with treatment received (p = 0.032), the MDM2 one only with p53 status (p = 0.031). No significant relationship with DFS or OS was found. In contrast, for the complete SNP panel, there was a significant association between OS and a SNP of alcohol dehydrogenase 3, ADH1C R272Q (corrected p value = 0.0023). A multivariate analysis showed that only ADH1C genotype and p53 status were significantly associated with OS (p < 0.001 and p = 0.012 respectively). Conclusion: SNPs in CYP1B1 and MDM2 could be markers of response to chemotherapy in breast cancer while a SNP in ADH1C may have a prognostic interest that needs be further validation. Disclosure: All authors have declared no conflicts of interest.

Development of new anticancer drugs has resulted in improved mortality rates and 5-year survival rates in patients with cancer. However, many of the modern chemotherapies are associated with cardiovascular toxicities that increase cardiovascular risk in cancer patients, including hypertension, thrombosis, heart failure, cardiomyopathy, and arrhythmias. These limitations restrict treatment options and might negatively affect the management of cancer. The cardiotoxic effects of older chemotherapeutic drugs such as alkylating agents, antimetabolites, and anticancer antibiotics have been known for a while. The newer agents, such as the antiangiogenic drugs that inhibit vascular endothelial growth factor signalling are also associated with cardiovascular pathology, especially hypertension, thromboembolism, myocardial infarction, and proteinuria. Exact mechanisms by which vascular endothelial growth factor inhibitors cause these complications are unclear but impaired endothelial function, vascular and renal damage, oxidative stress, and thrombosis might be important. With increasing use of modern chemotherapies and prolonged survival of cancer patients, the incidence of cardiovascular disease in this patient population will continue to increase. Accordingly, careful assessment and management of cardiovascular risk factors in cancer patients by oncologists and cardiologists working together is essential for optimal care so that prolonged cancer survival is not at the expense of increased cardiovascular events.La mise au point de nouveaux médicaments anticancéreux a permis de réduire le taux de mortalité et d'améliorer le taux de survie après 5 ans des patients atteints de cancer. Cependant, nombre de ces nouveaux anticancéreux sont associés à une toxicité cardiovasculaire qui accroît le risque cardiovasculaire de ces patients, notamment en ce qui a trait à l'hypertension, à la thrombose, à l'insuffisance cardiaque, à la cardiomyopathie et à l'arythmie. Cette problématique limite les choix de traitement et peut avoir une incidence négative sur la prise en charge du cancer. La cardiotoxicité des anticancéreux plus anciens comme les agents alkylants, les antimétabolites et les antibiotiques anticancéreux est connue depuis assez longtemps. Les nouveaux agents comme les antiangiogéniques, qui inhibent l'expression de facteurs de croissance endothéliale vasculaire, sont également associés à des pathologies cardiovasculaires, plus particulièrement à l'hypertension, à la thromboembolie, à l'infarctus du myocarde et à la protéinurie. Le mécanisme causal exact des complications liées aux antiangiogéniques demeure encore inexpliqué, mais la dysfonction endothéliale, les dommages vasculaires et rénaux, le stress oxydatif et la thrombose pourraient être des facteurs importants. Le recours de plus en plus fréquent aux nouvelles chimiothérapies et la prolongation de la survie des patients feront encore augmenter l'incidence des maladies cardiovasculaires dans cette population. Les oncologues devront donc travailler de pair avec les cardiologues afin de soigneusement évaluer et prendre en charge les facteurs de risque cardiovasculaire pour assurer les meilleurs soins possibles et ainsi éviter que la prolongation de la survie des patients se fasse au prix d'un nombre accru d'événements cardiovasculaires.

We investigated, in the panel of 60 human tumour cell lines of the National Cancer Institute (NCI-60), whether the R72P polymorphism of TP53 and the T309G polymorphism of MDM2 were associated to the in vitro cytotoxicity of anticancer agents, extracted from the NCI database. For validation, the same study was performed independently on a second panel of tumour cell lines, JFCR-45. Both SNPs were identified in cell DNA using PCR-RFLP techniques confirmed by direct sequencing and by pyrosequencing. For the analysis of the results, the mutational status of p53 was taken into account. In the NCI-60 panel, the TP53 rare-allele frequency was 32% and the MDM2 rare-allele frequency 39%. The MDM2 alleles were distributed according to Hardy–Weinberg equilibrium whereas this was only found, for the TP53 alleles, in p53 non-mutated cell lines. Comparable results were obtained in the JFCR-45 validation set. The TP53 SNP had low impact on anticancer drug cytotoxicity in either panel. In contrast, the MDM2 gene polymorphism had a major impact on anticancer drug cytotoxicity, essentially in p53 non-mutated cell lines. Presence of the rare allele was associated to significantly higher MDM2 protein expression and to increased sensitivity to DNA-interfering drugs. In the JFCR-45 panel, a similar effect of the MDM2 gene polymorphism was observed, but was less dependent on the p53 mutational status. We hypothesised that cell lines harbouring the MDM2 G allele presented a lower availability of p53 for DNA repair, translating into higher sensitivity to DNA-damaging agents.

4788 ERCC2 (XPD) and ERCC5 (XPG) belong to the nucleotide excision repair (NER) pathway. ERCC2 is part of the TFIIH transcription complex, which is stabilized by ERCC5. Both ERCC2 and ERCC5 are required for the association of the CAK subcomplex (CDK7-Cycline H-MAT1) with the core TFIIH. They bear several common polymorphisms that have been studied for their involvement in cancer risk. Some studies have suggested in addition a potential role of these polymorphisms in the efficiency of cisplatin. We have used the NCI-60 panel of human tumor cell lines as a model for establishing relationships between the presence of the polymorphism in the genome of the cell lines and their responsiveness to a large number of anticancer drugs. Two single-nucleotide polymorphisms (SNP) in ERCC2 (asp312asn and lys751gln) and one SNP in ERCC5 (asp1104his) were determined in DNA extracts of the NCI-60 panel, using PCR-RFLP techniques. These data were compared to the cytotoxicity of 135 core drugs against the panel as well as to ERCC2 and ERCC5 gene expression. Both cytotoxicity and expression data were extracted from the NCI database (http://dtp.nci.nih.gov). The variant homozygous ERCC2 cell lines expressed the ERCC2 gene to a significantly lower level than the common homozygous and heterozygous cell lines, while there was no significant difference in ERCC5 gene expression as a function of the genotype. We have shown that the lys751gln of ERCC2 and the asp1104his of ERCC5 SNPs were significantly associated to the cytotoxicity of taxanes, the variant ERCC2 genotype being associated to higher taxane activity and the variant ERCC5 genotype to lower taxane activity. This prompted us to generate isogenic cell lines differing only by one of these SNPs. We first cloned wild-type and variant cDNAs in a pTracer eukaryotic expression vector; then, transformed Xeroderma pigmentosum patient cell lines with either ERCC2 (XP6BE) or ERCC5 (XPCS1LV) deficiency were transfected to establish stable cell lines. Comparative cytotoxicity evaluation of paclitaxel and cisplatin were undertaken in the isogenic pairs, as well as drug- and UV-induced DNA damage. Since TFIIH is involved not only in DNA repair but also in mRNA transcription and cell cycle progression, we hypothesize that impaired ERCC2 or ERCC5 proteins may cause the liberation of CAK, allowing it to exert its activity on the G2/M transition. Enhancing mitosis entry would then lead to hypersensitivity to spindle poisons, explaining thus the effect of ERCC2 and ERCC5 polymorphisms on taxane sensitivity. The impairment of ERCC2 activity would be due to its lower expression level, the lys751gln polymorphism leading to an important rearrangement of mRNA secondary structure that we have evidenced in silico. The impairment of ERCC5 activity would be due to the fact that the common 1104his allele in Caucasians encodes an unusual aminoacid at this position. These hypotheses are under exploration.

Today's drug therapy regimens rely on the anticipated relation between drug dose, acquired plasma level and desired effect. However, the capacity of individual patients to absorb and metabolise drugs may differ significantly, part of which is due to genetic factors. These genetic factors can be used to predict the drug metabolizing potential of patients before starting therapy. A major challenge is to identify the genetic polymorphisms which are relevant for a particular therapy and to determine the clinical consequences with respect to dosing, or choice of drug, based on the outcome of pharmacogenetic analyses. For the cytochrome P450 system, which catalyzes oxidative reactions, many new genetic polymorphisms have been identified in the last years. In this review, the current knowledge of cytochrome P450 polymorphisms with respect to cancer treatment is described, highlighting the potential of reaching considerable benefit from personalized therapy, either by improving efficacy or reducing the toxicity of current treatment regimens. This review specifically addresses the knowledge today on cytochrome P450 pharmacogenetics for tamoxifen, docetaxel, paclitaxel, cyclophosphamide, ifosfamide, imatinib, gefitinib, irinotecan, etoposide, teniposide, thalidomide and vincristine therapies, discussing its current potential for individualized therapy based on cytochrome P450 genetic polymorphisms.

Le séquençage du génome humain est à l’origine de la découverte de millions de variations de séquence dans le génome humain. Ces variations de séquence d’ADN sont dans leur immense majorité limitées à des polymorphismes de nucléotide unique (Single Nucléotide Polymorphisms ou SNP); cette forme courante de polymorphisme se rencontre environ toutes les 1 000 bases dans le génome humain et 1,8 million de SNP sont actuellement répertoriés. L’accès à ces SNP dans les banques de données publiques a ouvert la possibilité d’étudier l’influence de ces polymorphismes répertoriés sur la prédisposition à certaines maladies génétiques ainsi que sur la réponse aux médicaments. L’identification de ces nombreux SNP tient beaucoup aux progrès des techniques d’analyse de l’ADN et nous décrivons, dans cette revue, différentes méthodes pouvant être utilisées pour l’étude de polymorphismes. Les techniques d’étude des SNP reposent sur deux principes fondamentaux: d’une part la création ou l’élimination par le SNP d’un site de coupure pour une enzyme de restriction donnée, et d’autre part la formation d’un mésappariement due à la présence d’une séquence variante lors d’une hybridation. La technique originelle d’étude des SNP est le Southern blot couplé à une digestion enzymatique permettant d’étudier les polymorphismes de taille de fragments de restriction (RFLP). Ce procédé s’est largement simplifié par l’arrivée de la technique de réaction en chaîne par polymerase (PCR). D’autres approches telles que la chromatographie à haute performance en conditions dénaturantes (dHPLC) ou la PCR en temps réel permettent une discrimination des allèles sauvages et variants. En revanche, le séquençage reste la seule technique permettant de déterminer la nature et la position des SNP connus et inconnus. Enfin, l’apparition des techniques à haut débit permet de prévoir des changements d’échelle considérables dans la recherche des SNP.