Can pharmacogenetics explain efficacy and safety of cisplatin pharmacotherapy?

Several recent pharmacogenetic studies have investigated the variability in both outcome and toxicity in cisplatin-based therapies. These studies have focused on the genetic variability of therapeutic targets that could affect cisplatin response and toxicity in diverse type of cancer including lung, gastric, ovarian, testicular, and esophageal cancer. In this review, we seek to update the reader in this area of investigation, focusing primarily on DNA reparation enzymes and cisplatin metabolism through Glutathione S-Transferases (GSTs). Current evidence indicates a potential application of pharmacogenetics in therapeutic schemes in which cisplatin is the cornerstone of these treatments. Therefore, a collaborative effort is required to study these molecular characteristics in order to generate a genetic panel with clinical utility.


INTRODUCTION
Cisplatin is an alkylating agent used to treat several types of cancers that works by causing DNA lesions via the formation of intrastrand and interstrand crosslinks, resulting in the activation of various signal-transduction pathways that block cellular processes, such as replication and transcription. The action of cisplatin is cell cycle-independent, although in some cases, prolonged G2 phase cell-cycle arrest occurs (Siddik, 2003;Kelland, 2007). Cisplatin has a central role in cancer chemotherapy for testicular, ovarian/cervical, head and neck, and nonsmall-cell cancers. The side effects include nephrotoxicity (Wong and Giandomenico, 1999), hematogenesis and neurotoxicity (Decatris et al., 2004).
From the beginning, cisplatin has presented variations in therapeutic response. While some tumors are hypersensitive to anticancer therapy, other tumors have an intrinsic resistance. Investigations have sought an explanation of this variation and have suggested that the major resistance mechanisms include reduction in drug levels that reach the target DNA due to reduced uptake and/or increased efflux; increased cellular thiol levels; enhanced DNA repair and/or increased damage tolerance; and failure of cell-death pathways after the formation of platinum-DNA adducts (Fojo, 2001;Siddik, 2003;Wang and Lippard, 2005). In each of these processes there exist potential sites of pharmacogenetics variability (Figure 1). Changes at the genetic level causing modifications in cellular phenotype could explain some of the variability in response and toxicity to cisplatin-included chemotherapy. In this review, we discuss associations between genetic variants in the germ line and in outcomes following cisplatin-based chemotherapy. We mainly focus on DNA repair and cisplatin detoxification through Glutathione S-Transferases (GSTs).
Cisplatin distorts the structure of the DNA that generate intrastrand 1, 2-crosslinks binding proteins into shallow minor groove [high-mobility group (HMG) box proteins, repair proteins, transcription factors, histone H1] (Kartalou and Essigmann, 2001;Wozniak and Blasiak, 2002;Zdraveski et al., 2002). It covalently binds DNA and forms DNA adducts through intra-and interstrand crosslinks (ICLs). Intrastrand crosslinks are repaired by nucleotide excision repair (NER) using the other strand as a template. As both strands are compromised in ICLs, other enzymes are involved in their repair. Two major pathways of ICL repair exist; one is replication dependent and mainly involves homologous recombination, the second is replication independent and involves NER (Ho and Schärer, 2010). At the start of both of these pathways, translesion (TLS) polymerases are needed to bypass ICLs and restore one of the two DNA strands. Translesion synthesis is a mechanism used by cells to prevent common DNA damage from stalling replication forks and rising apoptosis levels. The most important TLS polymerases are Pol ζ (Polymerase zeta) and REV1 (Reversionless 1). Studies have shown that disruption or suppression of expression of both REV3L, the gene encoding the catalytic subunit of Pol ζ, or REV1 modifies sensitivity to cisplatin (Lin et al., 2006;Doles et al., 2010). Goricar et al. (2014) recently determined in patients with malignant mesothelioma that the mutant allele in REV1 rs3087403 and REV1 TGT haplotype associated with increased risk for leukopenia and neutropenia. REV3L rs465646, rs462779, and REV3L CCGG haplotype associated with longer overall survival (Goricar et al., 2014).

DNA REPAIR ENZYMES
DNA damage repair mechanisms are as follows: direct repair of alkyl adducts; repair of base damage and single strand breaks by base excision repair; repair of double strand breaks by homologous recombination or by non-homologous end joining; repair of bulky DNA adducts by NER; and repair of mismatches and insertion/deletion loops by DNA mismatch repair (Camps et al., 2007). The NER pathway is one of the major DNA repair systems involved in the removal of platinum adducts. This pathway involves many proteins in lesion recognition, excision, DNA synthesis and ligation. Excision repair cross-complementary 1 (ERCC1) is a key protein involved in the process of NER and ERCC1-xeroderma pigmentosum (ERCC1-XPF) catalyzes incision on the incision 50 side to the site of DNA damage (Parker et al., 1991;Bessho, 1995). In addition to ERCC1, xeroderma pigmentosum complementary group D (XPD) encodes a helicase that participates in both NER and basal transcription as part of the transcription factor, IIH. Mutations destroying the enzymatic function of XPD protein are manifested clinically in combinations of three severe syndromes, including xeroderma pigmentosum, XP combined with Cockayne Syndrome and trichothiodystrophy (Lehmann, 2001;Clarkson and Wood, 2005). ERCC1 and ERCC2 (XPD) have pivotal roles in the NER pathway, this has been evidenced in studies where lower levels of intratumoral ERCC1 mRNA are significantly correlated with improved survival due to enhanced tumor cell sensitivity to cisplatin (Shirota et al., 2001). mRNA levels as well as the overexpression of ERCC1 and other enzymes have been implicated in the development of clinical resistance to platinum (Kirschner and Melton, 2010;Cheng et al., 2012).
Among these genes, the most studied is ERCC1 gene, mostly focused on the therapy of non-small cell lung cancer (NSCLC) and esophageal cancer. Polymorphisms in ERCC1 include mainly rs3212986 and rs11615. The polymorphism rs3212986 is located in the 3 untranslated region and therefore may affect mRNA stability resulting in a decreased expression levels . In relation to rs3212986, the C allele leads to a change that results in an increase in overall survival Krivak et al., 2008;Takenaka et al., 2010), progression free survival (Krivak et al., 2008;Kim et al., 2009;Erčulj et al., 2012;Chen et al., 2013), treatment response (Li et al., 2010) and prognosis (Takenaka et al., 2010;Okuda et al., 2011). However, opposite associations have been reported in other studies related to reduced responses with the C allele (Bradbury et al., 2009;Kalikaki et al., 2009;Park et al., 2011;Wang et al., 2011), as well as increased toxicity (Khrunin et al., 2010;Tzvetkov et al., 2011;Erčulj et al., 2012). Wang et al. (2011) andBradbury et al. (2009) showed that in esophageal cancer, patients with A/A or A/C genotype had improved outcomes compared with patients carrying wildtype genotypes. In addition, Park et al. (2011) have found similar results in metastatic cancer patients. On the contrary, opposite results have been found in NSCLC and ovarian cancer where the C allele relates to improved survival and treatment response. The variability in outcomes amongst these studies could be due to tumor characteristics (tissue-specific or organ-specific). The polymorphism C→T at codon 118 located on exon 4 of ERCC1 gene (rs11615) is expected to have the same effect. This polymorphism is associated with clinical response to platinum-based chemotherapy in NSCLC. The C allele is also related to an increase in overall survival (Isla et al., 2004;Ryu et al., 2004;Cheng et al., 2012;Joerger et al., 2012), progression free survival (Ryu et al., 2004;Cheng et al., 2012;Joerger et al., 2012), improved treatment response (Kalikaki et al., 2009) and prognosis (Okuda et al., 2011). Nevertheless, others authors detect opposite associations in larger-population studies, including amongst Chinese patients (Li et al., 2010;Ren et al., 2012): this should be considered in future research. Nephrotoxicity has been related to the C allele in rs3212986 ERCC1 (Tzvetkov et al., 2011), T allele in rs11615 ERCC1 (Tzvetkov et al., 2011) and C/T genotype in rs3212986 ERCC1 (Khrunin et al., 2010), independent of cancer type.
Another widely studied gene is ERCC2 (XPD). The presence of a variation in ERCC2 gene (rs13181 and rs1799793) reduces repair capacity, and results in greater efficacy of cisplatin treatment due to increased DNA damage and an enhanced cytotoxic effect. rs1799793 generates a positive effect in overall survival and progression free survival Bradbury et al., 2009;Biason et al., 2012). Erčulj et al. (2012) found that G/G genotype is related to an increase in various types of toxicity (Erčulj et al., 2012) while nephrotoxicity has been shown by Joerger et al. (2012) (Joerger et al., 2012). The A allele in the mutation rs13181 increases overall survival (Park et al., 2001;Quintela-Fandino et al., 2006;Caronia et al., 2009;Chew et al., 2009). However, other authors have found the C allele related to increased overall survival (Bradbury et al., 2009) in esophageal cancer and progression free survival in pancreatic cancer (Avan et al., 2013). These discrepancies suggest that associations with C allele are not fully clear in these types of cancers, and that patients factors, treatment modalities and ethnic population could influence the outcome. Nonetheless, the majority of the results support an association between both rs1799793 and rs13181 and clinical outcomes in patients with NSCLC, osteosarcoma, breast cancer, ovarian cancer, and colorectal cancer. These significant associations in ERCC2 polymorphisms and clinical outcomes have included studies with a larger number of patients and differing patient populations.
Other studies found associations between ERCC5 mutations (rs1047768 and rs751402), PFS (progression free survival) (Sun et al., 2013) and OS (overall survival) (He et al., 2013). These studies have indicated that ERCC5 polymorphisms are involved in the efficacy of cisplatin neoadjuvant chemotherapy. Also, ototoxicity has related to rs2228001 mutation in the Xeroderma Pigmentosum Complementation group C (XPC) gene (Caronia et al., 2009). More information is needed about these associations to reach more powerful conclusions, including a greater number of patients and amongst different ethnic populations.
Additional DNA repair genes have also shown variability, including X-ray repair cross-complementing group 1 (XRCC1). This protein is involved in base excision repair. Among the mutations, we highlight rs25487 and rs1799782 mutations. In relation to rs25487, the mutant G variant has been associated with decreased treatment response Giachino et al., 2007;Pacetti et al., 2009;Khrunin et al., 2010;Joerger et al., 2012;Ke et al., 2012;Miao et al., 2012), although opposite results exist (Quintela-Fandino et al., 2006;Sakano et al., 2006). Other evidence indicates associations between the G allele and neutropenia (Khrunin et al., 2010). T allele in rs1799782 mutation is related with an increase (Miao et al., 2012; and decrease in overall survival (Li et al., 2006;Shim et al., 2010).  and Miao et al. (2012) have performed studies in ovarian cancer with a large number of patients. Further data are required to confirm this association. Another finding is the relation between treatment response and the T allele (Wang et al., 2004;Yuan et al., 2006;Kim et al., 2009;Ke et al., 2012). This discrepancy may be due to cancer type or combined therapies. DNA repair enzymes might decrease the synergistic effects of combination of cisplatin and radiation and information from population should be added in future association specifics to subgroups . In addition, some studies have used cisplatin in combination with paclitaxel, gemcitabine, cyclophosphamide or 5-FU, depending on cancer type. Others factors that might affect variability in different populations are the stage of disease, patient status and period of follow-up in survival analysis.
With respect to X-ray repair cross complementing protein 3 (XRCC3), a protein involved in DNA double-strand breaks, the rs861539 mutation is the only one that relates to treatment outcome. Increased overall survival was associated with the T allele (De las Peñas et al., 2006; as was progression free survival (Font et al., 2008). However, Ren et al. (2012) have shown inverse results  including a large number of patient (n = 340) with NSCLC. More data are necessary to confirm these opposing results.
In summary, studies of association between genetic variants in the DNA repair system and clinical results show that these variants can be potential biomarkers for outcomes in the cisplatinbased therapies ( Table 1). Despite race and treatment regimen, associations testing the polymorphism in ERCC1 appear to follow a consistent direction. rs3212986 and rs11615 polymorphisms should be considered in a future genetic panel because results were obtained in several researches with different treatment and demographic characteristics. Additional research should be performed in order to replicate results found with polymorphisms in ERCC2, XRCC1, and XRCC3. In additional studies, the later polymorphism should be used to evaluate clinical outcomes (overall survival and disease progression) considering different subgroups of patient. In relation to specific toxicities, associations with nephrotoxicity have been described and characterized, but likewise require confirmation.

PHARMACOKINETIC MECHANISMS
Evidence indicates that reduced drug accumulation is a significant mechanism of cisplatin resistance (Kelland, 1993). The cause may be an inhibition in drug uptake, an increase in drug efflux, or both. Studies concerning the mechanisms of cisplatin uptake into the cell have focused on both passive diffusion (Hromas et al., 1987;Binks and Dobrota, 1990;Mann et al., 1991) and copper transporters (Katano et al., 2002;Ohashi et al., 2003;Safaei et al., 2004).
Recent studies have demonstrated that mutation or deletion of the CTR1 gene results in increased cisplatin resistance and reduction of platinum levels (Ishida et al., 2002). Copper-transporting P-type adenosine triphosphate (ATP7B) is associated with cisplatin resistance in vitro (Komatsu et al., 2000), and in various cancers (Nakayama et al., 2002(Nakayama et al., , 2004Ohbu et al., 2003). ATPbinding cassette sub-family C2 (ABCC2), another transporter protein, also has a role in cisplatin resistance, probably promoting drug efflux (Koike et al., 1997;Kool et al., 1997;Cui et al., 1999). ABCC3 is a member of the multidrug resistance protein (MRP) family. Caronia et al. (2011) found that rs4148416 was associated with low survival. In addition, the ABCB1 gene that is well-known and encodes P-glycoprotein, contains three polymorphisms (rs2032582, rs1045642, and rs1128503) that have been studied individually and as a haplotype, however, the results have been inconsistent (Caronia et al., 2011).

DETOXIFICATION
Cisplatin is inactivated by conjugation with glutathione through the GSTs. This phase II enzyme catalyzes the conjugation of reactive metabolites with negatively charged hydrophilic molecules for disposal in excretion processes. Genetic variations in GSTs have been implicated in cellular resistance to cancer chemotherapy and in outcomes of cisplatin-based treatments. When GSTs enzymes with reduced activity are present, the available concentration in the drug in tumor tissue increases. In these patients therapy might be more effective, but might also be severely toxic (Strange et al., 2000;Siddik, 2003;Quiñones et al., 2006). Several studies have shown significant association between polymorphic GSTs genes and cisplatin treatment response suggesting these polymorphisms as potential biomarkers ( Table 2).
In the GSTs superfamily there are eight cytosolic classes (Alpha, kappa, mu, omega, pi, sigma, theta, and zeta) (Katoh et al., 2008;Luo et al., 2011). GSTP1, GSTM1, and GSTT1 genes, have been the most widely studied in relation to the functional polymorphisms. GSTP1 is widely expressed in normal human epithelial tissues. A single nucleotide substitution (A→G) at position 313 (rs1695) of the GSTP1 gene, results in replacement of isoleucine with valine at codon 105 of the enzyme, substantially diminishes GSTP1 enzyme activity. On the contrary, GSTM1 and GSTP1 genetically delected (homozygous null allele) will lead to an absence of enzymatic activity (Stoehlmacher et al., 2002).
The GSTP1 gene has been the most studied in a wide number of cancers with controversial results related to cisplatinbased therapy. Some investigations have shown that patients with G/G genotype present less toxicity (Oldenburg et al., 2007a,b;Goekkurt et al., 2009;Kim et al., 2009) with more survival (Goekkurt et al., 2006;Ruzzo et al., 2006;Ji et al., 2013) and better therapy response Yang et al., 2012). On the other hand, the G allele has been associated with a risk of myelosuppression, polyneuropathy, and toxicity (Yokomizo et al., 2007;Joerger et al., 2012;Windsor et al., 2012;Rednam et al., 2013). In ovarian cancer, the A allele is related to better PFS and OS (Khrunin et al., 2010). GSTP1 A/A genotype has been found to predict suboptimal response to flurouracil/cisplatin chemotherapy and poor survival in patients with advanced gastric cancer (Ruzzo et al., 2006). The influence of rs1695 GSTP1 on toxicity to taxane-and platinum-based chemotherapy is in debate (Kim et al., 2009).
Polymorphism of GSTM1 and GSTT1 genes is associated with cisplatin-based treatments. GSTM1 null has been specifically related to an increase of OS and PFS (Medeiros et al., 2003;Petros et al., 2005;Beeghly et al., 2006;Ott et al., 2008). Concerning toxicity, it has been associated with a decrease in toxicity (Oldenburg et al., 2007a,b;Khrunin et al., 2010), although Dhawan et al. (2013) showed the opposite but with a small sample (n = 23) (Dhawan et al., 2013). On the GSTT1 gene, the non-null allele relates to an increase in overall survival and progression free survival (Goekkurt et al., 2009), however, Kim et al. (2009) showed the opposite but this contradiction apparently is caused by different definitions of patient response. Moreover, the null allele has also associated with an increase in ototoxicity (Jurajda et al., 2012;Choeyprasert et al., 2013). Finally, additional studies examining the GSTA1 gene showed the T/T genotype (rs3957357) associates with an increase of overall survival (Khrunin et al., 2010). Regarding to GSTM3 gene, the AGG/AGG haplotype (rs1799735) is related to less thrombocytopenia, anemia and neuropathy (Khrunin et al., 2010). Nevertheless, more evidence is needed in order to determine a clear role of GSTA1 and GSTM3 genes on cisplatin-based therapy.
Polymorphisms in the GSTP1 gene have shown controversial results among different types of cancer. Some studies found the polymorphic allele related to less toxicity, better therapy response and more survival but others found the opposite regarding to toxicity (Rednam et al., 2013). The results obtained by several authors demonstrate that the GSTM1 null allele is consistently   related to overall survival in different types of cancer. Concerning toxicity, few investigation have found associations, therefore the role of this polymorphism on toxicity is not clear. On the other hand, the GSTT1 null allele associates with toxicity in patients carrying this polymorphism. Regarding OS and PFS it appears that null allele is related to decreased OS and PFS, although one author showed the opposite (Ruzzo et al., 2006;Goekkurt et al., 2009). This contradiction apparently is caused by different definitions of patient response. Together, the evidence appears to indicate a strong association between GSTs polymorphisms and clinical response (overall survival and disease progression). However, the effects on toxicity do not appear to have a clear and dominant trend, and may be related to differing treatment modalities in each of the studies. Despite this, with the data presented we can conclude that the GSTP1 polymorphic allele and the GSTM1 and GSTT1 null alleles appear to result in enhanced overall survival and progression free survival, particularly in gastric cancer where the data have been more consistent. Lack of activity in GSTs enzymes appear to lead to a better treatment response.

CONCLUSION
Personalized therapy promises improved outcomes to treatment with respect to efficacy and toxicity of treatment. Ideally, sub-groups of patients that would require adjustment to therapy based on genetic information could be detected prior to commencing treatment, and therapy accordingly optimized. Pharmacogenetics, the study of the role of inheritance in individual variation in drug response, can address cisplatin cellular resistance, providing tools to achieve the modification of current treatments in different types of cancer, including lung, gastric, ovarian, testicular and, esophageal cancers (Weinshilboum, 2003).
Variable responses to different treatments, including cisplatin, have been seen from different points of view. When looking into the genetic variability in processes where cisplatin is involved, including pharmacokinetics and pharmacodynamics, efforts have delivered evidence regarding DNA repair systems and metabolization systems. Within the variability in DNA repair processes, key genes involved include ERCC1, ERCC2 (XPD), ERCC5, XRCC1, XRCC3, and XPC genes. Studies examining the genetic variability of cisplatin metabolism have shown that the main genes involved are GSTP1, GSTM3, GSTM1, and GSTT1. Currently there appears to be a group of genes that would influence variability in response and toxicity in cisplatin-based therapies which we present here in this up-dated review.
Diverse results have been found among the polymorphisms analyzed in both DNA repair enzymes and detoxification enzymes. These contradictions and variations are primarily due to the heterogeneity amongst studies (patient population, treatment and number of subjects). Another possibility is with the inclusion of a large number of candidate genes, there is always a risk of false positive associations. For example, recent studies showed a relationship between rs12201199 in thiopurine S-methyltransferase gene (TPMT) and rs9332377 in the catechol-O-methyltransferase gene (COMPT) with cisplatininduced hearing loss in children (Ross et al., 2009). Our opinion is that future studies in this line should include the genes we have highlighted, and that a collaborative effort is required to improve the quality and strength of evidence in order to achieve a validated panel of polymorphisms that guides therapeutic decisions.
Finally, prospective clinical studies employing polymorphism panels in these treatment procedures are required to determine whether adjustment of therapy based on genetic information can influence outcomes in these scenarios.

AUTHOR CONTRIBUTIONS
Ángela Roco: Review of intellectual content and Final approval, Juan Cayún: Substantial contributions, Stephania Contreras: Substantial contributions, Jana Stojanova: Substantial contributions, Luis Quiñones: Review of intellectual content and Final approval.

ACKNOWLEDGMENTS
The work in the author's laboratory has been financed by Grants FONDECYT 1140434, Chile.