Influence of Single-Nucleotide Polymorphisms on Clinical Outcomes of Capecitabine-Based Chemotherapy in Colorectal Cancer Patients: A Systematic Review

Simple Summary Colorectal cancer is one of the most prevalent neoplasms worldwide. Capecitabine is an oral fluoropyrimidine widely used to treat colorectal cancer in early and advanced stages. However, it shows high interindividual variability in its effectiveness and safety. This variability may be due to genetic variants in proteins involved in the pharmacokinetics and pharmacodynamics of the drug. Currently, only four variants of the DPYD gene are clinically relevant for the prediction of severe toxicity, and there are no validated predictive biomarkers of capecitabine effectiveness. Therefore, the search of potential predictive genetic biomarkers to personalize and optimize capecitabine therapy remains necessary. The aim of this study was to systematically review the literature published in the last 10 years on the influence of single-nucleotide polymorphisms in the main genes involved in capecitabine pharmacokinetics and pharmacodynamics on therapy outcomes in patients with colorectal cancer. Abstract The aim of this systematic review was to provide a comprehensive overview of the literature published in the last decade on the association of single-nucleotide polymorphisms in genes involved in the pharmacodynamic and pharmacokinetic pathways of capecitabine with treatment outcomes among colorectal cancer patients. A systematic search of the literature published in the last 10 years was carried out in two databases (Medline and Scopus) using keywords related to the objective. Quality assessment of the studies included was performed using an assessment tool derived from the Strengthening the Reporting of Genetic Association (STREGA) statement. Thirteen studies were included in this systematic review. Genes involved in bioactivation, metabolism, transport, mechanism of action of capecitabine, DNA repair, and folate cycle were associated with toxicity. Meanwhile, genes related to DNA repair were associated with therapy effectiveness. This systematic review reveals that several SNPs other than the four DPYD variants that are screened in clinical practice could have an impact on treatment outcomes. These findings suggest the identification of future predictive biomarkers of effectiveness and toxicity in colorectal cancer patients treated with capecitabine. However, the evidence is sparse and requires further validation.


Introduction
Colorectal cancer (CRC) is one of the most prevalent malignancies worldwide. In the United States, it is estimated that the incidence of CRC in 2022 will be 151,030 new cases [1]. According to data provided by the International Agency for Research on Cancer (IARC), CRC is the second leading cause of cancer death worldwide, accounting for 9.4% of deaths [2].
Capecitabine, an oral fluoropyrimidine (FP), is a prodrug of 5-fluorouracil (5-FU) used to treat CRC in both early and advanced stages [3]. It can be used as monotherapy and in combination with other chemotherapeutic agents, biologics, or radiotherapy (RT) [4,5]. Despite its widespread use, capecitabine therapy shows high interindividual variability in its effectiveness and safety. The observed variability may be due to patients' clinical or demographic factors, including genetic factors [6]. Genetic variants that affect the activity of proteins involved in the pharmacokinetics (PK) and pharmacodynamics (PD) of capecitabine could cause different intensities and durations of the observed response [7].
There are currently no validated predictive biomarkers of the effectiveness of capecitabine treatment, and only four variants of the DPYD gene, which encodes for dihydropyrimidine dehydrogenase (DPD), the rate-limiting enzyme for FP metabolism, are clinically relevant in relation to their influence on the presence of severe toxicity during capecitabine therapy (rs3918290 (DPYD*2A), rs55886062 (DPYD*13), rs67376798, and rs75017182-rs56038477 (HapB3)) [8]. However, the genotyping of these four variants only prevents 20-30% of severe toxicity events associated with FPs. It seems that the remaining 70-80% of toxicities may be due to other genetic variants in DPYD or other genes involved in the PK and PD of capecitabine [9]. Therefore, the study of potential predictive genetic biomarkers of effectiveness and toxicity of capecitabine-based therapy is still necessary.
Several genes are involved in the PK of capecitabine. They can be grouped according to the bioactivation, catabolism, and transport processes of the drug (Figure 1a). The bioactivation of capecitabine comprises three steps. In the liver, capecitabine is catalyzed mainly by the carboxylesterase 1 (CES1) enzyme to form 5 -deoxy-5-fluorocytidine (5 -dFCR). 5 -dFCR is subsequently metabolized by the enzyme cytidine deaminase (CDA) from which 5 -deoxy-5-fluorouridine (5 -dFUR) is obtained. This metabolite is finally catalyzed to the active principle 5-FU by the enzymatic action of thymidine phosphorylase (TP) or uridine phosphorylase (UPP). TP is expressed in both liver and tumor tissue, but its expression in the latter is greater [10,11]. There are many 5-FU catabolism routes, some of which lead to the formation of inactive metabolites and others to the production of metabolites with pharmacological activity. As mentioned previously, the rate-limiting step of 5-FU catabolism is mediated by DPD, an enzyme that transforms it into dihydrofluorouracil (DHFU), which is subsequently metabolized by the dihydropirimidinase (DPYS) and beta-ureidopropionase 1 (UPB1) enzymes to obtain fluoro-beta-alanine (FBAL), a metabolite excreted in the urine. There are three pharmacologically active metabolites of 5-FU: fluorodeoxyuridine monophosphate (FdUMP), fluorouridine triphosphate (FUTP), and fluorodeoxyuridine triphosphate (FdUTP). The conversion of 5-FU into its main active metabolite FdUMP can occur by several routes: (a) direct action of the TP and then thymidine kinase (TK) enzymes, (b) indirect action of the uridine monophosphate synthetase (UMPS) and amidophosphoribosyltransferase (ATase) enzymes, of which the latter is encoded by the phosphoribosyl pyrophosphate amidotransferase (PPAT) gene, or (c) by UPP and uridine-cytidine kinase (UCK) to form the intermediate metabolite fluorouridine diphosphate (FUDP), which is subsequently catalyzed via ribonucleotide reductase (RNR). FUDP can also be converted into FUTP and FdUTP. Transport of 5-FU is mediated by various proteins, including equilibrative nucleoside transporter 1 (ENT1), encoded by solute carrier family 29 member 1 (SLC29A1) gene, solute carrier family 22 member 7 (SLC22A7), ATP-binding cassette subfamily G member 2 (ABCG2), ATP-binding cassette subfamily C member 3 (ABCC3), ATP-binding cassette subfamily C member 4 (ABCC4), and ATP-binding cassette subfamily C member 5 (ABCC5) [12]. The ATP-binding cassette subfamily B member 1 (ABCB1) gene, also known as multidrug resistance 1 (MDR1), codes for P-glycoprotein (P-gp), a carrier protein that transports a great variety of substrates and plays a crucial role in maintaining intracellular levels of numerous antineoplastic agents [13]. Although capecitabine has not been clearly identified as a substrate of P-gp, its expression has been related to resistance to 5-FU in modified cell lines, so its potential influence on the outcomes of therapy cannot be ruled out [14]. sion of 5,10-methylenetetrahydrofolate (5,10-MTHF) to dihydrofolate, a key component of the folate cycle. Furthermore, FUTP and FdUTP metabolites are directly incorporated into RNA and DNA, respectively, causing direct damage to genetic material and consequently cell death. Therefore, as well as TYMS, the genes involved in (a) the folate cycle (methylenetetrahydrofolate reductase (MTHFR), dihydrofolate reductase (DHFR), methylenetetrahydrofolate dehydrogenase (MTHFD1), serine hydroxymethyltransferase 1 (SHMT1), gamma-glutamyl hydrolase (GGH), and folylpolyglutamate synthase (FPGS)), (b) DNA repair (ERCC excision repair 1 and 2 (ERCC1, ERCC2), single-strand-selective monofunctional uracil-DNA glycosylase 1 (SMUG1), thymine DNA glycosylase (TDG), Xray repair cross complementing 1 and 3 (XRCC1, XRCC3), and (c) the cell cycle (tumor protein P53 (TP53)), are of particular interest in the PD of capecitabine [12]. The objective of this study was to systematically review the literature published in the last decade on the influence of single-nucleotide polymorphisms (SNPs) in the main genes involved in the PK and PD of capecitabine on the effectiveness and safety of antineoplastic therapy in patients with CRC. As for the PD of capecitabine (Figure 1b), the FdUMP metabolite inhibits the action of thymidylate synthase (TS), an enzyme encoded by the TYMS gene that is crucial for the synthesis of pyrimidines and DNA. In turn, this action blocks the simultaneous conversion of 5,10-methylenetetrahydrofolate (5,10-MTHF) to dihydrofolate, a key component of the folate cycle. Furthermore, FUTP and FdUTP metabolites are directly incorporated into RNA and DNA, respectively, causing direct damage to genetic material and consequently cell death. Therefore, as well as TYMS, the genes involved in (a) the folate cycle (methylenetetrahydrofolate reductase (MTHFR), dihydrofolate reductase (DHFR), methylenetetrahydrofolate dehydrogenase (MTHFD1), serine hydroxymethyltransferase 1 (SHMT1), gamma-glutamyl hydrolase (GGH), and folylpolyglutamate synthase (FPGS)), (b) DNA repair (ERCC excision repair 1 and 2 (ERCC1, ERCC2), single-strand-selective monofunctional uracil-DNA glycosylase 1 (SMUG1), thymine DNA glycosylase (TDG), X-ray repair cross complementing 1 and 3 (XRCC1, XRCC3), and (c) the cell cycle (tumor protein P53 (TP53)), are of particular interest in the PD of capecitabine [12].
The objective of this study was to systematically review the literature published in the last decade on the influence of single-nucleotide polymorphisms (SNPs) in the main genes involved in the PK and PD of capecitabine on the effectiveness and safety of antineoplastic therapy in patients with CRC.
Duplicate articles were deleted, and the remaining articles were analyzed by title and abstract. Those that met the inclusion criteria were evaluated by reading the full text. Search and study selection was performed independently by two researchers. In the event of discrepancies in the comparison of results, a third researcher was consulted.

Eligibility Criteria
Studies were selected if they complied with the following inclusion criteria: patients treated solely with therapeutic regimens based on capecitabine (as monotherapy or in combination with other antineoplastic agents or RT) for all types of treatment (neoadjuvant, adjuvant, palliative); patients genotyped for SNPs in CES1, CES1P1, CES2, CDA,  DPYD, DPYS, PPAT, RRM2, RRM1, TK1, TYMP, UCK1, UCK2, UMPS, UPP1, UPP2, UPB1,  ABCB1, ABCC3, ABCC4, ABCC5, ABCG2, ABCB1, SLC22A7, SLC29A1, TYMS, ENOSF1,  MTHFR, DHFR, MTHFD1, SHMT1, GGH, FPGS, ERCC2, ERCC1, SMUG1, TDG, XRCC3, XRCC1, or TP53 genes; evaluation of therapy effectiveness using RECIST criteria or Dworak classification (for tumor response) or survival and toxicity analysis (using the Common Terminology Criteria for Adverse Events (CTCAE)); full-text availability; original design type: randomized clinical trial, non-randomized clinical trial or experimental study, cohort study (prospective or retrospective), or case-control study; published in the last 10 years in English, Spanish, Portuguese, or German. If a cohort (or part of a cohort) was described in more than one study, the most recent or extensive study was selected. Conversely, studies that presented the same cohort (or part of the same cohort) but evaluated different genes of interest were included. Exclusion criteria were extraction of genetic material from tumor tissue and studies based solely on haplotypes, mutation studies, gene expression studies. Case report/case series articles, editorials, letters to the editor, clinical guidelines, and reviews were excluded.

Data Extraction
Extracted data included study design, clinical data collection, ethnicity, study size, mean age, women percentage, CRC stage, treatment regimen and setting, genes of interest investigated, type of outcome studied, and statistical measures of outcome. The extraction of data from the articles included was performed and reviewed by two researchers. The terminology employed by each author of the studies included was used to record outcome measures. The measure of association was calculated when it was not present in a study only if the information necessary for its determination was available. If any of the selected studies presented results for genetic variants other than SNPs, only the results related to the latter were collected.

Quality Assessment
To assess the quality of the included studies, a descriptive analysis was performed by two researchers evaluating nine items obtained from the Strengthening the Reporting of Genetic Association studies (STREGA) criteria [16]. In case of any discrepancies, a third researcher was consulted. The items analyzed covered the following subjects: (a) laboratory methods, (b) number of samples genotyped and genotyping success rate, (c) population stratification methodology, (d) genotype or haplotype inference methods, (e) Hardy-Weinberg equilibrium (HWE), and (f) indication of the novelty of the genetic association study. A value of 1 was assigned if the study complied with the item (Y), 0.5 if the Item was incomplete (I), and 0 if it was not complied with (N). A total score was calculated for each study by adding all the scores for the items assessed (range 0 to 9). Total score was expressed as a percentage (points obtained/maximum points × 100) to classify the studies as high (>80%), moderate (50-80%), or low (<50%) quality, as done previously [17][18][19].

Search Results
The initial search delivered 1383 studies (Figure 2). After deletion of duplicates (n = 302) and removal of studies that did not comply with the inclusion criteria in the screening of titles and abstracts (n = 1020), 61 records were preselected for full-text review. Of these, 36 were eliminated because they included patients treated with other FPs (5-FU or S-1), 1 because DNA was obtained from tumor tissue, 1 because variant analysis focused on irinotecan-based therapy, 1 because its statistical analysis focused on association of genetic polymorphisms with plasma 5-FU levels, 1 because the association analysis was performed by haplotypes and not by single variants, 3 because patients with neoplasms other than CRC were included, and 5 for not complying with the study type for selection (3 case reports/case series, 1 editorial, 1 review). Finally, 13 studies were included in this review. study. A value of 1 was assigned if the study complied with the item (Y), 0.5 if the was incomplete (I), and 0 if it was not complied with (N). A total score was calculate each study by adding all the scores for the items assessed (range 0 to 9). Total score expressed as a percentage (points obtained/maximum points × 100) to classify the stu as high (>80%), moderate (50-80%), or low (<50%) quality, as done previously [17-19

Search Results
The initial search delivered 1383 studies ( Figure 2). After deletion of duplicates 302) and removal of studies that did not comply with the inclusion criteria in the scree of titles and abstracts (n = 1020), 61 records were preselected for full-text review. Of t 36 were eliminated because they included patients treated with other FPs (5-FU or S because DNA was obtained from tumor tissue, 1 because variant analysis focused o notecan-based therapy, 1 because its statistical analysis focused on association of ge polymorphisms with plasma 5-FU levels, 1 because the association analysis was formed by haplotypes and not by single variants, 3 because patients with neoplasms o than CRC were included, and 5 for not complying with the study type for selection (3 reports/case series, 1 editorial, 1 review). Finally, 13 studies were included in this rev

SNPs Associated with Capecitabine-Induced Toxicity
Genes, SNPs, and genotypes associated with capecitabine-induced toxicity are summarized in Table 2.

Cytidine Deaminase Gene (CDA)
For the CDA gene, two studies included in this review reported significant associations with severe toxicity in capecitabine-based therapy [22,25]. A study conducted in Spain found that the AA genotype of the CDA rs2072671 SNP (c.79A>C; missense; p.Lys27Gln) was associated with a greater risk of severe overall toxicity (grade 3-4) in 239 patients with CRC (stages I-IV) treated with various capecitabine-based therapeutic regimens (OR = 1.84; 95% CI = 1.06-3.18; p = 0.029 for AA vs. AC/CC) [22]. In line with these results, Pellicer et al. (European ancestry; Spain) found a protective effect of the C allele of this SNP against severe overall toxicity in 301 patients with CRC (stages I-IV) under treatment with capecitabine-based therapeutic regimens (OR = 0.5; 95% CI = 0.30-0.83; p = 0.007 for AC/CC vs. AA) [25]. They also found a significant association between this genotype and the presence of severe hand-foot syndrome (HFS) (OR = 0.27; 95% CI = 0.10-0.71; p = 0.008 for AC/CC vs. AA) [25]. Moreover, they reported that the T allele of the CDA rs1048977 SNP (c.435C>T; synonymous; p.Thr145Thr) was associated with the presence of severe hyperbilirubinemia (OR = 8.62; 95% CI = 1.05-70.24; p = 0.044 for CT/TT vs. CC) [25].
Moreover, two of the studies included in this review reported significant associations between SNPs of the ENOSF1 gene and toxicity during capecitabine-based therapy [21,22].  [31]. No significant association was reported between the MTHFR rs1801131 SNP (1286A>C; missense; p.Glu429Ala) and toxicity in CRC patients treated with capecitabine [20,31].

Gene Variants Associated with Capecitabine Effectiveness
The genes, SNPs, and genotypes associated with capecitabine effectiveness are summarized in Table 3.  (stages II-IV). This study found no significant association between the ABCB1 rs1128503 and rs1045642 SNPs (c.3435T>C; synonymous; p.Ile1145Ile) and treatment response [30].

ERCC Excision Repair 2 (ERCC2)
Two studies included in this review provided results related to the association of SNPs in the ERCC2 gene with the effectiveness of capecitabine-basedtherapy [29,30]

Methylenetetrahydrofolate Reductase Gene (MTHFR)
None of the studies that investigated the influence of SNPs in the MTHFR gene on treatment effectiveness reported statistically significant results [20,27,29].

Quality Assessment
The quality score assigned to each included study is available in Supplementary Table S2. Quality scores ranged between 33.33-72.22%. Most of the studies (7/13; 53.85%) had a moderate level of quality. Six studies (46.15%) fully reported the genotyping methods used. The seven studies (53.85%) that were considered incomplete did not provide information on DNA storage conditions or genotyping platforms used. All studies (100%) indicated whether their research reported new associations, replicated previous studies, or both. Nine studies (69.23%) reported both the number of samples to be genotyped and those that were successfully genotyped and considered HWE in the analysis. A substantial proportion of the included studies did not mention call rates and error rates (9/13; 69.23%), the center where genotyping was performed (12/13; 92.31%) or did not mention whether or not genotype or haplotype inference was performed (8/13; 61.54%).

Discussion
This systematic review identified numerous SNPs in genes involved in the PK and PD of capecitabine that may influence the toxicity or effectiveness of antineoplastic therapy in CRC patients. Genes involved in bioactivation, metabolism, transport, or mechanism of action of capecitabine, DNA repair, and the folate cycle have been associated with toxicity, while genes involved in DNA repair have been significantly associated with therapy effectiveness.
CDA gene plays an important role in the bioactivation of capecitabine into 5-FU and in the detoxification of other antimetabolite elements such as gemcitabine, decitabine, and cytarabine [33]. Studies included in this review reported that CDA rs2072671 and rs1048977 SNPs were significantly associated with overall toxicity, HFS, and severe hyperbilirubinemia [22,25]. Both these SNPs have been associated with altered PK, enzyme activity, and exposure to drugs metabolized by CDA [34,35]. It has been reported that the C allele of the CDA rs2072671 SNP has a detrimental effect on enzyme activity, which would hypothetically lead to a decrease in capecitabine activation, and consequently less exposure to 5-FU [36]. However, other studies provide contradictory information and indicate that the resulting catalytic activity depends on the substrate analyzed [33,37,38]. Studies conducted in patients of Asian (China) and European (Switzerland, The Netherlands) ancestry with gastric neoplasms treated with capecitabine-based regimens have not reported a significant association between this SNP and severe toxicity [39,40]. In line with the results of this review, two meta-analyses, in 1093 and 623 patients of mixed (international) ethnicity with pancreatic and non-small-cell lung cancer (NSCLC) treated with gemcitabine, found a significant association between the C allele of the CDA rs2072671 SNP and the presence of hematological toxicity [34,38]. Both studies highlighted this SNP as a potential predictive biomarker of antineoplastic therapy toxicity [34,38].
There are no other studies that have been conducted in patients treated with capecitabine for the CDA rs1048977 SNP. However, a narrative review reported that of a total of two included studies evaluating the association of this SNP with gemcitabine toxicity, only one reported that pancreatic cancer patients of mixed ethnicity (USA) carrying the T allele were significantly associated with neutropenia [34]. In turn, recent studies in patients with solid tumors of the pancreas, bladder, and lung in European (Poland) and Asian (China) populations have also reported a significant association of the T allele of this SNP with gemcitabine-related toxicity [41,42]. The DPD enzyme is the first and rate-limiting step in the catabolism of 5-FU, converting it into the metabolite DHFU. DPYD is the only gene that currently has SNP-validated biomarkers of FP toxicity in everyday practice [8]. No studies reporting significant results with regard to the association of the clinically relevant DPYD SNPs rs3918290, rs55886062, rs67376798, and rs75017182/rs56038477 with capecitabinebased treatment toxicity were included in this review. This may be due to (a) the fact that these are the first variants studied in relation to FP-associated toxicity, and therefore most of the studies focusing mainly on these SNPs could have been published more than 10 years ago, (b) the extremely low frequency of these SNPs in the overall population, requiring a larger sample size than the studies included in this review, and (c) the fact that the great majority of studies of these SNPs have been conducted in patients with various solid neoplasms treated with 5-FU based regimens or without distinction of FP agent [8,43,44].
Over the last decade, there has been a notable increase in the study of the influence of SNPs other than the four clinically relevant variants in the DPYD gene, which have a relatively higher frequency in the overall population, on capecitabine-based therapy toxicity. This review included five studies that reported contradictory results regarding the association of various SNPs in the DPYD gene with capecitabine toxicity [21,23,25,26,28,31]. These results agree with studies conducted in patients with other cancers or treated with other FPs [45][46][47][48][49][50][51][52]. A meta-analysis of six studies in 6119 European ancestry (international) patients with solid neoplasms (gastrointestinal, breast, pancreas, bile duct, among others) treated with FPs reported that the DPYD rs1801160 SNP was associated with an increased risk of toxicity, indicating that this SNP should be included in clinical practice [45]. In turn, a study in 503 CRC patients of European ancestry (Croatia) treated with FPs reported that patients carrying the DPYD rs1801160 and rs2297595 SNPs showed, respectively, a tendency towards and a significant association with severe adverse events. This study also mentioned that the inclusion of these SNPs in clinical practice should be considered [46]. Similar findings are reported by a study in 508 patients of European ancestry (Italy) with CRC (stages II-III) treated with FP. This study found a significant association between the DPYD rs1801160 and rs2297595 SNPs and severe adverse events during therapy [47]. Two recent studies in patients with gastrointestinal neoplasms (n = 80 and 93), of Jordanian (Jordan) and Latin American (Chile) origin, reported a significant association between the DPYD rs1801265 SNP and adverse events during FP therapy [49,50]. However, two other studies conducted in patients with gastrointestinal neoplasms (n = 503 and 113), of European and African American origin (Croatia, USA), found no significant association between this SNP and FP severe toxicity [46,51]. The differences in these findings may be due to several factors, such as sample size and ethnicity of the population being investigated. A similar situation occurred with the DPYD rs17376848 SNP regarding sample size. A previous study in 64 European patients (Switzerland) with metastatic gastrointestinal carcinomas treated with capecitabine-based therapeutic regimens reported a significant association between this SNP and diarrhea and HFS [52]. However, recent studies with larger sample sizes (n = 508 and 1254 patients, respectively) in European populations (Italy) with solid stomach, colon, and breast tumors did not demonstrate such association [47,48].
UMPS is an enzyme that metabolizes 5-FU into other metabolites with cytotoxic activity and is considered to be a major regulator of the cytotoxic effects of 5-FU [53]. One study included in this review reported an association of the UMPS rs4678145 and rs2279199 SNPs with asthenia and severe nausea/vomiting in CRC patients treated with capecitabine-based regimens [25]. However, a similar study in 338 Asian patients (China) with gastrointestinal cancers treated with capecitabine found no significant association between the UMPS rs4678145 and rs2279199 SNPs and severe toxicity [54].
SLC22A7 is the gene that encodes the organic anion transporter 2 (OAT2). This protein is responsible for transporting 5-FU within cells [55]. One study included in this review reported an association of the SLC22A7 LD SNPs rs2270860 and rs4149178 with cutaneous toxicity and severe diarrhea in CRC patients treated with capecitabine-based regimens [25]. Evidence on SNPs in this gene and their relationship with transporter activity is limited, and no significant association has been found between these variants and OAT2 hepatic expression [55]. There are no other studies that evaluate the influence of these SNPs on capecitabine-based treatment toxicity. However, a study in 344 pediatric cancer patients (mostly with leukemias and lymphomas) of mixed ethnicity (Canada) described an association between the G allele of the SLC22A7 rs4149178 SNP and a lower risk of cardiotoxicity during anthracycline treatment [56].
TS is a key enzyme for DNA biosynthesis and is the main therapeutic target of 5-FU [57]. ENOSF1 is an enzyme with numerous isoforms. While one isoform exhibits catalytic activity, others seem to have a regulatory role on TS activity [58]. ENOSF1 and TYMS genes partially overlap and are transcribed in opposite directions [59]. Four studies included in this review reported the association of several SNPs in these genes (ENOSF1 rs261091 and rs2741171, TYMS/ENOSF1 rs699517 and rs2790, and TYMS rs2853741 and rs3786362) with the risk of toxicity during capecitabine treatment, especially HFS [21,22,25,32]. A meta-analysis of 1912 patients of predominantly European ancestry (international) with gastrointestinal cancers reported that carriers of the G allele of the ENOSF1 rs2612091 SNP showed a higher risk of HFS during FP treatment. This study highlighted the essential role of ENOSF1 and of other variants in TYMS (6bp-indel and 28bp-repeat) in HFS development during FP treatment [59]. Another study in 342 Asian patients (China) with metastatic breast cancer under capecitabine-based therapy reported that patients carrying the CT genotype of the TYMS rs2853741 SNP had 2.25 times more risk of HFS [60].
MTHFR is a critical enzyme in the folate cycle and plays an important role in the PD of FPs. This enzyme catalyzes the irreversible conversion of 5,10-methylenetetrahydrofolate (5,10-MTHF) to 5-methyltetrahydrofolate, reducing the amount of 5,10-MTHF available. 5,10-MTHF is essential for the formation of a ternary complex with the active metabolite of 5-FU, FdUMP, and the TS enzyme, which results in the inhibition of the enzyme [61]. It has been suggested that MTHFR activity may be a crucial factor for predicting FP response and toxicity [61,62]. The MTHFR rs1801133 SNP results in decreased MTHFR activity leading to an increased intracellular concentration of 5,10-MTHF, which could enhance the formation of the 5,10-MTHF/FdUMP/TS ternary complex, increasing the risk of FP toxicity [63]. Only one study included in this review reported a significant association between the MTHFR rs1801133 SNP and toxicity during capecitabine therapy [31]. A study conducted in 50 patients of Latin American origin (Costa Rica) with metastatic CRC treated with FP-based chemotherapy reported that carriers of the T allele showed a higher risk of hematological, neurological, and HFS toxicity [62]. A meta-analysis of 20 publications revealed that in 1635 lung cancer patients of Asian (China) and European (Spain) origin, carriers of the T allele of the MTHFR rs1801133 SNP showed a higher risk of hematological toxicity during oxaliplatin-based therapy [64].
ERCC1 and ERCC2 are nucleotide excision repair genes. They are part of the socalled DNA repair genes and play a key role in tumor response to chemotherapy-induced DNA damage [29]. ERCC1 protein is involved in the DNA damage incision process and ERCC2 in the damage unwinding process. Polymorphisms in ERCC1 and ERCC2 could alter the ability to repair DNA, thereby affecting the response or survival of cancer patients [65]. Two studies included in this review reported a significant association of the ERCC1 rs11615, ERCC1 rs10412761, and ERCC2 rs1799787 SNPs with response to neoadjuvant capecitabine-based chemoradiation in European ancestry patients (Spain, France) [24,29]. However, contradictory results have been reported for the ERCC1 rs11615 SNP in patients of Asian origin (India) under adjuvant CAPOX treatment [30]. The most widely studied SNP of the ERCC1 gene is rs11615 [66][67][68]. It has been reported that this SNP has a negative impact on mRNA expression level, which is related to a decrease in ERCC1 repair function, resulting in an improved response to cytotoxic treatment [69]. A meta-analysis of six studies in 1137 patients with osteosarcoma suggested that in one Asian population (China), the ERCC1 rs11615 SNP is significantly associated with response to platinum-based chemotherapy, indicating that carriers of the C allele would benefit more from therapy [66]. In turn, a meta-analysis of 26 studies in 1401 patients of Asian and European ancestry (international) with NSCLC, revealed that the ERCC1 rs11615 SNP was associated with overall response rate (ORR) [67]. In contrast, a meta-analysis of 22 studies including 2846 patients of Asian and European ancestry (international) with advanced CRC treated with FPs and platinum found no significant association between the ERCC1 rs11615 SNP and ORR [68]. However, it did report a significant correlation between the T allele of this SNP and lower OS and progression-free survival (PFS). Notably, stratified analysis by ethnicity revealed that the T allele of this SNP was associated with worse survival profiles in patients of Asian origin, but with favorable prognostic outcomes in European ancestry patients. This finding reveals the high significance of ethnicity regarding genetic influence on therapy outcomes [68]. Concerning the ERCC1 rs10412761 SNP, previous studies on populations of predominantly European ancestry (USA) with pancreatic and ovarian cancer treated with several therapeutic strategies found no significant association between this SNP and survival or time to cancer recurrence [70,71]. Finally, no other studies evaluating the influence of the ERCC1 rs10412761 SNP on the effectiveness of chemotherapy was found.
The differences observed in the results obtained for certain SNPs with respect to their impact on toxicity or effectiveness of capecitabine-based therapy among the studies included in this review may be due to several factors, including (a) high methodological diversity among studies, such as design, population ethnicity, sample size, clinical data collection, response or toxicity assessment methods, and capecitabine therapeutic regimens evaluated, as well as (b) differences in the statistical methodology applied to control for confounding factors and whether or not adjustments for multiple comparisons were applied. This review has some limitations that need to be mentioned: (a) It only included studies that analyzed patients with CRC under capecitabine-based treatment, and this substantially reduced the possible number of results and impedes their generalizability to other FPs and neoplasms. (b) Moreover, it only examined the influence of SNPs on the effectiveness and safety of capecitabine-based therapy, excluding the possible effect of other genetic variants that have been associated with effects on capecitabine therapy outcomes (tandem repeats, copy number variations, insertions, deletions, etc.). (c) Due to the high variety of SNPs reviewed, it was not possible to perform a meta-analysis to observe variations in the level of association of genotypes with the outcomes of capecitabine-based therapy. (d) A large proportion of the SNPs examined for each gene were reported by only one study, revealing the need for further evidence to corroborate these findings. (e) The included studies were in the moderate to low range of methodological quality according to the STREGA statement criteria, thus interpretations of the findings of this review must be treated with caution.

Conclusions
In conclusion, the results obtained in this systematic review suggest that according to the recent literature, as well as the four SNPs of current clinical relevance in DPYD, there are other SNPs in genes related to the PD (TYMS, ENOSF1, MTHFR, ERCC1, and ERCC2) and PK (CDA, DPYD, UMPS, and SLC22A7) of capecitabine that could come to be considered in the future as predictive biomarkers of the outcomes of capecitabinebased therapy in patients with CRC. Specifically, current evidence suggests that SNPs TYMS/ENOSF1 rs699517, ENOSF1 rs2612091, and CDA rs2072671 appear to have a nearer future as biomarkers in clinical practice, although they still require further prospective validation. The remaining SNPs in the aforementioned genes require additional studies to elucidate their influence on capecitabine toxicity and effectiveness in CRC patients.  Funding: The Virgen de las Nieves University Hospital Biobank was supported by grants co-funded by ERDF funds (EU) from the Instituto de Salud Carlos III (PT13/0010/0039).

Institutional Review Board Statement: Not applicable.
Informed Consent Statement: Not applicable. Data Availability Statement: All data can be found in the text.

Conflicts of Interest:
The authors declare no conflict of interest.