Abstract
Background/Aim: Hand–foot syndrome (HFS) is a common adverse event associated with capecitabine. While the fibroblast growth factor receptor (FGFR) signaling pathway is involved in skin wound healing, its essential role in capecitabine-induced HFS remains unclear. We examined the association between polymorphisms in FGFR signaling pathway-related genes and capecitabine-induced HFS in patients with colorectal cancer (CRC).
Patients and Methods: This retrospective study included patients with CRC who received capecitabine and oxaliplatin. HFS was evaluated using CTCAE v5.0, up to 2 cycles. Polymorphisms were identified using whole-exome sequencing and confirmed using direct sequencing. The association between HFS and polymorphisms was analyzed using Fisher’s exact test. The Bonferroni correction for multiple testing was performed to calculate the corrected p-value (Pc).
Results: Overall, 937 polymorphisms were identified in 71 genes. Intronic FGFR2 rs2981460 C/C, and rs2981461 T/T genotypes, and haplotype II/II comprising the C and T alleles were associated with a lower HFS incidence (p=0.0161, Pc=0.113; p=0.0163, Pc=0.114; and p=0.0161, Pc=0.113, respectively). Synonymous FGFBP2 rs2286459 A/A was associated with a lower HFS frequency (p=0.0469, Pc=0.328). 3′-Untranslated region and nonsynonymous variants SPRY2 rs11911 T/G or G/G and rs504122 G/A or A/A, and homozygotes or heterozygotes of haplotype 2 comprising respective G and A alleles, were significantly associated with higher HFS incidence (p=0.0000803, Pc=0.000562; p=0.0000803, Pc=0.000562; and p=0.0000803, Pc=0.000562, respectively). A significantly higher HFS incidence was observed in patients with a combined risk genotype and diplotypes of FGFR2 any/II or any/any, FGFBP2 rs2286459 G/G or G/A, and SPRY2 2/2 or any/2 (p=0.0000314).
Conclusion: Variants in FGFR signaling pathway-related factors were significantly associated with capecitabine-induced HFS.
Introduction
Capecitabine is an orally administered anticancer agent that undergoes a three-step metabolic activation to produce its active metabolite, 5-fluorouracil (5-FU) (1). The final step involves metabolic conversion by thymidine phosphorylase, an enzyme predominantly expressed in tumor tissues, to form 5-FU, which in turn exerts antitumor effects. Capecitabine is used worldwide to treat various solid tumors, including colorectal cancer (CRC), and combination chemotherapy with capecitabine and oxaliplatin (CapeOX) has proven to be highly effective as an adjuvant treatment for CRC and metastatic CRC (2, 3), even in elderly patients (4). Despite its anticancer efficacy, capecitabine frequently induces adverse events including hand-foot syndrome (HFS), which occurs at a high rate (2, 5, 6). Patients with CRC who were treated with CapeOX developed grade ≥1 and grade ≥2 HFS with frequencies of 30% (2) and ~15% (2, 6), respectively. The development of HFS, especially at grade ≥2 severity, frequently leads to a reduction in capecitabine dose and sometimes discontinuation of the treatment.
HFS is characterized by symmetric erythema and edema accompanied by the onset of neuropathic pain, which is typically observed in the skin and pressure-prone areas such as the palms and soles. The erythema can progress to blistering with subsequent desquamation, erosion, and ulceration, which reflect the degree of clinical severity. Histopathological features, including basal layer vacuolar degeneration or full-thickness necrosis, spongiosis, hyperkeratosis, and/or parakeratosis, have been observed (7); however, the complete mechanism of HFS development remains unclear, despite several approaches having been performed (8, 9). Given that HFS causes damage to skin tissues due to internal factors stimulated by capecitabine treatment, processes related to wound healing may contribute to the development and worsening of HFS.
The normal mammalian response to injury occurs in three overlapping but distinct stages: inflammation, new tissue formation, and remodeling (10). The fibroblast growth factor receptor (FGFR) signaling pathway is associated with cutaneous wound repair through the positive regulation of reepithelialization (10, 11). The IIIb variant of FGFR2 (FGFR2-IIIb), predominantly expressed in epithelial tissues (12), particularly contributes to wound repair (10, 11); however, its role in HFS development remains unclear. Based on this evidence, we hypothesized that the factors associated with the FGFR signaling pathway contribute to the development of capecitabine-induced HFS. Our hypothesis is supported by the fact that the sprouty receptor tyrosine kinase signaling antagonist 2 (SPRY2) polymorphism rs117876855, which encodes a factor located downstream of FGFR2, is significantly associated with the development of capecitabine-induced HFS (13).
Therefore, in the present study, we retrospectively examined the effects of genetic polymorphisms in factors associated with the FGFR signaling pathway on capecitabine-induced HFS by analyzing clinical data and genome samples collected in our prospective study of patients with CRC who were treated with CapeOX (14).
Patients and Methods
Study design. This was a retrospective study using data and samples obtained from our previous prospective pharmacokinetic, pharmacodynamic, and pharmacogenetic studies. The primary endpoint of the study was to examine whether or not proton-pump inhibitor treatment with capecitabine affects the pharmacokinetics of capecitabine and its metabolites (15), and the secondary endpoint was to analyze the associations between capecitabine pharmacokinetics, pharmacodynamics, and pharmacogenetics (14). From the previous study, patients with stage III or high-risk stage II CRC who fully recovered after curative resection and were eligible for adjuvant therapy with CapeOX, or patients with metastatic CRC who were candidates for treatment with CapeOX, with or without bevacizumab therapy, were included (14). Two patients were newly enrolled following the prior analyses (14). The total number of patients in the present study was 39. Initially, we conducted whole-exome sequencing, focusing on factors related to the FGFR-related signaling pathway, to identify polymorphisms associated with capecitabine-induced HFS by comparing the genotype frequencies observed in patients who developed any grade of HFS with those in patients without HFS. Polymorphisms with minor allele frequencies of less than 0.10 in the Japanese or East Asian population were excluded (16). To identify polymorphisms that were significantly associated with capecitabine-induced HFS (p<0.05), we compared the frequencies of the three genotypes with zero, one, and two polymorphic alleles in patients with and without HFS. After identification, polymorphisms were confirmed by direct DNA sequencing, as described below. This study was conducted in accordance with the ethical standards outlined in the Declaration of Helsinki. The study protocol was approved by the Institutional Review Board of Showa Medical University and registered in the University Hospital Medical Information Network-Clinical Trials Registry (UMIN000031182). All the patients provided written informed consent for the use of their peripheral blood samples and medical information for research purposes.
Patients. Eligible patients were aged 20 years or older with histologically confirmed postoperative or metastatic CRC and were indicated for treatment with a CapeOX-containing regimen in an adjuvant or metastatic setting. The details of the eligibility criteria have been described previously (14).
Treatment. Oral capecitabine (1,000 mg/m2) was administered twice on day 1. For the pharmacokinetic analysis of the first capecitabine dose, oxaliplatin (and bevacizumab for metastatic CRC) administration on day 1 was shifted to day 2 in the first cycle. Details of the treatments have been described previously (14).
Toxicity evaluation. All patients who received CapeOX were assessed for clinical and laboratory adverse events including HFS, according to the Common Terminology Criteria for Adverse Events, version 5.0 (14) during the first two treatment cycles. Because, as described above, the association of pharmacokinetics on day 1 and toxicities of capecitabine was analyzed, the duration of toxicity evaluation was restricted to the first two treatment cycles, which the day 1 pharmacokinetics might influence.
Whole-exome and direct sequencing. Whole-exome sequencing was performed as previously described elsewhere (17). Briefly, genomic DNA extracted from peripheral blood was used for library preparation with the SureSelect Human All Exon V6 kit (Agilent, Santa Clara, CA, USA), and sequencing was conducted on the Illumina NovaSeq 6000 and NovaSeq X Plus platforms with 150 bp paired-end reads. Base calling, read trimming, and FASTQ conversion were performed using the Illumina package bcl2fastq v2.20.0., as described previously (17). Raw reads were aligned to the GRCh38/hg38 reference genome using the Burrows–Wheeler Aligner (bwa-0.7.17). Duplicate reads were removed with Picard (picard-tools-2.18.2-SNAPSHOT; http://broadinstitute.github.io/picard/), and variant calling was carried out using the Genome Analysis Toolkit (GATKv4.0.5.1; https://www.broadinstitute.org/gatk/). Variant annotation was performed using SnpEff (SnpEff_vSnpEff 4.3t 2017-11-24; http://snpeff.sourceforge.net/SnpEff.html). For the present study, variants were retained only if they met the following criteria: depth ≥10 and genotype quality ≥20. For single-nucleotide polymorphisms, QualByDepth (QD) <2.0, FisherStrand (FS) >60.0, MappingQuality (MQ) <40.0, MQRankSum <−12.5, and ReadPosRankSum <−8.0. For insertions and deletions, QD <2.0, FS >200.0, ReadPosRankSum <−20.0. The median of the mean depth of target regions across samples was 84.5 (range=70.8-125.3).
For direct sequencing, the following polymorphisms were analyzed: rs2981460 and rs2981461 in FGFR2, rs2286459 in FGFBP2, and rs11911 and rs504122 in SPRY2. The primers used for polymerase chain reaction (PCR) to amplify the gene fragments containing these polymorphisms and for sequencing are listed in Table SI. The reaction mixture for amplifying the genomic region consisted of genomic DNA (50 ng), 1×PCR buffer, 2 mM MgCl2, 0.2 mM dNTPs, 0.2 μM of each primer, and 1.25 U of AmpliTaq Gold® 360 DNA Polymerase (Roche, Branchburg, NJ, USA) in a final volume of 25 μl. PCR cycling conditions for amplifying this region were initial denaturation at 95°C for 10 min, followed by 30 cycles of denaturation at 95°C for 15 s, annealing at 60°C for 30 s, and extension at 72°C for 1 min, except for analysis of rs2981460 and rs2981461 in FGFR2 with extension at 72°C for 2 min. The final extension step was performed at 72°C for 7 min.
Statistical analysis. Allele and genotype frequencies for each polymorphism, and the significance of the deviations from the Hardy–Weinberg equilibrium were determined using SNPAlyze 9.0 (Dynacom, Yokohama, Japan). Linkage disequilibrium analysis was performed to estimate the correlation coefficient r-square and Lewontin’s coefficient D’ among the nucleotide polymorphisms. Associations between two or three groups of categorical variables were analyzed using Fisher’s exact test. Bonferroni correction for multiple genotype and diplotype comparisons was performed to obtain corrected p-values (Pc). All analyses were performed using JMP software (version 17.0; SAS Institute, Cary, NC, USA). Associations were considered statistically significant when the two-tailed p-Value was less than 0.05.
Results
Patient characteristics. The characteristics of the 39 patients are presented in Table I. All patients had adequate renal and liver function to meet the eligibility criteria. Twenty-five of the 39 included patients developed HFS after the initiation of capecitabine-containing therapy. Grades 1, 2, and 3 HFS were observed in 21 (53.8%), 2 (5.13%), and 2 (5.13%) patients, respectively. The characteristics of patients with HFS of any grade and those without HFS were not significantly different.
Patient characteristics.
Whole-exome sequencing to identify genes associated with HFS. Overall, 937 polymorphisms in 71 genes encoding factors related to the FGFR signaling pathway were identified using whole-exome sequencing (Table SII). Next, we examined the association between these polymorphisms and the HFS of any grade. Five polymorphisms, including rs2981460 (T>C, intron) and rs2981461 (C>T, intron) in FGFR2, rs2286459 (G>A, Phe84Phe) in FGFBP2, and rs11911 [T>G, 3′-untranslated region (3′-UTR)] and rs504122 (G>A, Pro106Ser) in SPRY2 were significantly associated with the any grades of HFS (Table II).
Associations between development of HFS and FGFR signaling pathway-related plymorphisms in whole-exome sequencing.
Allele and genotype frequencies of polymorphisms in FGFR2, FGFBP2, and SPRY2. Genetic polymorphisms, identified using whole-exome sequencing, were confirmed using direct DNA sequencing. The allele and genotype frequencies are summarized in Table III. The calculated allele frequencies of polymorphisms in FGFR2, FGFBP2, and SPRY2 were almost consistent with previously reported data (18-22) and confirmed to Hardy–Weinberg equilibrium (p>0.05). Polymorphisms rs2981460 (T>C) and rs2981461 (C>T) in FGFR2, and rs11911 (T>G) and rs504122 (G>A) in SPRY2 were linked (r-square=0.812, D’ value=1.00, and r-square=0.950, D’ value=1.00, respectively). The estimated haplotypes of FGFR2 and SPRY2 are shown in Table IV.
Allele and genotype frequencies of the polymorphisms in FGFR2, FGFBP2, and SPRY2 genes.
Haplotype structure of polymorphisms in FGFR2 and SPRY2.
Associations between HFS development and FGFR2, FGFBP2, and SPRY2 polymorphisms. Next, we examined the effects of FGFR2, FGFBP2, and SPRY2 polymorphisms on the development of HFS induced by capecitabine (Table V). The frequencies of FGFR2 rs2981460 C/C and rs2981461 T/T genotypes in patients who developed HFS was significantly lower than those in patients without HFS [8.00% (2/25) vs. 42.9% (6/14), odds ratio (OR)=0.116, 95% confidence interval (CI)=0.0193-0.695, p=0.0161; and 4.00% (1/25) vs. 35.7% (5/14), OR=0.0750, 95% CI=0.00767-0.733, p=0.0163, respectively]. Consistent with these results, the frequency of homozygotes for haplotype II of FGFR2, consisting of rs2981460 C and rs2981461 T, was significantly lower in patients with HFS than in those without HFS (OR=0.116, 95% CI=0.0193-0.695, p=0.0161). In addition, the frequency of FGFBP2 rs2286459 A/A in patients with HFS was significantly lower than that in patients without HFS [4.00% (1/25) vs. 28.6% (4/14), OR=0.104, 95% CI=0.0103-1.05, p=0.0469]. These results indicated the protective effects of rs2981460 C/C, rs2981461 T/T, and diplotype II/II in FGFR2 and FGFBP2 rs2286459 A/A against HFS development; however, these associations were no longer significant after Bonferroni correction for multiple testing (P c=0.113, 0.114, 0.113, and 0.328, respectively). When analyzed as three-genotype or diplotype groups, rs2981460, rs2981461, and haplotype II in FGFR2, and rs2286459 in FGFBP2 were not associated with the development of HFS after Bonferroni correction for multiple testing (Pc=0.301, 0.279, 0.301, and 0.265, respectively).
Associations between development of HFS and plymorphisms in FGFR2, FGFBR2, and SPRY2.
In contrast, patients with HFS showed significantly higher frequencies of SPRY2 rs11911 T/G or G/G, rs504122 G/A or A/A genotypes, and diplotype 2/2 or any/2 than patients without HFS [96.0% (24/25) vs. 35.7% (5/14), OR=43.2, 95% CI=4.42-422, p=0.0000803, respectively], indicating that SPRY2 rs11911 T/G or G/G, rs504122 G/A or A/A, and diplotype 2/2 or any/2 are risk genotypes or diplotype for HFS. These associations remained significant even after Bonferroni correction for multiple testing was performed (P c=0.000562 for all associations). In addition, the three genotype groups rs11911, rs504122, and haplotype 2 of SPRY2 were significantly associated with the development of HFS (p=0.000145, 0.000228, and 0.000145, respectively). These associations remained significant after Bonferroni correction for multiple tests (Pc=0.00102, 0.00160, and 0.00102, respectively).
Combined effect of risk genotype and diplotypes of FGFR2, FGFBP2, and SPRY2 on the HFS development. The combined effect of risk genotype or diplotype across FGFR2, FGFBP2, and SPRY2 on HFS development was evaluated. The risk diplotypes in FGFR2 and SPRY2 were any/any or any/II and any/2 or 2/2, respectively (Table V). The risk genotype for FGFBP2 was rs2286459 G/G or G/A (Table V). The frequency of patients carrying the three risk diplotypes and genotype was significantly higher in the HFS group than in the no HFS group [84.0% (21/25) vs. 14.3% (2/14), OR 31.5, 95% CI 5.00-198, p=0.0000314].
Discussion
This study showed for the first time that FGFR2 rs2981460 C/C and rs2981461 T/T genotypes, as well as diplotype II/II, and FGFBP2 rs2286459 A/A genotype were protective factors against capecitabine-induced HFS. In contrast, SPRY2 rs11911 T/G or G/G, rs504122 G/A or A/A, and diplotype any/2 or 2/2 were risk genotypes and diplotype for HFS, respectively. These SPRY2 polymorphisms may be stronger biomarkers than FGFR2 and FGFBP2 genotypes or diplotype for predicting capecitabine-induced HFS, because 1) three genotypes or diplotype groups of SPRY2, but not those of FGFR2 and FGFBP2, were significantly associated with the development of HFS, and 2) the associations between the two or three genotypes or diplotype groups of SPRY2 and the development of HFS remained significant even though Bonferroni correction for multiple testing was performed. Combining the risk genotypes or diplotype of FGFR2 rs2981460 and rs2981461, FGFBP2 rs2286459, and SPRY2 rs11911 and rs504122 was significantly associated with a higher incidence of capecitabine-induced HFS, even after Bonferroni correction; however, the functional roles of these polymorphisms remain unclear. These findings demonstrate that the combined evaluation of the FGFR2, FGFBP2, and SPRY2 polymorphisms identified in this study before initiating the capecitabine-including regimen may help predict patients predisposed to capecitabine-induced HFS. However, further clinical studies with larger numbers of patients are necessary to confirm the reproducibility of the present results.
Capecitabine-induced HFS cause damage to skin tissues, characterized by pathological features such as basal layer vacuolar degeneration or full-thickness necrosis, spongiosis, hyperkeratosis, and/or parakeratosis, which are caused by internal factors stimulated by capecitabine treatment (7). Therefore, as a biological response to capecitabine-induced HFS, processes related to wound healing, such as the FGFR signaling pathway, may play a role in the repair of damage (10, 11) and, in turn, are associated with the development and worsening of HFS. We found that polymorphisms in three factors of the FGFR2 signaling pathway, FGFR2, FGFBP2, and SPRY2, were significantly associated with the development of capecitabine-induced HFS, suggesting an essential role for the FGFR2 signaling pathway.
FGFR2-IIIb, a splice variant of FGFR2, is involved in wound repair (10, 11). The polymorphisms FGFR2 rs2981460 and 2981461 are located within the FGFR2-IIIb isoform and may play a role in altering its function. Although both FGFR2 rs2981460 and rs2981461 are intronic variants, and intronic variants are generally non-functional, some have been shown to enhance transcription or splicing (23). We evaluated whether these polymorphisms were located in genomic regions associated with transcription or splicing processes using fanta.bio (24), RegulomeDB v.2 (25, 26), and Genotype-Tissue Expression (GTEx). However, we could not determine whether these variants were associated with the transcription or splicing of FGFR2.
FGFBP2 is a member of fibroblast growth factor (FGF) binging proteins and has binding sites for FGF ligands (27). This structural feature allows the molecule to reversibly bind to and increase the bioavailability of FGF ligands, which may further affect the FGFR signaling pathway. FGFBP2 rs2286459 is a synonymous variant (Phe84Phe) whose functional roles remain unclear. Although synonymous variants are generally silent and do not alter gene expression or protein function, some synonymous mutations affect mRNA stability and splicing regulation (28) and influence co- or post-translational protein folding (28, 29).
SPRY2 is located downstream of the FGFR2 signaling pathway, which negatively regulates this pathway by inhibiting FGFR substrate 2α and RAS (12). SPRY2 rs11911 is located in the 3′-UTR, which is known to be associated with the stability of mRNA and microRNA-dependent gene expression (30). AU-rich elements within the 3′-UTR are involved in cis-regulatory sequences that destabilize mRNA through the binding of RNA-binding proteins (30). Given that SPRY2 rs11911 (T>G) alters the AU-rich sequence in SPRY2 mRNA from ‘GAUUUUUUCU U UCUUUUUUUA’ to ‘GAUUUUUUCU G UCUUUUUUUA’, this polymorphism may contribute to the change in the stability of the SPRY2 mRNA. In contrast, according to miRDB (31, 32), SPRY2 rs11911 does not appear to be associated with microRNA-mediated gene expression. SPRY2 rs504122 is a non-synonymous variant that causes an amino acid change at Pro106Ser. This amino acid residue is located close to the site associated with SPRY2 stability, where the phosphorylation of Ser112 and Ser121 may lead to a decrease in ubiquitination (33). The polymorphic serine residue at position 106 may be phosphorylated, which may contribute to the increased stability of SPRY2.
Patients who were treated with the FGFR2 tyrosine kinase inhibitor are known to develop HFS (34, 35). For example, 20-40% of patients with cholangiocarcinoma carrying FGFR2 fusions or rearrangements who received futibatinib or tasurgratinib had HFS of any grade (34, 35). Futibatinib and tasurgratinib inhibit the kinase activities of FGFR1 through FGFR4 (36, 37), with particularly potent inhibitions of FGFR2, as demonstrated by the lowest half-maximal inhibitory concentrations of 1.4 nM (36) and 0.5 nM (37), respectively. These results support our hypothesis that the FGFR2 signaling pathway plays an essential role in the development of capecitabine-induced HFS; however, further clinical and basic studies are necessary to confirm this hypothesis.
Study limitations. First, this was a retrospective study with a relatively small sample size. Therefore, future prospective multicenter studies with other independent cohorts involving larger numbers of patients are required to validate the present results for their application in clinical practice. Second, basic research to address the functional roles of factors related to the FGFR2 signaling pathway is also necessary.
Conclusion
We revealed for the first time that polymorphisms in three factors of the FGFR2 signaling pathway, namely FGFR2, FGFBP2, and SPRY2, are significantly associated with capecitabine-induced HFS, suggesting that this pathway plays an essential role in the development of HFS. Combining the risk genotypes or diplotype of FGFR2 rs2981460 and rs2981461, FGFBP2 rs2286459, and SPRY2 rs11911 and rs504122 was significantly associated with a higher incidence of capecitabine-induced HFS. These findings may have clinical value in identifying patients who are predisposed to HFS before capecitabine treatment, thereby facilitating appropriate treatment of such patients.
Acknowledgements
We thank Mr. Kengo Omata for his contribution to the discussion of the functions of the identified polymorphisms. This project was supported by the Common Fund of the Office of the Director of the National Institutes of Health, NHGRI, NHLBI, NIDA, NIMH, and NINDS. The data used for the analyses described in this paper were obtained from the GTEx Portal on October14, 2025.
Footnotes
Authors’ Contributions
Riku Yamanaka, Natsumi Matsumoto, and Ken-ichi Fujita wrote the manuscript; Riku Yamanaka, Natsumi Matsumoto, and Ken-ichi Fujita designed the study; Riku Yamanaka, Natsumi Matsumoto, Remi Murase, Yutaro Kubota, Hiroo Ishida, Ken Shimada, and Ken-ichi Fujita performed the research; Riku Yamanaka, Natsumi Matsumoto, Remi Murase, Yutaro Kubota, Hiroo Ishida, Ken Shimada, and Ken-ichi Fujita analyzed the data.
Supplementary Material
Supplementary tables are available at the Zenodo repository: https://zenodo.org/records/17645954
Conflicts of Interest
The Authors declare that they have no conflicts of Interest.
Funding
This study was supported in part by a JSPS Grant-in-Aid for Early-Career Scientists (grant number 22K15326) awarded to NM.
Artificial Intelligence (AI) Disclosure
No artificial intelligence (AI) tools, including large language models or machine learning software, were used in the preparation, analysis, or presentation of this manuscript.
- Received November 20, 2025.
- Revision received December 9, 2025.
- Accepted December 11, 2025.
- Copyright © 2026 The Author(s). Published by the International Institute of Anticancer Research.
This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited.
References
In this issue
Jump to section
Related Articles
Cited By...
- No citing articles found.