BACKGROUND: Colorectal cancer is still the second leading cause of cancer-related deaths and thus biomarkers allowing prediction of the resistance of patients to therapy and estimating their prognosis are needed. We designed a panel of 558 genes with pharmacogenomics records related to 5-fluorouracil resistance, genes important for sensitivity to other frequently used drugs, major oncodrivers, and actionable genes. We performed a target enrichment sequencing of DNA from tumors and matched blood samples of patients, and compared the results with patient prognosis stratified by systemic adjuvant chemotherapy. RESULTS: The median number of detected variants per tumor sample was 18.5 with 4 classified as having a high predicted functional effect and 14.5 moderate effect. APC, TP53, and KRAS were the most frequent mutated genes (64%, 59%, and 42% of mutated samples, respectively) followed by FAT4 (23%), FBXW7, and PIK3CA (16% for both). Patients with advanced stage III had more frequently APC, TP53, or KRAS mutations than those in stages I or II. KRAS mutation counts followed an increasing trend with grade (G1 < G2 < G3). The response to adjuvant therapy was worse in carriers of frameshift mutations in APC or 12D variant in KRAS, but none of these oncodrivers had prognostic value. Carriage of somatic mutations in any of the genes ABCA13, ANK2, COL7A1, NAV3, or UNC80 had prognostic relevance for worse overall survival (OS) of all patients. In contrast, mutations in FLG, GLI3, or UNC80 were prognostic in the same direction for patients untreated, and mutations in COL6A3, LRP1B, NAV3, RYR1, RYR3, TCHH, or TENM4 for patients treated with adjuvant therapy. The first association was externally validated. From all germline variants with high or moderate predicted functional effects (median 326 per patient), > 5% frequency and positive Manhattan plot based on 3-year RFS, rs72753407 in NFACS, rs34621071 in ERBB4, and rs2444274 in RIF1 were significantly associated with RFS, OS or both. CONCLUSIONS: The present study identified several putative somatic and germline genetic events with prognostic potential for colorectal cancer that should undergo functional characterization.

• Keywords
• Carcinoma, Colorectal, Drug, Oncodriver, Pharmacogene, Prognosis, Resistance, Sequencing,
• MeSH
• Drug Resistance, Neoplasm genetics   MeSH
• Adult MeSH
• F-Box-WD Repeat-Containing Protein 7 genetics   MeSH
• Pharmacogenetics methods   MeSH
• Fluorouracil therapeutic use   MeSH
• Class I Phosphatidylinositol 3-Kinases MeSH
• Colorectal Neoplasms * genetics   drug therapy   pathology   MeSH
• Middle Aged MeSH
• Humans MeSH
• Mutation genetics   MeSH
• Biomarkers, Tumor genetics   MeSH
• Tumor Suppressor Protein p53 genetics   MeSH
• Prognosis MeSH
• Adenomatous Polyposis Coli Protein genetics   MeSH
• Proto-Oncogene Proteins p21(ras) genetics   MeSH
• Aged MeSH
• High-Throughput Nucleotide Sequencing MeSH
• Check Tag
• Adult MeSH
• Middle Aged MeSH
• Humans MeSH
• Male MeSH
• Aged MeSH
• Female MeSH
• Publication type
• Journal Article MeSH
• Research Support, Non-U.S. Gov't MeSH
• Names of Substances
• APC protein, human MeSH Browser
• F-Box-WD Repeat-Containing Protein 7 MeSH
• Fluorouracil MeSH
• Class I Phosphatidylinositol 3-Kinases MeSH
• KRAS protein, human MeSH Browser
• Biomarkers, Tumor MeSH
• Tumor Suppressor Protein p53 MeSH
• PIK3CA protein, human MeSH Browser
• Adenomatous Polyposis Coli Protein MeSH
• Proto-Oncogene Proteins p21(ras) MeSH
• TP53 protein, human MeSH Browser

BACKGROUND: Colorectal cancer is a highly prevalent and deadly. The most common metastatic site is the liver. We performed a whole exome sequencing analysis of a series of metachronous colorectal cancer liver metastases (mCLM) and matched non-malignant liver tissues to investigate the genomic profile of mCLM and explore associations with the patients' prognosis and therapeutic modalities. METHODS: DNA samples from mCLM and non-malignant liver tissue pairs (n = 41) were sequenced using whole exome target enrichment and their germline and somatic genetic variability, copy number variations, and mutational signatures were assessed for associations with relapse-free (RFS) and overall survival (OS). RESULTS: Our genetic analysis could stratify all patients into existing targeted therapeutic regimens. The most commonly mutated genes in mCLM were TP53, APC, and KRAS together with PIK3CA and several passenger genes like ABCA13, FAT4, PCLO, and UNC80. Patients with somatic alterations in genes from homologous recombination repair, Notch, and Hedgehog pathways had significantly prolonged RFS, while those with altered MYC pathway genes had poor RFS. Additionally, alterations in the JAK-STAT pathway were prognostic of longer OS. Patients bearing somatic variants in VIPR2 had significantly shorter OS and those with alterations in MUC16 prolonged OS. Carriage of the KRAS-12D variant was associated with shortened survival in our and external datasets. On the other hand, tumor mutation burden, mismatch repair deficiency, microsatellite instability, mutational signatures, or copy number variation in mCLM had no prognostic value. CONCLUSIONS: The results encourage further molecular profiling for personalized treatment of colorectal cancer liver metastases discerning metachronous from synchronous scenarios.

• Keywords
• Colorectal cancer, Exome, Liver Metastasis, Metachronous, Prognosis, Therapy,
• Publication type
• Journal Article MeSH