Patients with bowel cancer gave written informed consent prior to their participation in General Hospital of Southern Theatre Command, PLA. Formalin-fixed, paraffin-embedded (FFPE) tumor tissues and matched blood samples in EDTA tubes (for germline tests) from 573 diagnosed bowel cancer patients (Information on clinicopathological status of patients is provided in Supplementary Table S2) were collected. All tumor FFPE sections were evaluated by pathologist to contain at least 20% tumor cells. Family history here is defined as the confirmed bowel cancer patient who has at least one family member (first and second-degree relatives) who had a history of tumor diagnosis. The immediate family member includes father, mother, brother(s), sister(s), son(s), daughter(s); second degree relatives include grandparent(s), uncle(s), aunt(s).

The FFPE tissues and peripheral white blood cells were collected to extract DNA using QIAamp DNA FFPE Tissue Kit and DNeasy Blood and Tissue Kit (Qiagen, Inc.), respectively. And the purified gDNA was quantified using the Qubit 3.0 Fluorometer (Life Technologies, Inc.).

For the matched germline and tumor samples, 100 ng of DNA was shared with a Covaris E210 system (Covaris, Inc.) to obtain an average of 200 bp fragments. Accel-NGS 2S DNA Library Kit (Swift Biosciences, Inc.) and xGen Lockdown Probes kit (IDT, Inc.) were used for NGS library preparation of the tumor gDNA matched germline gDNA. The custom xGen Lockdown probe was synthesized by IDT, Inc. to target the exons and selected intronic regions of 499 genes (Gene list is provided in Supplementary Table S1).

Pathogenicity of germline mutations was defined and predicted based on the five-grade classification system according to the American College of Medical Genetics and Genomics (ACMG) Guidelines for the Interpretation of Sequence (Li et al., 2017). It was modified here that pathogenic/likely-pathogenic germline variants were depicted as P/LP and the variant of undetermined significance (VUS), benign, likely benign, and somatic mutations were defined as the non-pathogenic group (Non-P) in our results. Therefore, all mutations were categorized into P/LP or Non-P groups in this study.

Tumor mutation burden of each sample was calculated according to a published and the method of Chalmers et al (2017).

Germline mutation data and incidence rates were obtained from the ChinaMAP database (http://www.GenomAD.org). The Cancer Genome Atlas (TCGA) database (https://tcga-data.nci.nih.gov/tcga/) provides several expression profiles and mutation data of CRC, as well as corresponding clinical data. Gene Ontology (GO) and the Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway terms were considered statistically significant when FDR < 0.01. CIBERSORT was used for evaluating diverse immune cell types in the cancer microenvironment. The violin software package was used to visualize differentially infiltrated immune cells between the two groups through the Wilcoxon test.

Statistical analyses were performed using the Statistical Package for the Social Sciences (SPSS) statistical package and Graphpad (Prism 8). Student’s t-test was performed when two groups were compared, and ANOVA and post hoc tests were performed when three or more groups were compared. Gene prevalence between different groups was analyzed by Chi-Square test or Fisher exact test under/with a dominant model. A two-sided p value of less than 0.05 was considered to be statistically significant.

Briefly, paired tumor/germline analysis was conducted using a customized next generation sequencing (NGS) panel of 499 selected genes (Supplementary Table S1). Somatic variants were determined by comparing the data between tumor and blood samples and all participants were included as a result of successful germline sequencing, which resulted in an evaluable population of 573 patients with bowel cancer. The demographic, clinical, and pathological characteristics of this patient cohort are shown in Table 1. In this cohort (n = 573), 21.64% (n = 124) participants who were diagnosed with bowel cancers were < 50 years old and 78.36% (n = 449) > 50 years old. Approximately 39.62% (n = 227) patients were female, 66.32% (n = 380) exhibited colon cancer, 27.4% (n = 157) owned rectal cancer, and 52.53% (n = 301) patients were reported to have the stage IV tumors. A total of 108 (18.85%) participants had one or more first-degree relatives with a history of tumor diagnosis, and 13 (2.27%) had one or more second-degree relatives with a history of tumor diagnosis.

Overall, 47 patients were found to carry P/LP germline mutations (Table 2). The age and sex of the patients were not associated with the presence or absence of a P/PL germline mutation (p = 0.19 and p = 0.21, respectively, Table 2). Interestingly, the ratio of patients with bowel cancer with at least one family member (first- and second-degree relatives) with family history of cancer(s), such as colon, breast, endometrium, ovary, and/or pancreas was significantly higher in the P/LP group than that of the Non-P group (p = 0.037, Table 2).

A total of 30 genes with P/PL germline variants were detected among 47 patients (Figure 1A). Besides, this study identified 5 out of 47 patients who carried MUTYH P/LP mutations (5/47), followed by APC (4/47) MLH1 (3/47), TP53(3/47), and ATM (3/47) (Supplementary Table S2, Figure 2A). Moreover, it was found that 8 patients carried LS related mutations (3 MLH1, 2 MSH2, 2 MSH6, and 1 EPCAM) and 39 carried non-LS mutations (Figure 1A). These findings were consistent with those reported in the previous studies (Latham et al., 2019). Notably, certain novel P/LP mutations were present, including those in FANCD2, RAD51D, BLM, CDH1, FLCN, MEN1, SDHB, and SLX4 were newly discovered in our cohort and have been rarely reported in the previous publications. The detailed distribution data and information on the germline mutations are presented in Supplementary Table S2. The functions of these genes with newly discovered P/LP mutations were mainly involved in DNA damage repair pathways (DDR-related genes were shown in Supplementary Table S3) and exerted a very broad impact. They included the following components: i) Those contributing to homologous recombination (HR), such as BRCA1, RAD51D, MRE11A, and RAD51B; ii) Those involved in Fanconi anemia (FA), such as FANCG, FANCA, SLX4, BLM, FANCD2, FANCL, and BRIP1; iii) Those involved in base excision repair (BER), such as MUTYH; iv) those involved in nucleotide excision repair (NER), such as ERCC2; v) those involved in mismatch repair (MMR), such as MLH1, MSH2, MSH3, MSH6, PMS1, and EPCAM; vi) those contributing to DNA sensor (DS), such as ATM and CHEK2 (Figure 1B).

Mutations in the DNA damage repair genes increase the risk of subsequent mutations and therefore confer high cancer susceptibility. Previous studies verified the association between alterations in 34 DDR genes that exhibited higher TMB levels in urothelial cancer and demonstrated that DDR inactivation was associated with higher levels of TMB (Wang et al., 2018). Therefore, the TMB of CRC patients harboring germline mutations was compared among different DDR subgroups. Although there were no statistically significant differences with regard to the incidence of TMB in these pathways, the average value of MMR was the highest with the exception of the group of others, while HR exhibited the lowest (Figure 1C). This phenomenon is in line with the previous studies showing that over 90% of MMR-deficient tumors exhibit high TMB levels (Stadler et al., 2016). In addition, the data indicated that the majority of germline mutations were located in or affecting protein functional domains and that they may have a significant impact on protein function (Supplementary Figure S1).

To investigate the risk of bowel cancer in individuals carrying P/LP germline mutations, the mutation prevalence of all germline mutations was searched in the total population and in different populations derived from the China_MAPs database (Table 3). Eight genes were significantly associated with CRC compared with control subjects derived from the China_MAPs database. These included the following: i) APC, with mutations in 0.70% of cases and in 0.01% of control subjects (OR, 74.32); ii) ATM, with mutations in 0.52% of cases and in 0.03% of control subjects (OR, 18.56); iii) MLH1, with mutations in 0.52% of cases and in 0.01% of control subjects (OR, 111.50); iv) FANCD2, with mutations in 0.35% of cases and in 0.03% of control subjects (OR, 12.35); v) MSH3, with mutations in 0.35% of cases and in 0.04% of control subjects (OR, 9.27); vi) MSH6, with mutations in 0.35% of cases and in 0.01% of control subjects (OR, 74.07); vii) PMS1, with mutations in 0.35% of cases and in 0.30% of control subjects (OR, 10.59); viii) RAD51D, with mutations in 0.35% of cases and in 0.01% of control subjects (OR, 74.07) (Table 3). These findings suggested that these P/LP germline mutations were risk factors for the development of bowel cancer.

Furthermore, the bowel cancer germline mutation frequency found in the present study was compared with that reported in other studies including the investigations performed in Japan (Fujita et al., 2020) and in America (Yurgelun et al., 2017). For the incidence of APC and MUTYH in the Japanese cohort were significantly lower than those noted in the present (p = 0.02, p < 0.001 respectively). The P/PL prevalence did not differ significantly of genes including MLH1, MSH2, MSH6, BRCA1, TP53, CHEK2, ATM, and BRIP1 among the three cohorts investigated. (Table 4).

The relationship between germline mutation carriers and patients with somatic mutations has been studied in other cancer types, such as lung cancer (Liu et al., 2020). However, the connection of germline variations and somatic mutations in bowel cancer has not been explored in detail. The somatic mutation spectrum was classified by pathogenicity/likely pathogenicity of germline mutations for all patients with bowel cancer (P/LP and Non-P groups) (Figures 2A,B). TP53, APC, KRAS, NOTCH1, ARID1A, FBXW7, PIK3CA, KMT2C, NOTCH3, and TCF7L2, were found to be the top 10 mutated genes in the P/LP group. With regard to the Non-P group, the top 10 mutated genes were TP53, APC, PIK3CA, SMAD4, NOTCH3, FAT1, ARID1A, KMT2D, RECQL4, and FBXW7. According to the different mutation classification categories, the missense mutation was the one that obtained the highest proportion in the bowel cancer samples in the presence or absence of germline variants.

To determine the presence of germline variants in the patients examined, the comparison of somatic alterations was conducted between P/LP and Non-P groups, and the results revealed dramatic differences (p < 0.001) in several gene mutations between these two groups (Figure 3A). A total of 17 gene mutations such as TCF7L2, KMT2D, PRKDC, NOTCH1, KMT2C, ERBB3, and TSC2, and others were more common in patients with bowel cancer with P/LP germline mutations; strikingly, SMAD4 exhibited the opposite prevalence and its mutation frequency in the Non-P group was significantly higher than that in the P/LP group (Figure 3A). Subsequently, Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment analyses were further conducted to explore the biological roles of the identified differential genes. GO as well as KEGG enrichment analysis revealed that these genes were mainly enriched in the biological process (BP) terms, such as PI3K signaling, DNA recombination, HR, and MMR (Supplementary Figure S2). Furthermore, the identification of specific driver genes in the present study has been previously reported (Eklöf et al., 2013; Dow et al., 2015). The mutation rate (frequency) of APC was the highest among all genes both in the P/LP and non-P groups followed by the mutation rate of TP53, KRAS, PIK3CA, SMAD4 and BRAF (Figure 3B).

TMB is considered a vital biomarker in a variety of cancer types (Samstein et al., 2019; Wu et al., 2019), which may reflect the degree of genomic instability at the nucleotide level. Therefore, the TMB was initially compared in the P/LP and Non-P groups and the data indicated a significantly higher level of TMB in the P/LP group (14.56 vs. 6.39 mutations/Mb, p = 0.0056, Figure 4). Furthermore, the DDR system plays an important role in maintaining genome stability, based on this notion, a focused analysis was performed, specifically on the DDR-altered genes in all bowel cancer cases. In the P/LP group, DDR alterations were present in 68.1% (n = 32) of cases and were involved in the DDR pathways including HR (31.36%), DS (20.82%), FA (19.28%), MMR (12.85%), BER (9%) and NER (6.68%), respectively (Figure 5A). In the Non-P group, DDR gene mutations occurred in 30% (n = 157) of cases, including HR (26.97%), MMR (20.22%), FA (20.22%), BER (13.48%), DS (11.24%) and NER (7.67%) (Figure 5B). In the P/LP group, FANCA, ATM, and MUTYH were the most commonly altered DDR genes, followed by BRCA2, MSH3, and POLE; whereas in the Non-P group, BRCA2 exhibited the highest DDR gene alterations, followed by ATM, POLE, and MSH6 (Figure 5C). Furthermore, the highest average TMB value was observed in DDR-altered Non-P cases with bowel cancer compared with that of DDR-altered P/LP cases pf bowel cancer or Non-DDR altered cases (p < 0.001, Figure 5D). The data demonstrated that tumors with DDR mutations in the Non-P group exhibited increased genomic instability than that of the P/LP group. The lowest average TMB appeared in the non-DDR group, which is in line with the evidence reported in the previous studies.

A total of 223 bowel cancer samples with somatic mutation profiles were downloaded from the TCGA database. The relevant clinical information was listed in Supplementary Table S4. DDR gene mutation analysis was also conducted in the TCGA cohort, and the FA pathway accounted for the highest proportion (25.22%), while NER accounted for the lowest proportion (7.42%) (Figure 6A). Moreover, a similar finding was obtained regarding specific gene mutations in these pathways between TCGA cohort and the Chinese cohort, such as ATM (the highest mutation frequency) and EPCAM (relative lower mutation frequency) (Figure 6B).

Subsequently, the TMB in the Chinese cohort was compared with the mutation count in the TCGA cohort (TMB was not accessible and therefore the mutation count was adopted here). The selection of mutation genes with P/LP was performed via the cBioportal website. In concordance with the previous results, a significantly higher level of TMB was observed in bowel cancer harboring DDR somatic mutations compared with cases with non-DDR bowel cancer in TCGA cohort (p < 0.0001, Figure 7). A similar finding was obtained from the Chinese cohort (Figure 5D). Especially, although there was no statistically significant difference among the DDR pathways examined in the TCGA and the Chinese cohort, the lowest TMB or mutation count was observed in NER pathway. In addition, the HR pathway displayed an increased mutation count in the TCGA cohort, while the Chinese cohort harbored the highest TMB of somatic mutations in BER pathway (Supplementary Figure S3).

Previous study has demonstrated that mutations can generate novel peptide sequences, which may affect the immune response (Chalmers et al., 2017). The TMB noted in the DDR mutation group was higher in the Chinese cohort and the TCGA cohort (Figure 5D, Figure 7). Thus, higher immune cell abundance was expected in the DDR somatic mutation. By applying the CIBERSORT algorithm, the differential variation of immune cell infiltration was estimated in the DDR and non-DDR groups of bowel cancer. The Wilcoxon rank-sum test indicated that the proportion of B cell naive (p = 0.017), T cell follicular helper (p = 0.0069), Macrophage M1 (p = 0.0038), and Neutrophils (p = 0.0163) were significantly elevated in the DDR group. By contrast, the infiltration levels of T cell regulatory Tregs (p = 0.0291) and Myeloid dendritic cell resting (p = 0.0163) were lower in the DDR group (Figure 8A). However, no association was observed between the different DDR pathway alterations and immune cell abundance (Figure 8B).

In addition, the present study further investigated whether the number of somatic mutations in the DDR genes of P/LP group was associated with improved survival following programmed cell death protein 1 (PD-1)/programmed death-ligand 1 (PD-L1) or cytotoxic T-lymphocyte-associated protein 4 immunotherapy in CRC patients. The clinical data of immunogenomic studies were downloaded from the cBioportal website. The overall survival (OS) was defined as the time from initial surgery to the date of death or last contact (censored). As expected, alteration in the mutation status of P/LP DDR conferred superior OS compared with patients without altered DDR gene(s) (HR, 0.3358; 95% CI, 0.1767 to 0.638; p = 0.0009) in the immunotherapy cohort (Figure 9A). However, patients with DDR mutations did not obtain a significantly prolonged OS when the MMR pathway was excluded (HR = 0.2998, 95%CI 0.1051 to 0.8556; p = 0.0549, Figure 9B).