Colorectal cancer (CRC) has one of the highest rates of aggregation within families (familial CRC), with up to 35% of CRC thought to be caused by inherited genetic risk factors [1]. However, the underlying cause of CRC can be assigned to one of the inherited CRC and polyposis syndromes in only 5–10% of cases [2], therefore, the genetic cause of the majority of familial CRC remains unknown. The term Familial Colorectal Cancer Type X (FCCTX) was proposed to define families with a strong CRC family history that meet the Amsterdam I criteria [3] where the tumors are DNA mismatch repair (MMR)-proficient/microsatellite stable and do not carry a germline pathogenic variant in one of the MMR genes (Lynch syndrome) [4, 5]. The genetic factors underlying FCCTX are poorly understood and are likely to be heterogeneous involving multiple susceptibility genes [6].

Serrated polyposis syndrome (SPS) is characterized by the presence of multiple serrated colorectal polyps (hyperplastic polyp, sessile serrated lesion (SSL) and traditional serrated adenoma) resulting in an increased risk of developing CRC [7,8,9]. The diagnostic criteria for SPS was re-defined by the World Health Organization (WHO) in 2019 [10] to include (i) 5 or more serrated polyps proximal to the rectum, all 5 mm or greater in size, with 2 or more 10 mm or greater in size or (ii) more than 20 serrated polyps of any size in the large bowel, with 5 or more proximal to the rectum. The progression from serrated polyp to carcinoma, referred to as the serrated neoplasia pathway of tumorigenesis, is characterized by distinct molecular features, including the presence of microsatellite instability (MSI), high levels of the CpG island methylator phenotype (CIMP) and somatic mutations in the oncogenes BRAF or KRAS [11]. However, the genetic etiology of SPS remains poorly understood [12, 13]. Recently, germline pathogenic variants in RNF43 have been proposed to underlie SPS [14,15,16,17,18], but they account for only a small proportion of cases [19]. As such, expert groups are yet to recommend the inclusion of RNF43 in multi-gene testing panels for patients with SPS [12].

The BRCA1 gene acts as a tumor suppressor through its role in DNA repair [20]. Germline pathogenic variants in BRCA1 confer high risks of breast and ovarian cancers [21]. The association between BRCA1 pathogenic variants and CRC development is more uncertain [22]. Multiple studies have investigated whether carriers of germline BRCA1 pathogenic variants have an increased risk of developing CRC, with mixed results [23, 24].

It has been suggested that digenic inheritance may account for some cases of familial CRC and polyposis syndromes, however there are few reports in the literature [25,26,27]. In this study, we describe a family meeting FCCTX criteria where multiple cancer-affected individuals carried germline pathogenic variants in both the BRCA1 and RNF43 genes on chromosome 17q. The tumor characteristics from carriers were assessed to characterize the drivers of tumorigenesis. Our findings demonstrate a possible role for digenic inheritance in the predisposition to familial CRC.


Study cohort

The family presented was identified from the Australasian Colorectal Cancer Family Registry (ACCFR) (HREC:13,094) [28,29,30]. The ACCFR recruited multiple-member CRC-affected families from Family Cancer Clinics across Australia and New Zealand between 1998 and 2008. Participants provided written consent to access their tumor tissue and provided a blood sample [30]. Methodology for germline MMR and MUTYH gene testing and tumor characterization have been described previously [28].

Germline sequencing and variant detection

CRC-affected individuals 009 and 014 had germline whole exome sequencing (WES) performed. Briefly, 50ng of genomic DNA was fragmented to an average size of 180 bp in length using a Covaris focused-ultrasonicator (Covaris, Woburn, MA, USA). An Illumina sequencing technology compatible whole genome library was created using Kapa Biosystems Hyper Prep Kits (Kapa Biosystems Inc., Wilmington, MA, USA). These libraries were then subjected to whole exome target enrichment using Agilent SureSelect hybrid capture version 4 kits (Agilent Technologies, Santa Clara, CA, USA). Parallel sequencing of libraries was performed on Illumina HiSeq2000/2500 system using version 1.5 or version 3 chemistry using paired-end 2 × 100 bp reads (Illumina, San Diego, CA, USA). All sequencing reads were converted to industry standard FASTQ files using BCL2FASTQ v1.8.4. FASTQ files were processed using a pipeline based on industry standard software packages and programs. Sequencing reads were aligned to the GRCh37 human genome reference using v0.7.8 BWA-MEM aligner [31] to generate BAM files. SAMtools v0.1.19 [32] was used to sort BAM files and Picard v1.111 ( to mark duplicate read pairs. Post alignment joint insertion/deletion (indel) realignment and base quality scores recalibration was performed on the BAM files using GATK v3.1-1 [33]. Variants were called from germline BAM files individually using GATK Haplotype Caller v3.1-1 [34] and SAMtools v0.1.19 [35].

Germline variant annotation

Germline variants were annotated with the Ensembl Variant Effect Predictor release 105 (December 2021) for the human genome reference GRCh37 including the CADD predicted pathogenicity scores for each variant [36, 37]. The RefSeq transcript NM_007300.4 was used for BRCA1. The RefSeq transcript NM_017763.5 was used for RNF43. Sanger sequencing of the BRCA1 and RNF43 pathogenic variants was used for confirmation of the variants in persons 009 and 014 and to segregate the variants in 19 other family members with available DNA.

Tumor tissue sample processing and nucleic acid preparation

Where available, formalin-fixed paraffin-embedded (FFPE) tumor tissue blocks were obtained. MMR status was determined with immunohistochemistry as previously described [28]. Sections were stained with haematoxylin and eosin and prepared for pathological review. Tumor, polyp and histologically normal mucosa were macrodissected and processed independently. DNA was extracted with the QIAamp DNA FFPE Tissue kit following standard protocols (Qiagen, Hilden, Germany).

Tumor tissue sequencing and variant detection

CRC tumor tissue and matched blood-derived DNA from person 009 were prepared according to the procedure for Hybridization Capture using the Agilent SureSelectXT Low Input Clinical Research Exome v2 kit. The prepared libraries were sequenced with Illumina sequencing technology comprising 150 bp paired. Raw FASTQ files underwent adapter sequence trimming using trimmomatic v.0.38 [38] and alignment to the human genome reference GRCh37 using BWA v.0.7.12 [31]. Duplicate reads were identified with Picard v2.8.2. Mean on target coverage for the tumor and buffy coat samples was 499.3 and 79.5 respectively. Germline variants were called with HaplotypeCaller from GATK 4.0.0 [39] using GATK’s recommended workflow. Somatic single-nucleotide variants (SNVs) and short insertions and deletions (indels) were called with Mutect2 [40] with the recommended GATK practices and Strelka v.2.9.2 [41] with Illumina’s recommended workflow. Mutations reported by both callers were filtered to PASS variants with a minimum variant allele frequency of 0.1 and minimum depth of 50 reads.

Tumor loss of heterozygosity analysis

Two CRC and two polyp tissue DNA samples from two carriers of both variants were assessed for loss of heterozygosity (LOH) of the wildtype alleles of the BRCA1 and RNF43 variants using standard Sanger sequencing protocols. Short (179 bp) BRCA1 amplicons were generated using GCAGAAGAGGAATGTGCAACATTCT and TTATCTTTCTGACCAACCACAGGAA with sequencing occurring in the reverse direction. Short (182 bp) RNF43 amplicons were generated using ACAGGCTACTCAGGGTCAAATAGAT and CGAATGAGGTGGAGTCTTCGA with sequencing occurring in the forward direction. Tumor tissue DNA was available for a single CRC from person 009 for extended LOH assessment using WES tumor data. The captured regions of the genome were assessed for evidence of LOH by interrogating heterozygous germline variants in the tumor for their presence as homozygous reference or homozygous alternative in the tumor tissue. A tumor cellularity estimate of 80% was used. Germline variants with an allele frequency of between 0.4 and 0.6 were considered heterozygous. An allele frequency difference of 0.3 or greater in the somatic tissue, limited to variants with a germline depth ≥ 10 and tumor depth ≥ 30, was considered evidence of LOH. Individual variants suggesting the presence of LOH were aggregated to determine likely genomic regions of LOH. The algorithm used is available at

Tumor mutational signature analysis

SNVs and indels were filtered to those in the capture region. These filtered SNVs and indels were used to calculate tumor mutational signatures according to the method given by SignatureEstimation [42] from the set of COSMIC version 3.2 signatures [43] limited to signatures observed in CRC tissue comprising 15 single base substitution (SBS) signatures and 5 indel (ID) signatures [44] as commonly recommended [45], including SBS3 and ID6 given their association with BRCA1 mutations. SBS3 or ID6 present at > 10% or > 20% proportion in the tumor signature profile, respectively, was considered positive for defective homologous recombination-based DNA damage repair (HRD).


Two germline pathogenic variants were identified; one in BRCA1:c.2681_2682delAA, a frameshift pathogenic variant located in exon 10 encoding p.Lys894ThrfsTer8, and another in RNF43:c.988 C>T, a nonsense pathogenic variant located in exon 9 encoding p.Arg330Ter, in one family meeting the FCCTX criteria. The family pedigree with cancer-affected and carrier status is shown in Fig. 1. No other loss of function or predicted pathogenic variants were identified in established hereditary CRC and polyposis genes. Ten individuals carried the BRCA1:c.2681_2682delAA variant and eight individuals carried the RNF43:c.988 C>T variant. Seven individuals carried both pathogenic variants of whom six were cancer-affected (4 CRC, 1 breast/ovarian cancer, 1 metastatic cancer of unknown primary). All four of the CRC-affected relatives tested carried both pathogenic variants. Only a single carrier of both variants was cancer-unaffected at age 58 (person 026). Where both variants were tested, two individuals were found to carry only a single variant, person 018 carried only the BRCA1 variant and person 028 carried only the RNF43 variant, where each likely represents a separate homologous recombination event on chromosome arm 17q. Details of carrier status and their tumors are provided in Table 1.

Fig. 1
figure 1

Pedigree diagram for a family with colorectal cancer, serrated polyposis syndrome and BRCA1:c.2681_2682delAA and RNF43:c.988 C>T germline pathogenic variants. The indicated carriers include obligate carriers

Table 1 Cancer and colonic polyp history together with the carrier status of the BRCA1:c.2681_2682delAA and RNF43:c.988 C>T germline pathogenic variants in people from a family meeting FCCTX criteria

The proband (person 001), a carrier of both the BRCA1 and RNF43 variants, was diagnosed with an adenocarcinoma of the caecum at age 53, a peritoneal cancer at age 62 and an ovarian cancer at age 63. MMR immunohistochemistry (IHC) of the metastatic lymph nodes indicated the CRC tumor was MMR-proficient. Three colonoscopies performed between the ages of 52 and 62 identified “numerous small metaplastic polyps” although the number and specific pathology were not reported, and, therefore, unclear if this person met the criteria for SPS. She died at age 67.

Two of the proband’s daughters (009 and 010) carried both the BRCA1 and RNF43 variants and both were CRC-affected. Person 009 was diagnosed with an MMR-proficient adenocarcinoma of the transverse colon at age 44. There was no report of synchronous polyps. Person 010 was diagnosed with a 15 mm moderately differentiated adenocarcinoma of the sigmoid colon at age 56, which appeared to have arisen from an SSL. Nine colonoscopy procedures between the ages of 40 and 63 revealed multiple serrated and adenomatous polyps. At the age of 56, a colonoscopy revealed a 10 mm hyperplastic polyp in the transverse colon in addition to the CRC. At the age of 59, a repeat colonoscopy showed a 6–8 mm adenomatous polyp and a 6–8 mm SSL in the ascending colon and two 6–8 mm hyperplastic polyps in the left colon. At the age of 62, a further colonoscopy showed a 5–8 mm hyperplastic polyp in the rectum. Including the SSL from which the adenocarcinoma had arisen from, person 010 met the 2019 WHO diagnostic criterion 1 for SPS [10].

Person 014 (a brother of the proband) was a carrier of the BRCA1 and RNF43 variants. He was diagnosed with an MMR-proficient adenocarcinoma of the transverse colon at age 56 and a prostate cancer at age 71. Person 018 (another brother of the proband) was a carrier of the BRCA1 variant but not the RNF43 variant. He was diagnosed with pharyngeal squamous cell carcinoma at age 57 and prostate adenocarcinoma at age 58. Two of the BRCA1 carriers developed breast cancer (persons 025 and 056), one of whom was diagnosed at 34 years of age; the subtype was unavailable.

Tumor analysis

The CRCs from persons 009 and 014 were both MMR-proficient by IHC, wildtype for BRAF p.V600 and KRAS codon 12 and 13 somatic mutations and were CIMP-negative (Table 2) suggesting they had not developed via the serrated pathway of tumorigenesis. The MMR-proficient CRC and contiguous SSL from person 010 were both BRAF p.V600E mutation positive and CIMP-high, consistent with development via the serrated pathway (Table 2). Sanger sequencing of the BRCA1 and RNF43 variants in the tubular adenoma, SSL and CRC from person 010 showed evidence of LOH of the wildtype allele for both variants in the SSL and adenocarcinoma but not the tubular adenoma (Table 2; Fig. 2).

Table 2 Molecular characteristics of tumors from the family co-segregating the BRCA1:c.2681_2682delAA and RNF43:c.988 C>T germline pathogenic variants
Fig. 2
figure 2

Sanger sequencing of the BRCA1:c.2681_2682delAA (left column) and RNF43:c.988 C>T (right column) pathogenic variants in a tubular adenoma, sessile serrated lesion (SSL), and colorectal cancer (CRC) for person 010 showing loss of heterozygosity (LOH) of the wildtype allele for both variants in the sessile serrated lesion and CRC but not the tubular adenoma

To further investigate tumor etiology, the CRC from person 009 underwent WES. No somatic mutations in BRCA1 and RNF43 were observed, however, loss of the wildtype allele was evident for both variants. LOH of a larger region across chromosome arm 17q was detected that included the BRCA1 and RNF43 genes (Fig. 3). Analysis of COSMIC tumor mutational signature profiles revealed SBS3 (61.8%), SBS1 (11.3%) and SBS30 (8.9%) as the SNV-derived signatures with the highest proportion. The observed indels in this tumor were decomposed into the signatures ID6 (65.2%), ID5 (30.5%) and ID1 (4.3%), with the predominance of both SBS3 and ID6 indicative of defective homologous recombination-based DNA damage repair (HRD); the contexts of SBS3 and ID6 are compared to those observed in 009 in Fig. 4.

Fig. 3
figure 3

Allele frequency plot for a colorectal tumor of a person (person 009) with BRCA1:c.2681_2682delAA and RNF43:c.988 C>T germline pathogenic variants showing loss of heterozygosity across chromosome 17, including BRCA1 and RNF43

The top plot covers the whole of chromosome 17. The middle plot covers a region around BRCA1. The bottom plot covers a region around RNF43.

Fig. 4
figure 4

Comparing SNV-derived mutational contexts of a person with BRCA1:c.2681_2682delAA and RNF43:c.988 C > T germline pathogenic variants (person 009) (A) with defective homologous recombination-based DNA damage repair associated signature SBS3 (B), and similarly, indel-derived contexts of person 009 (C) with ID6 (D)


This study identifies a family meeting the criteria for FCCTX where a germline BRCA1:c.2681_2682delAA p.Lys894ThrfsTer8 pathogenic variant and a germline RNF43:c.988 C>T p.Arg330Ter pathogenic variant co-segregated with CRC in four carriers, one of whom was confirmed to meet the WHO2019 diagnostic criteria 1 for SPS. Tumor analysis demonstrated loss of the wildtype allele for both variants in the two CRCs tested. As both BRCA1 and RNF43 reside on chromosome 17q, the LOH observed across the region encompassing both these tumor suppressor genes (Fig. 3) confirms that both BRCA1 and RNF43 had biallelic inactivation. The presence of both the tumor mutational signatures SBS3 and ID6 at high levels (> 50%), which is associated with HRD, and the absence of serrated pathway molecular characteristics, namely the BRAF p.V600E mutation and CIMP-high, suggests that tumorigenesis for the CRC from person 009 was driven by HRD deficiency related to BRCA1 inactivation. In contrast, biallelic inactivation of BRCA1 and RNF43 was also present in the CRC from person 010 with the tumor demonstrating characteristics of the serrated pathway (BRAF p.V600E mutation and high levels of CIMP), suggesting that for this tumour tumorigenesis may have been driven by RNF43 deficiency.

Germline pathogenic variants in BRCA1 predispose carriers to significantly elevated risks of breast and ovarian cancers [46], but the relationship between BRCA1 and CRC susceptibility is less clear [22]. A recent meta-analysis and systematic review showed BRCA1 and/or BRCA2 pathogenic variant carriers did not have a higher risk of developing CRC [47]. Past studies have suggested BRCA2 may underlie CRC development in FCCTX families, however, there is little evidence implicating BRCA1 [48, 49]. In the current study, ten family members carried the BRCA1 variant, four developing CRC with only three developing a breast or ovarian cancer. Tumor WES derived analysis from the single CRC from person 009 demonstrated that tumorigenesis was dominated by the BRCA1 variant-related HRD process, evidenced by the high proportion of HRD-related SBS3 and ID6 mutational signatures. Despite this, it is possible that RNF43 deficiency has also contributed to the initiation and/or progression of tumorigenesis in this person together with HRD.

Somatic mutations in RNF43 play a role in colorectal tumorigenesis including in the serrated pathway [50,51,52,53,54]. Furthermore, although rare in SPS [19], several studies have now provided evidence that germline RNF43 variants are associated with SPS [14,15,16,17,18,19, 53]. Only a few of these studies have investigated segregation of the RNF43 variant with SPS in the family. Of note, a study by Taupin et al. [15] identified a germline nonsense variant in RNF43 (c.394 C>T p.Arg132Ter) in two siblings affected with SPS, one developed CRC and a study by Yan et al. [17] identified a germline splice site variant (c.953-1 G>A) in RNF43 carried by six people from the family. Five of the six carriers met the WHO2010 criteria for SPS with a second somatic hit in RNF43 (predominantly LOH) identified in all 22 cancers/polyps analyzed [17]. There were eight carriers of the RNF43:c.988 C>T p.Arg330Ter variant in the family from this study, four were CRC-affected and a single carrier was confirmed to meet the WHO2019 criteria for SPS. Furthermore, LOH was observed as the second somatic hit in both CRCs tested and in an SSL polyp. [16, 17] Our findings add further support for the association between germline RNF43 variants and susceptibility to SPS and CRC.

Tumor mutational signature analysis is an important tool for understanding tumor etiology and for predicting response to cancer therapies, including the use of PARP inhibitors for cancers with HRD [55]. Of the current COSMIC mutational signatures, SBS3 and ID6 are associated with HRD, which are associated with defects in BRCA1, BRCA2 or other genes involved in the homologous recombination pathway [55, 56], although HRD in CRC is not commonly observed [57]. In the CRC from person 009, both SBS3 and ID6 were the dominant mutational signatures, supporting HRD related to the germline BRCA1 variant.

This study has several limitations. Phenotype data was not available from all family members including incomplete or historic colonoscopy and/or pathology reports that meant some of the colonic polyp number and morphological classification was not definitive or equivalent to contemporary polyp classification. Little data was obtained from earlier generations as those generations were deceased prior to commencing a detailed investigation. Furthermore, the tumor tissue for molecular testing was limited with only a single CRC with sufficient DNA for WES and therefore, confirmation that HRD associated mutational signatures were the dominant mutational process in the other CRCs from BRCA1 carriers could not be determined. Further investigation of HRD in CRC tumorigenesis is needed.


In summary, we have identified coinheritance of pathogenic germline variants in BRCA1 and RNF43 segregating with CRC in a family previously characterized as FCCTX. One individual satisfied the diagnostic criteria for SPS, and there was evidence for a somatic second-hit in BRCA1 and RNF43 in the form of LOH. Bioinformatic analysis showed that the tumorigenesis was predominantly driven by the BRCA1 variant with LOH, as indicated by the HRD-related mutational signatures in the tumor. Our study highlights a possible role of digenic inheritance underlying FCCTX.