Association Between Loss-of-Function Mutations Within the FANCM Gene and Early-Onset Familial Breast Cancer

Key Points

Question

Do inactivating germline mutations within the FANCM gene increase breast cancer and/or ovarian cancer risk?

Findings

This case-control study included 2047 BRCA1 and BRCA2–negative familial breast cancer cases and 2187 controls and revealed an association of FANCM mutations with breast cancer. More pronounced associations were identified for early-onset (before age 51 years) breast cancer and triple-negative breast cancer. Analysis of 628 unselected ovarian cancer cases revealed no significant association.

Meaning

We suggest FANCM be included in diagnostic gene panel testing for individual breast cancer risk assessment.

Abstract

Importance

Germline mutations in established moderately or highly penetrant risk genes for breast cancer (BC) and/or ovarian cancer (OC), including BRCA1 and BRCA2, explain fewer than half of all familial BC and/or OC cases. Based on the genotyping of 2 loss-of-function (LoF) variants c.5101C>T (p.GIn1701Ter [rs147021911]) and c.5791C>T (p.Arg1931Ter [rs144567652]), the FANCM gene has been suggested as a novel BC predisposition gene, while the analysis of the entire coding region of the FANCM gene in familial index cases and geographically matched controls is pending.

Objectives

To assess the mutational spectrum within the FANCM gene, and to determine a potential association of LoF germline mutations within the FANCM gene with BC and/or OC risk.

Design, Setting, and Participants

For the purpose of identification and characterization of novel BC and/or OC predisposition genes, a total of 2047 well-characterized familial BC index cases, 628 OC cases, and 2187 geographically matched controls were screened for LoF mutations within the FANCM gene by next-generation sequencing. All patients previously tested negative for pathogenic BRCA1 and BRCA2 mutations. All data collection occurred between June 1, 2013, and April 30, 2016. Data analysis was performed from May 1, 2016, to July 1, 2016.

Main Outcomes and Measures

FANCM LoF mutation frequencies in patients with BC and/or OC were compared with the FANCM LoF mutation frequencies in geographically matched controls by univariate logistic regression. Positive associations were stratified by age at onset and cancer family history.

Results

In this case-control study, 2047 well-characterized familial female BC index cases, 628 OC cases, and 2187 geographically matched controls were screened for truncating FANCM alterations. Heterozygous LoF mutations within the FANCM gene were significantly associated with familial BC risk, with an overall odds ratio (OR) of 2.05 (95% CI, 0.94-4.54; P = .049) and a mutation frequency of 1.03% in index cases. In familial patients whose BC onset was before age 51 years, an elevated OR of 2.44 (95% CI, 1.08-5.59; P = .02) was observed. A more pronounced association was identified for patients with a triple-negative BC tumor phenotype (OR, 3.75; 95% CI, 1.00-12.85; P = .02). No significant association was detected for unselected OC cases (OR, 1.74; 95% CI, 0.57-5.08; P = .27).

Conclusions and Relevance

Based on the significant associations of heterozygous LoF mutations with early-onset or triple-negative BC, FANCM should be included in diagnostic gene panel testing for individual risk assessment. Larger studies are required to determine age-dependent disease risks for BC and to assess a potential role of FANCM mutations in OC pathogenesis.

This case-control study analyzed the entire coding region of the FANCM gene to find its association with breast and ovarian cancer risk.

Introduction

Monoallelic mutations within several Fanconi anemia complementation group genes, including FANCD1/BRCA2, FANCJ/BRIP1, FANCO/RAD51C, and FANCN/PALB2, confer a moderate to high risk for breast cancer (BC) and/or ovarian cancer (OC), while the role of other Fanconi anemia–associated genes in BC and/or OC pathogenesis remains elusive. Among those genes, FANCM (OMIM 609644) is a plausible candidate because the FANCM protein and its binding partner FAAP24 (OMIM 610884) are essentially required to anchor the multi-subunit Fanconi anemia core complex to chromatin after DNA damage. So far, 2 truncating germline variants within the FANCM gene, c.5101C>T (p.GIn1701Ter [rs147021911]) and c.5791C>T (p.Arg1931Ter [rs144567652]), have been associated with an increased BC risk.

A study in the Finnish population revealed the nonsense variant c.5101C>T to be associated with familial BC. With a carrier frequency (CF) of 1.83% (38 in 2080) in female control individuals, this variant appears to be particularly frequent in the Finnish population. Overall, the c.5101C>T mutation was found in 3.39% (45 in 1327) of BRCA1 (OMIM 113705) and BRCA2 (OMIM 600185)–negative familial BC cases (odds ratio [OR], 2.11; 95% CI, 1.34-3.32; P = .001) with an even higher CF of 5.88% (12 in 204) observed in a subgroup of mainly unselected cases with a triple-negative (TN) BC tumor phenotype (OR, 3.56; 95% CI, 1.81-6.98; P < .001). No significant correlation was established in 548 unselected OC cases (OR, 1.56; 95% CI, 0.75-3.26; P = .23). A multinational study investigating c.5791C>T, the second variant shown to be associated with BC, in 8635 BRCA1 and BRCA2–negative familial BC cases and 6625 control individuals confirmed an association between truncating FANCM mutations and BC risk, with a CF of 0.21% (18 in 8635) in cases and 0.06% (4 in 6625) in controls and an OR of 3.93 (95% CI, 1.28-12.11; P = .02). The c.5791C>T variant has shown to create an exonic splicing silencer, resulting in incomplete exon 22 skipping and protein truncation (p.Gly1906Alafs12Ter). Thus, the c.5791C>T nonsense variant is only barely detectable on transcript level.

Because all studies described to date rely on only 2 truncating germline variants, mutational analysis of the entire coding region of the FANCM gene in patients and geographically matched controls is pending. Moreover, a higher cumulative number of patients known to carry truncating FANCM mutations is required to establish genotype-phenotype correlations, especially regarding the TNBC tumor phenotype and cancer site.

Methods

In the course of a whole exome sequencing approach, we identified the c.5101C>T germline mutation in a BRCA1 and BRCA2–negative familial BC index case from a high-risk family of German origin, which prompted us to analyze the FANCM gene in detail. The mutation was also present in the affected mother and maternal aunt, both affected by BC (eFigure in the Supplement). Based on the whole exome sequencing data provided by the Exome Aggregation Consortium, 0.65% of more than 22000 individuals of non-Finnish European origin carried heterozygous loss-of-function (LoF) mutations within the FANCM gene (excluding The Cancer Genome Atlas data; eTable 1 in the Supplement). To assess the cumulative CF of FANCM in the German population, we analyzed the whole exome sequencing data of a cohort of 2187 independent individuals not enriched for cancer phenotypes. Concordant with the data provided by the Exome Aggregation Consortium, 11 in 2187 controls carried heterozygous germline LoF mutations, resulting in a cumulative CF of 0.50% (eTable 1 in the Supplement).

Between May 1, 2016, and July 1, 2016, we analyzed the entire coding region of the FANCM gene in a well-characterized cohort of 2047 BRCA1 and BRCA2–negative familial index patients with BC of German origin by focusing on LoF alterations. Family history was considered positive when the inclusion criteria of the German Consortium for Hereditary Breast and Ovarian Cancer for genetic germline testing are fulfilled (eTable 2 in the Supplement).

Written informed consent was obtained from all patients, and ethical approval was given by the Ethics Committee of the University of Cologne.

Results

Overall, 21 in 2047 index cases carried heterozygous truncating FANCM mutations, resulting in a cumulative CF of 1.03% (OR, 2.05; 95% CI, 0.94-4.54; P = .049), as compared with the whole exome sequencing data from German controls (Table). A total of 1547 (76%) in 2047 patients were affected by BC and reported no personal or family history of OC. Among these 1547, we identified truncating mutations in 16 patients, resulting in a cumulative mutation frequency of 1.03%. Of the remaining 500 patients with BC who had either a personal history (62 [12.4%]) or family history (438 [87.6%]) of OC, 5 (1.00%) carried truncating FANCM alterations. In addition to the 62 familial index cases diagnosed with BC and OC, we subsequently analyzed 628 BRCA1 and BRCA2–negative OC cases not selected for family history. Comparable with the data reported by Kiiski et al, a marginally increased cumulative CF of 0.870% (6 in 690) was observed in OC cases vs controls (OR, 1.74; 95% CI, 0.57-5.08; P = .27), although the differences did not reach statistical significance. In summary, this first investigation of the entire coding sequence of the FANCM gene revealed a weak but significant overall association with familial BC.

Table. Frequencies of Heterozygous Loss-of-Function Germline Mutations Within the FANCM Gene in Cases and Controls.

Cohort	Total	Neg, No.	Pos, No. (%)	OR (95% CI)a	P Valueb
Geographically matched controls (Germany)	2187	2176	11 (0.504)
All familial BC index cases, BC or BC/OC family history	2047	2026	21 (1.026)	2.05 (0.94-4.54)	.049
Familial BC index cases
Age <51 y, BC or BC/OC family history	1393	1376	17 (1.220)	2.44 (1.08-5.59)	.02
Age ≥51 y, BC or BC/OC family history	653	649	4 (.613)	1.22 (0.33-4.15)	.73
BC only family history	1547	1531	16 (1.034)	2.07 (0.91-4.77)	.06
BC/OC family history	500	495	5 (1.000)	2.00 (0.60-6.24)	.19
Familial cases with TNBC	215	211	4 (1.860)	3.75 (1.00-12.85)	.02
Patients with OC, mainly unselected	690	684	6 (.870)	1.74 (0.57-5.08)	.27

Abbreviations: BC, breast cancer; Neg, negative; OC, ovarian cancer; OR, odds ratio; Pos, positive; TNBC, triple-negative breast cancer.

Frequencies of heterozygous loss-of-function germline mutations within the FANCM gene in familial BC index cases are according to family history and age at first onset, familial cases with TNBC tumor phenotype, mainly unselected OC cases, and geographically matched controls.

^aUnivariate logistic regressions were performed to estimate OR and 95% CI.

^bPearson χ² test.

When we stratified the 2047 familial BC index cases for age at onset (AAO), 1393 (68%) cases had an AAO before 51 years. In this subgroup (mean AAO, 41 years), 17 mutation carriers were identified, resulting in a cumulative CF of 1.22% (17 in 1393) (OR, 2.44; 95% CI, 1.08-5.59; P = .02). In patients with BC whose AAO was 51 years or older (mean AAO, 59 years), the cumulative CF was 0.613% (4 in 653), which is similar to that observed in controls. Based on genotyping of 204 mainly unselected TNBC cases, Kiiski et al reported a significant association between the FANCM nonsense variant c.5101C>T with the TNBC tumor phenotype in the Finnish population (OR, 3.56; 95% CI, 1.81-6.98; P < .001). In our cohort of 2047 familial BC index cases, a TNBC tumor phenotype was reported in 215 (10.50%) cases, of which 4 carried a heterozygous FANCM mutation (CF, 1.86% [4 in 215]; OR, 3.75; 95% CI, 1.00-12.85; P = .02). Thus, our results confirm the association of heterozygous LoF mutations within the FANCM gene with the TNBC tumor phenotype and establish an association with an early-onset BC disease in familial index cases. In total, we identified 21 index patients with BC who carried heterozygous LoF mutations within the FANCM gene. In this cohort of 21 mutation carriers, we observed a mean AAO of BC as 46 years (age range, 32-65 years). One mutation carrier also developed serous OC at age 62 years. Most BC cases are of no special type and classified as grade 2 (eTable 3 in the Supplement). Three additional affected mutation carriers were identified by segregation analysis in 2 families (eTable 3 and eFigure in the Supplement).

Discussion

Screening the entire coding sequence of the FANCM gene revealed its weak but overall significant association with familial BC. The study established an association between FANCM and early-onset BC disease in familial index cases. In addition, it confirmed an association of heterozygous LoF mutations within the FANCM gene with the TNBC tumor phenotype. However, this study focused on familial cases, which may lead to a significant selection bias in determining mutation frequency. For other moderately penetrant risk genes, such as CHEK2, it has been demonstrated that ORs are higher in familial than in unselected BC cases. Thus, it might be worthwhile to analyze unselected BC cases and then stratify for early-onset BC or TNBC. Moreover, additional studies will be required in larger sample sets to establish a potential role of FANCM mutations in OC pathogenesis. Nevertheless, based on the present analyses and previously published findings, we recommend including FANCM in diagnostic gene panel testing.

Supplement.

eFigure. Segregation Analysis of the FANCM c.5101C>T (p.GIn1701Ter) Mutation in Independent Families

eTable 1. Loss-of-Function Mutations Identified Within the FANCM Gene

eTable 2. Definition of Positive Family History

eTable 3. Genotypes and Phenotypes of Heterozygous FANCM Mutation Carriers