Rates of Actionable Genetic Findings in Individuals with Colorectal Cancer or Polyps Ascertained from a Community Medical Setting

Abstract

As clinical testing for Mendelian causes of colorectal cancer (CRC) is largely driven by recognition of family history and early age of onset, the rates of such findings among individuals with prevalent CRC not recognized to have these features is largely unknown. We evaluated actionable genomic findings in community-based participants ascertained by three phenotypes: (1) CRC, (2) one or more adenomatous colon polyps, and (3) control participants over age 59 years without CRC or colon polyps. These participants underwent sequencing for a panel of genes that included colorectal cancer/polyp (CRC/P)-associated and actionable incidental findings genes. Those with CRC had a 3.8% rate of positive results (pathogenic or likely pathogenic) for a CRC-associated gene variant, despite generally being older at CRC onset (mean 72 years). Those ascertained for polyps had a 0.8% positive rate and those with no CRC/P had a positive rate of 0.2%. Though incidental finding rates unrelated to colon cancer were similar for all groups, our positive rate for cardiovascular findings exceeds disease prevalence, suggesting that variant interpretation challenges or low penetrance in these genes. The rate of HFE c.845G>A (p.Cys282Tyr) homozygotes in the CRC group reinforces a previously reported, but relatively unexplored, association between hemochromatosis and CRC. These results in a general clinical population suggest that current testing strategies could be improved in order to better detect Mendelian CRC-associated conditions. These data also underscore the need for additional functional and familial evidence to clarify the pathogenicity and penetrance of variants deemed pathogenic or likely pathogenic, particularly among the actionable genes associated with cardiovascular disease.

Introduction

Colorectal cancer (CRC) affects 4.3% of the US population and is a leading cause of cancer-related death in the US, second only to lung cancer (see Web Resources). Some 5% of CRC is attributable to known Mendelian genetic conditions with high penetrance.¹ Clinical testing for genetic etiology is focused on those with early age of onset and a positive family history.^{2, 3} Rates of positive genetic results in individuals with CRC without these red flags for hereditary cancer have not been well reported.

Additionally, most studies of the rate of incidental findings (IFs) for clinically important genes have been done in convenience samples, often without the availability of phenotype data.^{4, 5, 6, 7} It is well known that ascertainment will influence both the rate of positive findings and the derived penetrance estimates. As genomic screening for high penetrance gene-disease pairs is proposed and implemented, understanding the rate of positive tests is important for informed consent and planning adequate resources for follow-up. This is particularly true for the gene-disease pairs designated as “actionable” by the American College of Medical Genetics and Genomics (ACMG).⁸

With the goal of evaluating the influence of ascertainment on the rate of positive (pathogenic or likely pathogenic) CRC/P-associated and non-CRC IF genes in a community setting, we evaluated participants ascertained through three strategies: those with (1) current or past diagnosis of CRC (regardless of polyp history), (2) history of one or more adenomatous colon polyps without CRC, and (3) neither of those (control subjects).

Subjects and Methods

Identification of Participants

Participants were drawn from the Northwest Institute for Genomic Medicine biorepository (NWIGM). The NWIGM biorepository is co-managed by the Kaiser Permanente Washington Health Research Institute (KPWHRI) and the University of Washington (UW). The biorepository consists of DNA samples from Kaiser Permanente Washington members (KPWA, formerly Group Health Cooperative) linked with extensive longitudinal electronic health record (EHR) data to enable high-resolution phenotype analysis. The NWIGM biorepository includes DNA samples for 4,478 current and former KPWA members recruited between 2008 and 2017. Informed consent was obtained from all participants under the guidance of the KPW institutional review board.

Among these participants, we identified 1,165 KPWA members with a history of CRC or colon polyps but no known prior genetic CRC testing, as documented in their EHR and the Washington state cancer registry (WSCR). CRC-affected case subjects were recruited from all individuals receiving healthcare at KPWA identified by the WSCR as having CRC, and those with both a diagnosis code of CRC in their EHR (International Classification of Disease, 9^th Revision codes, ICD 9, 153.^∗, 154.0, or 230.7) and evidence of a surgical or radiation treatment for CRC within 1 year of any such diagnosis code. Polyps-affected case subjects were identified by ICD 9 codes 211.2, 211.3, V12.72. As these results did not distinguish the number or type of polyps; a single polyp could lead to inclusion in this group. Participants with both CRC and polyps were placed in the CRC group and not the polyps group. Participants with a diagnosis of Crohn disease or ulcerative colitis, pregnant women, as well as individuals undergoing chemotherapy treatment within 2 weeks of recruitment were excluded from the study. CRC or polyp diagnosis was confirmed by manual chart review for participants with a positive genetic result; two misclassified participants were removed from analyses. Control subjects were selected at random from the NWIGM biobank after excluding individuals who had CRC or polyps diagnoses as defined above and were aged less than 59 years, and who matched the CRC group as best as possible by mean age, sex, and ethnicity.

Sequencing and Interpretation

All sequencing and variant pathogenicity interpretation was performed in a CLIA-certified environment. CRC and colon polyp participants were sequenced on the custom eMERGEseq panel which was designed in 2015 to contain coding region sequences for the 56 genes recommended for IF return by ACMG at that time as well as 53 other genes of interest to the eMERGE consortium. Details of the eMERGEseq panel and reporting pipeline, including variant interpretation, strategy, and coverage, has been previously reported.⁹ CRC/P-associated genes among these included APC (MIM: 611731), MLH1 (MIM: 120436), MSH2 (MIM: 609309), MSH6 (MIM: 600678), MUTYH (MIM: 604933), PMS2 (MIM: 600259), PTEN (MIM: 601728), POLD1 (MIM: 174761), POLE (MIM: 174762), SMAD4 600993), BMPR1A (601299), STK11 (MIM: 602216), TP53 (MIM: 191170), and CHEK2 (MIM: 604373). In addition to the original 56 ACMG genes, which includes 48 non-CRC actionable gene-disease pairs, the loci (PALB2 (MIM: 610355), CACNA1A (MIM: 601011), COL5A1 (MIM: 120215), HNF1A (MIM: 142410), KCNE1 (MIM: 176261), KCNJ2 (MIM: 600681), and HFE (MIM: 613609; GenBank: NM_000410.3; c.845G>A [p.Cys282Tyr], homozygotes only) were also deemed actionable by the eMERGE network on review of the underlying evidence. Sequencing and variant pathogenicity classification was performed at the Partners Laboratory for Molecular Medicine (LMM) according to ACMG/AMP guidelines with disease and gene specifications from ClinGen Expert Panels; clinical reports and variant summaries were accessed through the GeneInsight platform.¹⁰ VUS findings in genes associated with CRC were returned to individuals with CRC or polyps. Exome sequencing and variant interpretation of control subjects were performed by the University of Washington (UW) Center for Precision Diagnostics. Briefly, DNA captured using the xGen Exome Research Panel v1.0 (IDT) was paired-end (2x 100 bp) sequenced on HiSeq 4000s (Illumina). Resulting sequences were aligned to hg19 with BWA v.0.7.17, genotyped with GATK v3.8, and annotated with an in-house tool based on SNPeff (v4.3) prior to clinical interpretation. No VUS findings were returned to participants in the control cohort. To standardize variant pathogenicity classifications among the CRC, polyps, and control groups, all variant pathogenicity classifications were reviewed by UW prior to return. While testing occurred from 11/2016 to 11/2017, variant interpretations are reviewed and updated by LMM on a rolling basis, triggered each time the variant is detected in the course of their research and clinical testing. All returned variants and their interpretations can be found in Table S1.

Results

The three groups (CRC, polyps, and control) varied in age distribution, with CRC-affected case subjects being older than polyp-affected case subjects (Table 1). These groups were 54%, 56%, and 61% female, respectively. Participants’ self-reported race/ethnicity was primarily “White” across all three groups (CRC, 90%; polyps, 70%; controls, 83%); the next most frequent responses were “Asian” (CRC, 4.3%; polyps, 20%; controls, 8.7%) and “Black or African American” (CRC, 2.1%; polyps, 3.3%; controls, 5.8%). Despite our effort to exclude participants who previously had a genetic test, one participant in the polyp group was found to have previously had a genetic test revealing a pathogenic variant in PTEN. The PTEN test had not been ordered as a result of polyps, however. The remaining participants with positive results did not have a hereditary cancer genetic test as part of their medical care.

Table 1.

Demographics, and Rate of CRC/P-Associated and Non-CRC/P-Associated Actionable IFs that Were Either Pathogenic or Likely Pathogenic Findings by Group

	CRC	Polyps	Control
n	529	636	590
Female	54%	56%	61%
European ancestry	90%	70%	83%
Mean age at sampling, years	73 (33–100)	60 (29–91)	70 (59–74)
Mean age of onset	63 (18–89)	52 (18–83)	n/a
CRC/P positive	20 (3.8%)	5 (0.8%)	1 (0.2%)
CRC/P VUS	78 (14.8%)	96 (15.1%)	n/a
Non-CRC/P positive	15 (2.8%)	16 (2.5%)	12 (2.0%)

n/a = not applicable. Specific variants and interpretations can be found in Table S1.

The CRC-affected case subjects had a higher rate of positive (P or LP) findings for CRC/P-associated genes (20/529, 3.8%), when compared to the polyps (5/636, 0.8%; two-tailed chi-square p = 0.0004) or control (1/590, 0.2%, p = 0.00001) groups (Figure 1 and enumerated in Table 2). The rate of positive findings did not differ between polyp-affected case subjects and control subjects (p = 0.12). Of these positive results in CRC-affected case subjects, 17/20 were in Lynch syndrome (MIM: 120435)-associated genes (MLH1, MSH2, MSH6, and PMS2), while 3/20 were compound heterozygotes for P or LP MUTYH variants associated with the autosomal-recessive inheritance of polyp and CRC risk. The age of diagnoses of the CRC participants was slightly younger among those with a positive finding (mean = 58 years, range 18–87) compared to those with VUS (mean = 64, range 26–86), or negative findings (mean = 63, range 25–89), lending support to other data indicating that individuals with earlier onset cases are more likely to receive a positive genetic finding.¹¹ However, there were positive CRC genetic results in older individuals that had not had prior testing for a genetic syndrome. Of the positive results in the polyps-affected case subjects, 4/5 were in Lynch syndrome-associated genes and 1/5 was in PTEN, which is associated with Cowden syndrome (MIM: 158350). Among control subjects, there was only a single CRC/P-associated finding, a likely pathogenic variant in the Lynch syndrome gene PMS2. Variants of uncertain significance in CRC/P genes did not differ between CRC-affected (78/528, 14.8%) and polyps-affected (96/635, 15.1%; p = 0.93) case subjects, and were not called for control subjects.

Figure 1.

Number and Type of Findings by Ascertainment

n.s. = not significant.

Table 2.

Positive Findings (P or LP) in CRC-Associated Genes by Cohort

Gene	CRC (n = 529)	Polyps (n = 636)	Controls (n = 590)
MLH1	3	1	0
MSH2	3	1	0
MSH6	5	0	0
PMS2	6	2	1
All Lynch syndrome	17	4	1
PTEN	0	1	0
MUTYH (compound het)	3	0	0
MUTYH (carrier)	1	0	0
Non-Lynch	4	1	0

Specific variants and interpretations can be found in Table S1.

We evaluated family history relevant to Mendelian CRC given at time of post-test genetic counseling. Sixteen participants with positive results in MLH1 (five individuals), MSH2 (three), MSH6 (four), or PMS2 (four) were seen by a medical geneticist. Family history was ascertained from the participant, including three generations and third-degree relatives (e.g., cousins), and cancer status for pedigree members was recorded. Eleven were ascertained due to personal history of CRC and five due to one or more polyps. Of these, five reported no family history of CRC; positive findings in these individuals included MSH6 (three individuals), MLH1, and PMS2. One participant without a CRC family history did report two first-degree relatives with uterine cancer. In eight individuals there was no reported family history of CRC onset at age ≤50 years. In the remaining eight participants, a history of CRC at age ≤50 was reported. Of these eight families, in two only the proband had CRC at age <50 (positive result in MSH6 or PMS2), in three both the proband and a relative had CRC at age <50 (all associated with MLH1), and in three only a relative had CRC at age <50 year (associated with MLH1, MSH2, and PMS2; for the PMS2 the CRC at age <50 was in a third-degree relative only). By contrast, 2 of 17 individuals who were negative for CRC findings, but seen for findings unrelated to CRC (including HFE) reported such a family history (i.e., CRC at age ≤50). Three individuals found to have variants causing Lynch syndrome reported a family history of uterine cancer, and three others reported a family history of ovarian cancer, all six including first-degree relatives. Three of these six families with uterine or ovarian cancer did not have CRC at age ≤50 in any member. As MUTYH is associated with autosomal-recessive disease, it was evaluated separately. No family history of CRC was reported by three participants with two pathogenic variants in MUTYH, although two of the three probands were affected with CRC. The participant positive for PTEN was excluded from family history analysis, as the results had been clinically identified prior to this study.

Table 3 shows non-CRC/P IFs across the three groups. The CRC-affected case subjects had a rate of 2.8% (15/529), the polyps-affected case subjects had 2.5% (16/636), and the control subjects had 2.0% (12/590). These overall rates did not differ significantly. However, we note a higher rate of hemochromatosis (MIM: 235200) risk findings (HFE c.845G>A [p.Cys282Tyr] homozygotes) among CRC-affected case subjects (6/529, 1.13%) compared to either polyps-affected or control groups, which each had a rate consistent with reported disease prevalence in European ancestry individuals (0.2% in polyps-affected case subjects and control subjects, prevalence: 1 in 500). This was not a prespecified hypothesis, however. The ages of onset of CRC in these individuals were 43, 56, 61, 63, 64, and 72 years. Three were female. Only one of these participants (a female) had the diagnosis of hemochromatosis prior to the diagnosis of CRC. Four had a clinical diagnosis of hemochromatosis prior to the return of these study results.

Table 3.

Non-CRC/P-Associated Actionable Incidental Findings

Gene (Associated Phenotype)	CRC	Polyps	Control	All
n	529	636	590	1755
ACMG
Cardiomyopathy
MYBPC3 (HCM)	3 (2 LP)	5 (3 LP)	1 (1 LP)	9 (6 LP)
MYL3 (HCM)	0	1	0	1 (0 LP)
MYH7 (HCM/DCM)	1 (1 LP)	1 (1 LP)	0	2 (2 LP)
DSC2 (ARVC)	0	1 (1 LP)	1 (1 LP)	2 (2 LP)
DSG2 (ARVC)	0	0	1 (1 LP)	1 (1 LP)
Cardiac Electrical
SCN5A (and DCM)	1 (1 LP)	1 (1 LP)	1	3 (2 LP)
KCNQ1	0	0	2 (1 LP)	2 (1 LP)
Breast/Ovarian Cancer
BRCA2	0	4	1	5 (0 LP)
BRCA1	0	0	1	1 (0 LP)
Other
RYR1 (MH)	2 (2 LP)	1 (1 LP)	1	4 (3 LP)
LDLR (FH)	0	1 (1 LP)	1 (1 LP)	2 (2 LP)
MYH11 (AA)	0	0	1 (1 LP)	1 (1 LP)
Non-ACMG
PALB2 (breast cancer)	2	0	0	2 (0 LP)
HFE c.845G>A homozygotes (hemochromatosis)	6	1	1	8

Values in parentheses, where present, indicate any likely pathogenic results among the listed total positive findings. Abbreviations: HCM, hypertrophic cardiomyopathy; DCM, dilated cardiomyopathy; ARVC, arrhythmogenic right ventricular cardiomyopathy; FH, familial hyperlipidemia; MH, malignant hyperthermia; AA, aortic aneurysm.

There was a higher than expected rate of positive results for cardiomyopathy related IFs, 15 out of 1,755, or 0.85%. Notably 11 of these 15 were LP variants versus 4 P variants. Three participants were heterozygous for the MYBPC3 variant GenBank: NM_000256.3 (c.3628-41_3628-17del), reported by this clinical lab as LP, a South Asian founder allele which has been described as causing mild disease in heterozygotes and severe disease in homozygotes.¹² Though hypertrophic cardiomyopathy (HCM) has a prevalence of 1/500 or 0.2%,¹³ 10/1,755 participants (1/176 or 0.6%) were found to have a positive finding in an HCM-associated gene (2 P, 8 LP). The prevalence of dilated cardiomyopathy (DCM), including asymptomatic, has been estimated as 1/250 to 1/500,¹⁴ although more than half of the cases do not have an identifiable genetic basis. Thus, our positive rate for DCM of 2/1,755 (1/878) is difficult to evaluate. The prevalence of ARVD has lower confidence, but is often cited as 1/5,000,¹⁵ though we identified 3 positives (1/585).

Discussion

In this community health system ascertained cohort of generally older individuals with prevalent CRC and without prior genetic testing, the rate of CRC/P-associated positive (P or LP) findings was 3.8%. This is similar to the 5% of CRC believed attributable to known Mendelian genetic conditions with high penetrance.¹ This compares to a prior report of 4 positive results in 116 individuals (3.4%) with CRC presenting to a cancer center¹⁶ and is lower than the rate found (8%) in a recent study of 92 individuals with CRC/P presenting to a genetics clinic.¹⁷ A recent report of pathogenic variant detection in individuals with advanced cancers found a positive result on a 76 gene panel in 8 of 65 (12%) individuals with CRC;¹⁸ however, that study included moderate and lower penetrance genes. Our study did not evaluate genes more rarely and more recently implicated in hereditary CRC, including AKT1 (MIM: 164730), PIK3CA (MIM: 171834), GALNT12 (MIM: 610290), GREM1 (MIM: 603054), AXIN2 (MIM: 603816), RPS20 (MIM: 603682), NTHL1 (MIM: 602656), PDGFRA (MIM: 173490), MSH3 (MIM: 600887), and CTNNA1 (MIM: 116805). Inclusion would be expected to increase our positive rate very slightly.

These findings suggest that CRC associated with Mendelian conditions may be found at a similar rate in other community-based settings. In particular, our cohort includes older individuals with CRC who were found to have hereditary cancer conditions that had escaped detection. The majority of these positives (85%) were in Lynch syndrome genes, putting the individuals at risk for non-CRC cancers which could be screened for, particularly endometrial cancer in women. Further, these results have significance for those family members who may share the pathogenic variant. As these data suggest that age alone is not a reliable factor in evaluating the need for a genetic test among those with a CRC diagnosis, new strategies to identify appropriate individuals for CRC genetic testing should be considered to improve detection of these cases.

We ascertained participants who were not known to have had prior genetic testing. It is expected that our 3-generation pedigrees taken in-person by a medical geneticist for individuals with CRC-associated positive genetic tests will be more complete than routine family history taken in a clinical setting. Assuming the family history information obtained at the genetics visit was correct, routine genetic testing for individuals with CRC at age ≤50 years would have identified only 5 of 16 probands and 8 of 16 Lynch-affected families. One additional family could have been identified by universal genetic testing for ovarian cancer. The remaining 7 of 16 families would have been difficult to identify as at risk of Lynch. Testing of individuals with CRC at age ≤50 would have only identified 1 of the 4 PMS2-associated results. As PMS2 variants are known to have lower penetrance for CRC and uterine cancer than pathogenic variants in other Lynch-associated genes,¹⁹ family history is expected to be less reliable as an indicator of an underlying genetic cause. CRC onset at age ≤50 would not have identified any of the compound heterozygotes for P or LP variants in MUTYH. It is unsurprising that family history is not informative for the autosomal-recessive disorder associated with MUTYH. While many of these families had concerning CRC history, which could potentially have been captured by careful assessment of family history, this risk was not identified or did not lead to genetic testing prior to this study. These results are similar to a larger study of individuals found to harbor a pathogenic or likely pathogenic variant in BRCA1/2 in a large biobank. In that study, of those who harbored such a variant that had not been detected clinically, approximately 50% did not meet personal or family history criteria for testing.²⁰ A similar result of ∼50% of those who harbored a pathogenic variant in BRCA1/2 not having a suggestive family history was reported in Israel.²¹ Nonetheless, in these cases the health systems did not identify or act on those indications when they were present.

The rate of positive CRC/P results is lower in those with any number of polyps or controls, 0.8% and 0.2%, respectively. The low threshold for inclusion (any number of polyps) is likely responsible for the low rate of findings in the polyps-affected group. A similar smaller study of 77 individuals who had “few” hamartomatous polyps did not yield any positive genetic results.²² These results may imply that individuals with few polyps will benefit less from evaluation for genetic etiology than CRC-affected individuals. However, stratifying by number of polyps, which is not possible with these data, would be expected to clarify the rate of positive findings.

Rates of detected non-CRC/P actionable IF were similar across groups; however, they were higher than disease prevalence suggested for some disorders. Notably, we found an excess of hemochromatosis risk genotype (HFE c.845G>A [p.Cys282Tyr] homozygotes) among CRC-affected case subjects (1.13%) compared to either polyps-affected or control groups, which each had a rate consistent with the population allele frequency in European ancestry individuals (MAF ∼4.96%). Hemochromatosis has been associated with excess risk of CRC;²³ however, this is not well known in the genetics community and is not included in most hemochromatosis review papers used by physicians. This finding warrants further estimation of the risk of cancer in individuals with hemochromatosis, including both risk and age of onset. If increased, the risk of cancer in hemochromatosis could be better incorporated into the care of these individuals and there may be utility in adding this single variant to CRC gene panels. Increased CRC risk in HFE c.845G>A (p.Cys282Tyr) homozygotes could add to the growing body of evidence supporting broader return of results for this specific genotype as an incidental finding.^{24, 25}

None of the participants identified as having an IF in a cardiomyopathy-associated gene had clinical cardiomyopathy, despite most being of older ages (range 31 to 76; mean 59). While some may have had subclinical disease, the prevalence of echocardiogram-detected cardiomyopathy is too low for subclinical disease to be a major factor. There are two possible explanations for the excess of P and LP IFs in cardiomyopathy-associated genes. First, some of the variants called P or, particularly, LP, may truly be benign. Second, some of these variants may be disease associated, but not with high penetrance.¹⁴ The LP classification category is intended to be used when variants have a greater than 90% chance of being pathogenic;²⁶ however, current methods are decidedly non-quantitative. Variant classification has evolved over the last 12 years and the data and methods to allow proper classification has notably changed with the introduction of large-scale and diverse population allele frequency data, which assists in classifying variants as too frequent to be high penetrance and pathogenic as well as more stringent approaches to variant classification.²⁶ However, even with these newer pathogenicity classification methods and databases, we find an excess of IFs; these data raise concern about the validity of the methods for LP classification and/or the expected penetrance of pathogenic cardiomyopathy variants.

These concerns could have major repercussions when considering return of IFs. In clinical medicine, P and LP results are treated in the same manner, with the same clinical treatments and the same follow-up in families. It is possible that LP class variants’ pathogenicity is over-estimated and these results should be clinically treated as more tentative. This may be particularly true for cardiovascular IFs as opposed to cancer-associated IFs, as the latter have a longer history of variant curation and a larger testing base to draw data from. In addition, variants in tumor suppressor genes are often easier to interpret given a loss-of-function mechanism leading to a majority of pathogenic variants being obviously disruptive (e.g., nonsense, frameshift, canonical splice site) compared to most cardiac disorders which are primarily caused by gain-of-function missense variants, a class much more difficult to interpret. Furthermore, we have less data to inform the penetrance of cardiac disease associated ACMG IF gene variants. Penetrance is a major factor in the ClinGen system for evaluating the actionability of pathogenic variants. Returning pathogenic ACMG IFs has been shown to be cost effective.²⁷ The cost benefit ratio is reduced when the penetrance is low. If an individual with a LP or P cardiomyopathy-associated pathogenic variant is unlikely to develop disease, due to either variant misclassification or low penetrance, then return of this information reduces its overall value to both the individual and the health system. Interventions to diagnose or prevent disease, such as echocardiograms and medications to reduce cardiac burden, may not be worthwhile. Better data on penetrance and the value of returning LP variants is therefore critically needed to balance the risks of intervention against the likelihood of specific disease outcomes.

The ACMG recommendations do not include return of LP results as IFs, but only include the return of P and “expected pathogenic” (EP) variants.⁸ The latter include predicted loss-of-function variants (e.g., nonsense, frameshift, canonical splice site) in genes where haploinsufficiency or biallelic LOF is a known mechanism of disease. Given that the more recent ACMG/AMP variant classification system does not include an EP category, it is unclear how to distinguish the EP from LP class for return. While the ACMG/AMP recommendations do consider an EP variant to be supporting evidence for a subsequent P or LP classification (code PVS1), variants that achieve an LP result that are EP may be, as a group, more likely to truly be pathogenic than variants that achieve the LP classification without being EP. Despite the current ACMG recommendations, some labs do offer to report LP IFs. These data suggest more confidence in return of cancer- than cardiac-associated LP variants.

A limitation of this study is that the CRC-affected and polyp-affected case subjects were sequenced and had variant pathogenicity called by a different clinical lab using a different sequencing platform than the control subjects, although all participants were ascertained from the same health care system. Although coverage was consistent across the two platforms, CNVs were not called or returned in the control cohort. These limitations are mitigated by the ability to compare the CRC and polyp groups directly, and the low rate of CNV IFs. An additional limitation is the age distribution difference among groups. While germline genotypes do not change over the lifespan, survival could vary among these groups. However, these results are meant to address the findings in prevalent case subjects; the older distribution of case subjects is typical of that. The polyp-affected group had a larger Asian ancestry proportion than other groups. This is unlikely to have significantly affected the results; the low rate of positive findings in this group is likely due to the fact that a single polyp resulted in inclusion. Finally, the results in this predominantly European ancestry study may not be generalizable to other ancestry groups. Similar studies should be repeated in diverse populations.

In summary, we find a rate of positive results for CRC-associated genes to be 3.8% (n = 20/529) in a community-based cohort of individuals with prevalent CRC. None of these positive genetic results had been clinically identified prior to the study and their presence changes medical care for both the participants and their families. These results demonstrate that usual care is not sufficient to identify many individuals with Mendelian CRC and suggest that a more liberal genetic testing strategy in those affected with CRC, but not polyps, may improve on this deficit. The overall IF rate in this cohort was approximately 2.5% (n = 44/1,755) and did not vary significantly based on ascertainment. We noted an unexpected excess of CRC-affected case subjects that had the HFE c.845G>A (p.Cys282Tyr) homozygous genotype; this merits further study. Finally, that cardiac IFs were more frequent than expected based on population prevalence suggesting that pathogenic variants in these genes may be of lower penetrance or some of the reported variants, particularly the LP, may not be pathogenic. Clinical return of incidental LP variants for these less-studied genes may be imprudent, particularly if the biological mechanism underlying a given disease association remains poorly understood.

Declaration of Interests

The authors declare no competing interests.

Published: August 15, 2019

Data and Code Availability

Data are currently accessible via dbGaP (accession: phs001616.v1.p1).

Web Resources

• Colorectal Cancer Stat Facts, https://seer.cancer.gov/statfacts/html/colorect.html

• GenBank, https://www.ncbi.nlm.nih.gov/genbank/

• OMIM, https://www.omim.org/

Supplemental Data

Table S1. All Variants Returned to Participants

Shown are all variants across CRC, polyps, and control cohorts, the clinical lab’s reported interpretation of each variant, and in how many individuals each variant was found. Variants of Uncertain Significance were not assessed or returned for control individuals.