Testing Positive on a Multigene Panel Does Not Suffice to Determine Disease Risks

Historically, cancer genetic testing has been limited to single-gene testing in families who fulfill susceptibility syndrome criteria. Today, next-generation sequencing-based multigene panel tests are commonly used when the syndrome is unclear, multiple genes might explain the phenotype, or when single-gene testing fails to detect a pathogenic germline mutation (1). Panel prices have plummeted, and families with minimal cancer family history in the syndromic spectrum of the mutated gene are currently being tested. However, the penetrance of strongly pathogenic germline mutations varies dramatically by testing indication, family history pattern, and other risk factors. This underappreciated fact will require a paradigm shift regarding how clinicians counsel patients about disease risks implied by mutations found on multigene panel testing.

Li-Fraumeni syndrome (LFS), a prototypic hereditary cancer syndrome mainly caused by pathogenic germline TP53 mutations, is illustrative. In this issue of the Journal, Rana et al. (2) report that individuals with pathogenic TP53 mutations detected by multigene panels have substantially different phenotypes, including cancer onset up to 25 years later, than when mutations are found by single-gene TP53 testing. “Confounding by indication” drives these discrepancies: subjects tested for only TP53 mutations more often met LFS clinical criteria. Those given multigene panel testing had less specific symptoms, necessitating a wider search for deleterious mutations.

The less severe phenotypes in TP53 mutation carriers identified by multigene panel testing may portend less classical phenotypes, especially as screening and treatment improve (eg, the Toronto protocol [3]), and atypical patients are diagnosed at older ages. Some individuals who present later in life may require different management than younger mutation carriers. An analgous situation is HIV-positive individuals in the United States, who now survive into their 60s without manifesting AIDS. Today, their cancer spectrum has shifted from AIDS-defining cancers (Kaposi’s sarcoma, non-Hodgkin lymphoma, and cervical cancer) to more typical aging-associated cancers (lung, prostate, and colorectum) and very high risk of anal cancer (4).

Furthermore, pathogenic TP53 mutations are substantially more common in the general population than indicated by studies of classical LFS families. A recent general population-based cohort showed a surprisingly high pathogenic and likely pathogenic TP53 germline mutation prevalence (0.2%), regardless of family history, 10 times higher than prior family study estimates (5). Another example is pathogenic germline mutations in DICER1, a genetic cause of childhood pleuropulmonary blastomas. Historically thought to be extremely rare, DICER1 mutation prevalence in an asymptomatic population was surprisingly high, at 1 in 10 600 (6). Thus, most TP53 and DICER1 mutation carriers in the general population likely have lower cancer risks than mutation carriers in highly affected families. These findings may surprise clinicians who are trained that pathogenic germline mutations in highly penetrant genes reliably define a cancer predisposition syndrome that confers extremely high cancer risks for all mutation carriers.

Multigene panel testing is more likely to identify the mutation carriers at lower disease risk because those who get multigene panel testing present with less personal and family history of disease, demonstrating that other risk factors remain important. The observed family history is a key clue to the importance of unknown, disease-modifying factors, as definitively demonstrated in BRCA1/2. In 2007, a controversial finding was reported, that women who tested negative for their family’s BRCA1/2 mutation presented greater than population average breast cancer risk (7). We suggested that this finding was genuine and hypothesized the presence of other unknown cancer risk–modifying mutations in these families (8). Families with 85% breast cancer penetrance to age 70 years probably have additional breast cancer risk factors because population-average penetrance is around 50% (9). Vice versa, families with a BRCA1/2 mutation but minimal family history of breast/ovarian cancers might lack these risk modifiers or carry unknown protective factors. In the Washington Ashkenazi Study (WAS), breast cancer penetrance was substantially increased, 1.5-fold (95% confidence interval = 1.15 to 1.97), if a BRCA1/2 mutation carrier had at least one first-degree relative with breast cancer (8). Although 1.5-fold risk increases are usually considered small, when acting on a high baseline of 50% penetrance, they substantially increase risk by 25%. Increasing recognition of “missing heritability” and polygenic risk further substantiate the legitimacy of these observations (10).

The population-based WAS had minimal ascertainment bias, but definitive proof for this risk increase required prospective data. In 2017, BRCA1/2 mutation penetrance for breast cancer was reported as an increase of approximately 1.7-fold if the patient had at least one first/second-degree relative with breast cancer (11), replicating the WAS findings. For BRCA2 carriers, breast-cancer penetrance to age 70 years decreased 25% if the woman had no first/second-degree relatives with breast cancer (64% vs 39%) (11). Because that cohort was largely clinic based, their penetrances overestimate general population rates (9), so risks for mutation carriers in the general population who have no first/second-degree relatives with breast cancer should be even lower. Similarly, the 23% of mutation carriers having no breast cancer–affected first/second-degree relatives (11) should also underestimate that percentage in the general population. In the BRCA1/2 context, it took 10 years to validate what, in retrospect, should have been the default presumption: the effect of rare, highly penetrant susceptibility gene mutations can be greatly amplified, or muted, by even weak personal/family cancer history and other nongenetic risk factors.

These lessons from TP53 and BRCA1/2 apply to risk assessment for any mutation identified by multigene panel testing. Multigene panels are identifying individuals with pathogenic germline mutations, but with little or no personal or family history of cancer in the syndromic spectrum of the mutated gene. Such individuals cannot be advised that they have the same disease risks as those from highly affected families. Consideration of personal/family cancer history and other risk factors will become paramount, because risk factors with small relative risks are magnified in the presence of powerful mutations (12). Accurately counseling individuals with mutations found on panel testing is very challenging, as research studies are usually based on moderate, or small, samples of mostly highly affected families (https://ask2me.org/). Thus, it is crucial that patients and providers report their test results to registries such as the Prospective Registry of Multi-Plex Testing (PROMPT; https://www.promptstudy.org) if we are to have any hope of collecting sufficient data to refine these questions. Ideally, registries would capture information about personal and family history of diseases, and all known disease risk factors. If the diagnostic blood sample were retained for research purposes, it could be examined for more common, less penetrant mutations that modify cancer risk.

Without such an effort, we are at risk of inaccurately counseling many, perhaps even most, people with mutations found by multigene panel testing. Pending the development of more refined, precise cancer risk estimates, it is incumbent upon us to be transparent with patients about what we know and what we don’t yet know, and to carefully avoid basing aggressive management options on risk estimates that have been inappropriately generalized from the era of single-gene testing.

Funding

This was supported by the Intramural Research Program of the US National Institutes of Health/National Cancer Institute.

Notes

Affiliations of authors: Division of Cancer Epidemiology and Genetics, National Cancer Institute, National Institutes of Health, Department of Health and Human Services, Bethesda, MD.

The funder had no role in the writing of the editorial or the decision to submit it for publication.

The authors have no conflicts of interest to report.

References

Author notes