Genomic medicine is expanding from a focus on diagnosis at the patient level to prevention at the population level. As test costs fall, it could become feasible to screen pre-symptomatic individuals through public health-based approaches, rather than restricting testing to high-risk groups. Indeed, pilot initiatives in which hundreds of thousands of individuals will undergo genomic screening are being launched in health systems in the United States (U.S.) (Carey et al., 2016; Schwartz et al., 2018; Lacaze et al., 2019; Grzymski et al., 2020), the United Kingdom (U.K.) (Genomics England, 2021), and Australia (Rowley et al., 2019; Lacaze et al., 2022). Leading hereditary conditions for consideration in population screening include hereditary breast and ovarian cancer syndrome (HBOC), Lynch syndrome (LS), and familial hypercholesterolemia (FH). These conditions are prioritized for screening due to their under-ascertainment using current screening approaches and the availability of evidence-based interventions to reduce morbidity and mortality (Centers for Disease Control and Prevention OoPHG, 2022).

Traditional methods to identify cases with HBOC, LS, and FH include genetic testing for patients meeting clinical, ethnicity or family-history based criteria (Hampel et al., 2008; Schofield et al., 2014; Klančar et al., 2015; Tognetto et al., 2017; Groselj et al., 2018; Gupta et al., 2019; Daly et al., 2020; Kunnackal John et al., 2021; Zuurbier et al., 2021). However, these targeted approaches have been found to miss a substantial proportion of individuals who harbor pathogenic variants. For example, >50% of individuals with pathogenic BRCA1 and BRCA2 (BRCA1/2) variants are missed by family history-based criteria (Metcalfe et al., 2010a; Gabai-Kapara et al., 2014; Manchanda et al., 2015a). The availability of large-scale next-generation sequencing (NGS) strategies and preventive options for these conditions makes it increasingly feasible to screen pre-symptomatic individuals through public health-based approaches, rather than restricting testing to high-risk groups.

This raises anew, and with urgency, questions about the limits of screening as well as the capacity to screen for genetic conditions at a population level, or in other words, population genomic screening. We use the term “population genomic screening” to refer to germline DNA testing among an unselected, asymptomatic population with the aim of identifying individuals with pathogenic/likely pathogenic (henceforth, “pathogenic”) variants. Key issues to scaling up population genomic screening include the optimal testing approach, penetrance of these conditions in the general population, clinical effectiveness, cost-effectiveness, acceptability, health system capacity to implement such a program, ethical issues such as overdiagnosis, access challenges and equity.

Decisions about screening are expected to align with the World Health Organization principles of screening. These criteria, developed by Wilson and Jungner in 1968, inform decision-making around disease screening and generally include considerations of the nature of the disease, test characteristics, and the availability, effectiveness and acceptability of preventive interventions or treatments (Table 1) (Wilson and Jungner, 1968). Since its publication, Wilson and Jungner’s criteria have been widely accepted, modified and used by decision-makers across the world to guide screening decisions. Whereas the Wilson and Jungner criteria were developed for programs aiming to enable early detection and intervention for individuals with early stages of a disease, population genomic screening programs would identify those with a genetic predisposition to disease. The identification of a pathogenic variant in an asymptomatic individual through genetic screening is not equivalent to a clinical diagnosis of the associated disease (Murray, 2016; Murray et al., 2021). Given the complexity of policy decision-making for genetic tests and genetic screening programs, various frameworks and sets of decision criteria have been developed to guide these decisions (Sanderson et al., 2005; Burke and Zimmern, 2007; Andermann et al., 2008; Teutsch et al., 2009; Andermann et al., 2011; National Academies of Sciences Engineering and Medicine, 2017; Pitini et al., 2019). While these newer frameworks and decision criteria share core elements with Wilson and Jungner such as those related to the natural history of the condition, the effectiveness of the test, and effectiveness of preventive interventions, newer frameworks extend Wilson and Jungner’s criteria to include considerations related to implementation issues such as health service delivery, ethics, and equity. However, these more recent criteria for genomic evaluation have not been universally adopted, and different health systems vary in which criteria are used in policy decisions, if any. Given the lack of a universally accepted set of decision criteria for genomic screening, and the continued relevance of the fundamental principles of Wilson and Jungner, we will use the Wilson and Jungner criteria to guide a synthesis of the current evidence on population genomic screening for leading gene-condition pairs. In addition, we also discuss ethical and equity considerations. While these are absent from the original Wilson and Jungner criteria, they are increasingly important in decision frameworks for genomic screening programs (Andermann et al., 2008; Pitini et al., 2019) and are commonly considered across various frameworks and sets of decision criteria for genomic technologies (Burke and Zimmern, 2007; Andermann et al., 2008; Teutsch et al., 2009; Andermann et al., 2010; Botkin et al., 2010; Andermann et al., 2011). We highlight policy and practice issues as well as future research priorities to inform the design of population genomic screening programs to maximize population benefits and minimize harms.

HBOC, LS and FH are characterized by their high penetrance, evidence-based interventions for prevention/treatment and subsequent benefits from the early detection, in line with fundamental principles of screening. The CDC Office of Public Health Genomics (OPHG) designates screening for HBOC, LS, and FH as Tier 1 genomic applications (Centers for Disease Control and Prevention OoPHG, 2022). Tier 1 genomic applications are those which could have a substantial, positive impact on public health based on: 1) A high prevalence of 1 in 200 for HBOC, 1 in 340 for LS and in 1 in 250 for FH in the general populations (exact frequency may vary in certain populations); 2) the under-ascertainment of current strategies; and, 3) established risk-reducing interventions that reduce morbidity and mortality (Abul-Husn et al., 2016; Akioyamen et al., 2017; Manickam et al., 2018; Grzymski et al., 2020; Manickam et al., 2021).

The natural histories of HBOC, LS, and FH are relatively well understood. HBOC is caused by pathogenic variants in BRCA1/2 which confer substantially elevated risks for female breast cancer, ovarian cancer, and male breast cancer (in particular for BRCA2 carriers), in addition to increased risks for pancreatic cancer, prostate cancer, and melanoma (The Breast Cancer Linkage Consortium, 1999; Brose et al., 2002; Levine et al., 2003; Lindor et al., 2008; Lynch et al., 2009; Moran et al., 2012; Mavaddat et al., 2013; McKay et al., 2016). While pathogenic variants in other genes including PALB2, RAD51C, RAD51D, and BRIP1 also cause hereditary breast and ovarian cancer, we focus this review on BRCA1/2 because of the higher frequency of pathogenic variants in the population in these genes, and established clinical management guidelines (Manickam et al., 2018; National Comprehensive Cancer Network (NCCN), 2021a). LS is caused by pathogenic variants in mismatch repair (MMR) genes MLH1, MSH2, MSH6, and PMS2, as well as deletions in EPCAM which lead to silencing of MSH2. Affected individuals are at increased risk for colorectal cancer (CRC), endometrial cancer, ovarian cancer, and other cancers (Lindor et al., 2008; Senter et al., 2008; Baglietto et al., 2010; Bonadona et al., 2011; Giardiello et al., 2014). FH, caused by pathogenic variants in LDLR, PCSK9, and APOB, is characterized by elevated plasma low-density lipoprotein cholesterol (LDL-C) levels, which leads to risks for cardiovascular disease and premature mortality (Youngblom et al., 2016).

Two key issues that inform natural history are penetrance and age of onset. HBOC, LS and FH exhibit high but incomplete penetrance. Although the penetrance (the chance that an individual with the condition will manifest particular features) of the causative genes has been estimated in cohorts ascertained with strong personal and family history of disease, it has yet to be well-established in the general population (Murray et al., 2021). Some studies suggest penetrance in the general population may vary from estimates from family-based studies (Forrest et al., 2022). However, the risk to those identified through population screening will likely still be high enough to warrant clinical intervention, at least in BRCA1/2 carriers where there is substantial evidence demonstrating high penetrance even among unselected cases (Chatterjee et al., 2001; Chatterjee and Wacholder, 2001; Antoniou et al., 2005; Chatterjee et al., 2006; Kuchenbaecker et al., 2017; Chen et al., 2020). These studies highlight the importance of evaluating the appropriateness of population genomic screening and subsequent interventions, given the potential for overdiagnosis and overtreatment (to be discussed in a subsequent section, Ethical considerations). Adding another layer of complexity to risk prediction, other genetic factors, such as polygenic background, and non-genetic risk factors (e.g., diet, environmental exposures, and clinical risk factors) can also influence the penetrance of these conditions (Fahed et al., 2020).

Based on the age of onset and availability of age-appropriate preventive interventions, the optimal age to initiate screening will vary across target conditions. For example, surveillance and risk-reducing surgeries for HBOC and LS are recommended in adulthood (National Comprehensive Cancer Network (NCCN), 2021a; National Comprehensive Cancer Network (NCCN), 2021b), while pharmacologic treatment of FH can begin in childhood (Gidding et al., 2015). The health outcomes and costs of population screening programs will likely vary depending on the age at which screening and intervention is initiated. Specific considerations related to the target population for each condition are provided throughout the subsequent sections.

One element of test performance is validity, which encompasses both “analytic validity” (accuracy in detecting the target genetic variant) and “clinical validity” (accuracy in identifying patients with the target condition) (Bombard et al., 2013). Test selection for population genomic screening should consider what type of genetic variation primarily causes the target condition, and testing laboratories should be equipped to manage gene-specific technical challenges [e.g., PMS2 pseudogenes (Hegde et al., 2014; Li et al., 2015; Lee et al., 2021a)]. Several laboratory considerations for population genomic screening include whether to perform full gene sequencing or targeted variant testing, whether to test for only known pathogenic variants or also novel variants, and whether to perform deletion/duplication analysis in addition to sequence analysis; each of these decisions will impact test costs and post-test residual risk (Lu et al., 2019). NGS has very high analytic sensitivity and specificity for detecting single-nucleotide variants and small insertions/deletions (Baudhuin et al., 2015; Judkins et al., 2015; Toland et al., 2018), and could be coupled with gene-targeted deletion/duplication analysis to increase detection of disease-causing variants for HBOC, LS and FH (Petrucelli et al., 1998; Idos et al., 2004; Ison et al., 2014). Deletion/duplication analysis is necessary to identify disease-causing variants in EPCAM. The use of array-based genotyping in population genomic screening has been found to result in false positives and false negatives compared to NGS or Sanger sequencing (Blout Zawatsky et al., 2021; Bowling et al., 2021). For HBOC, in the Ashkenazi Jewish (AJ) population, there are three founder variants (BRCA1 c.68_69delAG, BRCA1 c.5266dupC and BRCA2 c.5946delT) which are prevalent in ∼2.5% (Roa et al., 1996) of the population. While these variants do account for the majority of pathogenic BRCA1/2 variants in the AJ population (Walsh et al., 2017), some BRCA1/2 carriers would be missed if targeted founder variant testing as opposed to NGS was used in population genomic screening among the AJ population (Rosenthal et al., 2015; Solano et al., 2018).

Another aspect of genetic test performance is variant interpretation (Richards et al., 2015). Key issues related to variant interpretation include variants of uncertain significance (VUS) (Burke et al., 192022), discordant variant interpretations between diagnostic laboratories (Garber et al., 2016; Harrison et al., 2017; Iacocca et al., 2018; Lebo et al., 2018; Amendola et al., 2020; Mighton et al., 2021a), variant reclassification over time (Macklin et al., 2018; Mersch et al., 2018; Slavin et al., 2018; Turner et al., 2018; Esterling et al., 2020; Chiang et al., 2021) and recontacting patients with updated results (e.g., changes from VUS to likely pathogenic or pathogenic which may warrant medical follow-up) (Otten et al., 2015; El Mecky et al., 2019). While these issues exist in standard clinical genetic testing, they will be magnified if genomic screening is conducted at a population scale, and will need to be considered in program design/implementation.

A further aspect of test performance is the positive predictive value (PPV), the probability that a patient with a positive result (a reported pathogenic or likely pathogenic variant) has the associated condition (Hagenkord et al., 2020). PPV depends on test characteristics (specificity, sensitivity) and condition prevalence (Akobeng, 2007; Oleske, 2010; Hagenkord et al., 2020). As HBOC, FH, and LS have a lower prevalence in the general population compared to populations ascertained based on family history, this would reduce the PPV of a positive result obtained from population genomic screening compared to a positive result from genomic testing among high-risk populations (Hagenkord et al., 2020). Estimates of PPV for Tier 1 conditions range from 80% to 91%, assuming 99.95% specificity and that one-third of the overall positive rate is likely pathogenic variants and two-thirds are pathogenic variants (Hagenkord et al., 2020). Increasing test specificity can increase the PPV, which laboratories could accomplish by adjusting the reporting cut-off between a positive and a negative result (Lu et al., 2019; Hagenkord et al., 2020). For example, reporting only high confidence likely pathogenic variants can increase specificity (Hagenkord et al., 2020).

Among these three conditions, there is a pre-symptomatic state that is identifiable by molecular testing for pathogenic variants in the relevant genes (Youngblom et al., 2016; Petrucelli et al., 2022). Therefore, population genomic screening for HBOC, LS, and FH can be used to identify individuals with pathogenic variants in the causative genes who would not otherwise be identified through routine clinical care and could gain benefits from early intervention (Grzymski et al., 2020). Multiple studies have found that population genomic screening identifies carriers of pathogenic variants for HBOC, LS, and FH who were previously unaware of their variant (Buchanan et al., 2020; Grzymski et al., 2020; Abul-Husn et al., 2021; Lee et al., 2021b; Blout Zawatsky et al., 2021).

There are various surveillance and prevention options endorsed by clinical practice guidelines to guide the management of individuals with HBOC, LS and FH.

There are evidence-based clinical practice guidelines for the management of individuals with pathogenic variants in genes for HBOC, LS, and FH, as described above. It is important to consider that the evidence used to develop these guidelines is largely from cases ascertained through standard diagnostic approaches, as opposed to through population screening-based ascertainment (Murray et al., 2021). Over time, as evidence on penetrance in unselected populations accumulates, management guidelines may need to be updated with specifications for how to manage individuals with disease risk identified through population genomic screening, given the potential reduced penetrance (Murray et al., 2021). This is less likely to be necessary for the genes included in this review than for moderate penetrance genes, given that the penetrance is likely still high in unselected populations and sufficient to warrant clinical intervention.

Current models of genetics care are personnel- and time-intensive and not feasible at a population-scale. Key challenges include critical workforce shortages, which contribute to long wait times, a lack of genetics education among non-genetics specialist healthcare providers, and fragmentation of care (Suther and Kiros, 2009; Hann et al., 2017; Office of the Auditor General, 2017; Hoskovec et al., 2018; Stoll et al., 2018; Dragojlovic et al., 2020). These challenges persist in urban areas and are exacerbated in remote and under-served communities (Office of the Auditor General, 2017). Capacity to sustain population genomic screening must also include laboratory infrastructure, secure data storage, as well as bioinformatic and analytic pipelines to support population-scale genomic analyses (Kelly et al., 2021). There is a paucity of data on the availability and distribution of laboratory infrastructure and personnel including clinical laboratory geneticists and medical laboratory technicians (Dragojlovic et al., 2020). This is critical to understand as it will be variable across jurisdictions and will be important for decision-makers to determine how to deliver the program (i.e. the distribution of testing centres).

The possibility for variants to be reclassified over time means that case finding must be an ongoing process. Most reclassifications are from VUS to likely benign or benign, and reclassification of variants initially classified as pathogenic/likely pathogenic is very rare (Macklin et al., 2018; Mersch et al., 2018; Mighton et al., 2019). In the context of population genomic screening, reclassifications from VUS to pathogenic/likely pathogenic are particularly relevant, as an upgrade from VUS to pathogenic/likely pathogenic could impact medical management. This raises questions about the need for periodic reanalysis and recontact of patients for the return of updated results. The issues of reclassification and recontact already present practical and resource challenges in the context of targeted, clinical testing (Otten et al., 2015; El Mecky et al., 2019), and would be magnified if testing were implemented at the population scale. This is critical to note as non-European populations consistently have higher VUS rates due to lack of representation in databases, leading to higher rates of reclassification and the need for recontact in these populations (Popejoy and Fullerton, 2016; Slavin et al., 2019; Buchanan et al., 2020; Popejoy et al., 2020). There are currently variation in recontact guidelines and practices across jurisdictions, laboratories, and health systems (Bombard and Mighton, 2018; Sirchia et al., 2018), despite recontact being expected by patients (Linderman et al., 2016; Mighton et al., 2021b).

It is important to consider the potential harms and unintended consequences of population genomic screening. Early detection and preventive strategies for HBOC, LS, and FH such as high intensity surveillance, prophylactic surgeries, and pharmacotherapy are not without risks including exposure to radiation, false positives, surgical complications, and adverse drug reactions.

For HBOC, there is some observational evidence to suggest that exposure to diagnostic radiation, including mammography, at a young age is associated with increased risk for breast cancer among females with disease-causing BRCA1/2 variants (Pijpe et al., 2012). A systematic review of the harms of breast cancer screening among average-risk females found that harms included overdiagnosis (at rates of 11%–22% from randomized controlled trials [RCTs]) and false positive results which were associated with elevated anxiety, distress, and breast-cancer specific worry; however, the review only included females at average-risk and excluded those with pathogenic BRCA1/2 variants (Nelson et al., 2016). Psychological harms have been identified among BRCA1/2 carriers, related to false positives and living at risk for disease (Metcalfe et al., 2020). With respect to LS, a systematic review of colorectal cancer screening among average-risk individuals found serious adverse events from colonoscopy including perforations and major bleeds, but these events were uncommon in average-risk populations (Lin et al., 2021). However, high-risk patients such as those with LS were excluded from the review (Lin et al., 2021). For FH, the safety profile differs across pharmacologic therapies. For statins and PCSK9 inhibitors, RCTs have found that treatment-related adverse events did not significantly differ between therapy and placebo (Kastelein et al., 2015; Lozano et al., 2016b), though for statins there are sporadic reports of systemic, immunologic, and pain-related adverse events (Lozano et al., 2016b). Bile acid sequestrants have been commonly associated with adverse GI symptoms, and poor palatability (Lozano et al., 2016b).

Across all conditions, potential harms include genetic discrimination which can arise in a variety of settings. This includes insurance discrimination, which is especially relevant in countries such as the U.S. where much of the population must purchase private health insurances (Ridic et al., 2012; Maynard, 2013). Harms may also be caused when carriers face challenges in accessing risk-reducing strategies in jurisdictions without universal healthcare coverage or among historically underserved populations (e.g., rural populations) (Nguyen-Pham et al., 2014; Chandak et al., 2019; Villegas and Haga, 2019). This raises the question of whether it is ethical to offer population genomic screening in the absence of universal coverage of downstream risk-reducing management. Patient harm may also arise if patients who receive negative results from screening are falsely reassured and forego recommended scheduled screening for average risk populations (i.e., age- and family history-recommended screening) although recent evidence suggests this risk may be minimal (Burnell et al., 2022). Conversely, false positive results may lead to overdiagnosis and overtreatment, where patients and family members may undergo unnecessary investigations and potentially life-altering procedures such as prophylactic surgeries. Although these issues also affect patients undergoing family-history based testing, the higher rates of false positive results associated with population screening coupled with a larger number of patients undergoing genetic testing translates to a larger volume of patients who may receive inappropriate and unnecessary medical care.

At present, the balance of benefits and harms of population genomic screening are not well-characterized. This calls into question whether and to what degree the balance of benefits and harms of screening and subsequent interventions for HBOC, LS and FH should be discussed with patients to ensure informed decision-making. Likewise, it remains unclear how to meaningfully obtain informed consent at the population level given the diversity of literacy, health literacy, socioeconomic status, geography, and culture among screened populations. Genomic screening might not be desirable for all people based on their values and preferences, further highlighting the importance of informed decision-making.

Return of results at the population level presents a further issue. Genomic information is uncertain and complex; delivering this information may lead to adverse psychological outcomes (Mighton et al., 2021c). Among patients receiving positive results after genetic testing, there is evidence of increased risk of anxiety, distress and depression (Rew et al., 2010; Wade et al., 2010; Wade, 2019). Certain populations may face additional risks, such as children feeling a loss of autonomy and women who feel burdened with the responsibility of sharing results with relatives (Gaff et al., 2007; Wade et al., 2010; Wade, 2019). Moreover, parents become overprotective of genetically at-risk children and recognize a disruption of the parent-child relationship (Rew et al., 2010). Although these harms are typically rare and transient, genomic screening at a population level will result in a large number of individuals with psychological harms. In addition to high-quality genetic counseling support, there will also be a need for mental health professionals to support these patients and their families. Furthermore, how to manage VUS in population screening remains unresolved, though there is growing consensus that VUS should not be reported in screening contexts (Murray et al., 2021; Burke et al., 192022). An alternative approach is to examine strategies for return of VUS findings, reclassification, and follow-up, a focus of current investigation.

Equity is an important consideration. There are currently disparities in access to and outcomes of genetics services. Racialized and underserved populations often have lower referral rates, differential rates of service uptake, more frequent misdiagnoses or inconclusive test results, older age and more advanced disease stage of diagnosis, and higher mortality rates (Armstrong et al., 2005; Maddison et al., 2011; Cragun et al., 2015; Kerner et al., 2015; Purificacion et al., 2015; Manrai et al., 2016; Vohnout et al., 2016; Amrock et al., 2017; Landry and Rehm, 2018; Muller et al., 2018; Hendricks-Sturrup and Lu, 2019; Ndugga-Kabuye and Issaka, 2019; Ehrenberg et al., 2021). These disparities are present worldwide, highlighting the pervasiveness of health inequities and an urgency for strategies to address them prior to adoption of population screening, to avoid exacerbating these issues. In addition, many underserved populations have limited guidelines on risk factors or treatment recommendations, making it difficult for clinicians to provide appropriate and effective care (Hann et al., 2017). For example, there is a scarcity of guidelines for breast cancer screening in transgender individuals undergoing gender-affirming hormone therapy (Berro et al., 2020; Rolle et al., 2021).

Furthermore, availability of risk-reducing strategies is not consistent across jurisdictions. For example, the extent (if any) of reimbursement for these interventions will vary by healthcare systems, leading to out-of-pocket costs for high-risk individuals, likely exacerbating existing inequities for underserved populations and undermining the effectiveness of the screening program.

Our review has several limitations. This was not a systematic review, nor was a formal quality appraisal of studies conducted. Moreover, this review was limited to Tier 1 conditions-future research and evidence synthesis will be needed to address other actionable gene-condition pairs (e.g., other genes for hereditary breast and ovarian cancer including PALB2, RAD51C, RAD51D, and BRIP1 (Manchanda et al., 2018); TTR for hereditary transthyretin amyloidosis (Soper et al., 2021); endocrine tumour genes (Savatt et al., 2022); arrhythmia syndrome genes (Walsh et al., 2022)) and their suitability for population genomic screening.

Despite these limitations, our review suggests that there is evidence that population genomic screening for HBOC, LS, and FH would improve detection of individuals with pathogenic variants in the causative genes compared to traditional approaches to case ascertainment. For outcomes beyond detection rate, HBOC has the strongest support for population genomic screening, with evidence demonstrating clinical and cost-effectiveness in the general population; real world implementation studies in the general population are needed. In the AJ population, there is substantial evidence on acceptability, satisfaction, different models of implementation, psychological/quality of life outcomes, uptake of preventive strategies, and cost-effectiveness in support of population BRCA1/2 screening.

LS and FH both have preliminary evidence supporting population genomic screening, but major gaps remain in the literature. For FH, although there is evidence suggesting population genomic screening programs would have clinical and cost-effectiveness, the evidence on long-term outcomes is limited. Furthermore, the evidence on cost-effectiveness is limited to modelling studies. Real-world studies establishing cost-effectiveness and clinical effectiveness over longer follow-up periods are needed. Economic models suggest population genomic screening for LS may only be cost-effective at a very high cost-effectiveness threshold. Further evidence is critical to establish clinical effectiveness of screening for LS in asymptomatic individuals and cost-effectiveness in lower- and middle-income jurisdictions.

In addition to filling in the evidence gaps, ethical concerns such as potential overdiagnosis, as well as issues related to equity and access to testing and follow-up interventions will need to be considered at the program design stage. Adoption of population genomic screening will require major restructuring and investments to scale up the workforce, build capacity in non-genetics providers, adapt alternative delivery models (mainstreaming, digital tools), optimize IT infrastructure and prioritize an approach that is inclusive of historically underrepresented populations to ensure the full potential of population genomic screening can be realized.