Contemporary Insights Into the Genetics of Hypertrophic Cardiomyopathy: Toward a New Era in Clinical Testing?

Abstract Genetic testing for hypertrophic cardiomyopathy (HCM) is an established clinical technique, supported by 30 years of research into its genetic etiology. Although pathogenic variants are often detected in patients and used to identify at‐risk relatives, the effectiveness of genetic testing has been hampered by ambiguous genetic associations (yielding uncertain and potentially false‐positive results), difficulties in classifying variants, and uncertainty about genotype‐negative patients. Recent case‐control studies on rare variation, improved data sharing, and meta‐analysis of case cohorts contributed to new insights into the genetic basis of HCM. In particular, although research into new genes and mechanisms remains essential, reassessment of Mendelian genetic associations in HCM argues that current clinical genetic testing should be limited to a small number of validated disease genes that yield informative and interpretable results. Accurate and consistent variant interpretation has benefited from new standardized variant interpretation guidelines and innovative approaches to improve classification. Most cases lacking a pathogenic variant are now believed to indicate non‐Mendelian HCM, with more benign prognosis and minimal risk to relatives. Here, we discuss recent advances in the genetics of HCM and their application to clinical genetic testing together with practical issues regarding implementation. Although this review focuses on HCM, many of the issues discussed are also relevant to other inherited cardiac diseases.

H ypertrophic cardiomyopathy (HCM) is an inherited disease of cardiac muscle characterized by substantial heterogeneity in morphology, clinical manifestation, genetic etiology, and outcome. 1 HCM is diagnosed in the presence of a maximum left-ventricular wall thickness of at least 15 mm when this is not solely explained by abnormal loading conditions (such as hypertension and aortic stenosis), with other typical features including myocyte disarray and interstitial fibrosis. 2 Although HCM is considered a Mendelian disease with largely autosomal dominant inheritance, incomplete penetrance and variable expressivity are commonly observed in affected families.
As well as being a relatively common (estimated to affect 1 in 500 people) [2][3][4] and medically important condition, HCM has been an important model disease for studying issues that are shared among many inherited cardiac diseases. Research into the genetic basis of HCM has been ongoing for almost 30 years and early implementation of clinical genetic testing has yielded extensive data sets for use in research studies. However, several issues have limited the impact of HCM genetic testing in clinical practice: 1. Over half of patients tested have a negative genetic test result (ie, no definitive or putative pathogenic variant is detected). 2. The significance of many detected variants has been difficult to ascertain. 3. The limited correlation between genotype and clinical phenotype has constrained the use of genetic knowledge in directing clinical management. 4. The ever-increasing number of genes apparently associated with disease has clouded our understanding of the genetic architecture of HCM.
Despite these open challenges, recent findings have transformed our understanding of the genetic basis of HCM and offered the potential for more accurate, sensitive, and impactful application of genetic testing. These findings have been driven by the development of population genetic databases that have revolutionized our understanding of rare genetic variation in health and disease and by the accumulation of large patient cohorts and disease registries. Collaborative models of data sharing, together with the development of consensus guidelines for variant interpretation and validation of disease-gene associations, have proven invaluable for improving our knowledge of rare genetic diseases.
Here, we discuss the achievements earned in understanding the genetics of HCM in the past 30 years, the clinical implications of the current reassessment of genetic associations, the need to manage the genotype-negative HCM patient in a coherent and effective manner, and the challenges that need to be addressed before we can adopt a patient-specific, individualized approach in HCM. While focusing on HCM, the issues discussed are also broadly relevant to our understanding of other cardiomyopathies and inherited cardiac diseases.

GENETIC ASSOCIATIONS
HCM was first recognized over 50 years ago as a familial myocardial disease with increased risk for sudden death, variable disease expressivity, and natural history. 5,6 After more than 2 decades during which the cause of HCM remained unknown, the dawn of a molecular era arrived in 1989, when the first genetic locus (14q1) associated with the condition was detected by means of a linkage analysis on a large kindred group comprising 96 individuals and spanning 4 generations. 7 Shortly after, the associated genomic region was narrowed further by additional linkage studies on the same family, restricting the region containing the genetic defect to locus 14q11-12, 8 and consequently defining a potential causal role for one of the two highly homologous genes encoding myosin heavy chains: MYH6 and MYH7. The responsible gene was identified the same year as MYH7, with the identification of the causative missense substitution p.Arg403Glu. 9 The same month, another study from the same research group was published to demonstrate the genetic heterogeneity of HCM, as cosegregation of variants at the MYH7 locus with HCM was observed in only 2 of 4 affected pedigrees, highlighting the existence of alternative genetic causes at other loci. 10 After these pivotal discoveries, the genetic repertoire of HCM was gradually expanded throughout the decade to include seven additional sarcomeric genes (MYBPC3, TNNT2, TPM1, MYL2, MYL3, TNNI3, ACTC1) [7][8][9][10][11][12][13][14][15][16][17][18][19] (Table 1), incontrovertibly associated with HCM by studies based on linkage analyses in large and/or multiple affected families and functional experiments. We refer to these eight "core" disease genes with the term "validated genes" in this review. Simultaneously, definitive genetic associations with genocopies of HCM (ie, metabolic diseases characterized by a phenotype mimicking HCM but caused by genetic variants in different genes) were demonstrated, starting with Pompe disease [20][21][22][23][24][25][26][27][28][29][30][31][32][33] (Table 2). Besides Pompe disease, Fabry disease, PRKAG2-cardiomyopathy, Danon disease, and TTR-amyloidosis, other conditions (due to mitochondrial dysfunctions) can seldom present resembling isolated HCM, such as Friedreich's ataxia and Leigh syndrome, caused by variants in the FXN and COX15 genes, respectively. Improvements in sequencing technologies and their throughput, particularly with the advent of nextgeneration sequencing (NGS), revolutionized the ability to investigate the genetic architecture of Mendelian diseases such as HCM. Case-control rare variant comparisons became the design of choice for most studies, where usually relatively small groups of patients and controls were sequenced on one or a few genes of interest, encoding proteins within or outside the sarcomere, implicated in HCM with varying levels of supportive evidence 48,49 (Figure 1). Whenever a protein-altering variant was observed uniquely in patients, the absence of such variants in controls was often used as stand-alone evidence in support of the claim for pathogenicity, with little convincing segregation or functional evidence in support. Although at the time these approaches were generally considered sufficient to implicate genes in Mendelian disease, it is now recognized that more substantial evidence is necessary to demonstrate association between variants in a given gene and disease.

Nonstandard Abbreviations and Acronyms
The consequence of these efforts was an exponential increase in the number of proposed genetic associations in the years between 2000 and 2015 ( Figure 2). By 2016, 64 genes had been claimed to be causative in HCM (Human Gene Mutation Database v.2016.3, Table S1) with few of the additional genes being supported by sufficiently comprehensive and robust evidence [50][51][52][53][54][55][56][57] (Table 3).
Population databases of whole-exome sequencing (WES)-and in part, whole-genome sequencing (WGS)-such as the 1000 Genomes Project 58 (≈2500 individuals) and National Heart, Lung, and Blood Institute Exome Sequencing Project 59 (ESP, ≈6500 individuals) enabled early observations that many variants previously associated with cardiomyopathies were instead likely benign, as their population frequencies were incompatible with the prevalence of disease. 60,61 The definitive proof of this arrived in October 2014 with the release of the Exome Aggregation Consortium (ExAC) data set, a large collation of WES data from over 60 000 individuals, 62 in which the average individual was shown to carry ≈54 variants previously reported as pathogenic in widely used databases of disease-causing variants.
In the case of HCM and dilated cardiomyopathy (DCM), large-scale analysis based on ExAC demonstrated that the frequency of rare variants found in many genes implicated in the condition was comparable to background population variation, 49,63,64 casting doubts on the veracity of many purported associations ( Figure 2). In addition, a number of independent disease-and gene-specific approaches have recently been developed to aid decisions on which genes to include in clinical diagnostic panels. One recent example is the diagnostic effectiveness (DEff), 65 a gene-and disease-specific metric combining measures of actionability for cascade screening and prior likelihood of pathogenicity of variation detected in patients. DEff scores of validated and putative disease genes can be compared, in order to assist the decision on which of the latter should be routinely assessed in clinical practice and which would yield, at most, inconclusive results (variants of uncertain significance). Rigorous, disease-specific gene curation efforts that assess all available lines of evidence for disease association are now essential for genetically heterogeneous Mendelian diseases like HCM. In spite of the lack of a standardized approach to weight different types of evidence for gene-disease associations, results obtained by curation/classification efforts are characterized by general concordance, with the 8 validated genes unanimously considered as incontrovertibly associated with HCM, and genes such as PLN and CSRP3 supported by strong evidence in favor of HCM pathogenicity. 48,49,65 The ClinGen initiative 66 (https://clini calge nome.org/)-founded to define the clinical relevance of genes and genetic variants-has recently established a framework to systematically assess all gene-disease claimed associations' veracity and their clinical validity based on published evidence. The curation effort for HCM, assessing >50 genes, has recently classified evidence in support of their role in HCM as "definitive" for the 8 validated sarcomeric genes, PLN and FLNC (also associated with myofibrillar myopathy), "strong" for ALPK3 and "moderate" for ACTN2, CSRP3, TNNC1 and JPH2 (Tables 1 and 3). 48 Other 20 genes were assigned a ≥ moderate evidence classification for their role in genocopies of HCM (Table 2) or syndromes involving HCM.

GENETIC TESTING STRATEGIES
The primary purpose of genetic testing in HCM is to facilitate cascade screening in relatives of the index patient in order to identify those at risk and those not at risk of developing disease. As guidelines currently recommend regular clinical evaluation for first-degree relatives of all HCM patients, 2 discharging those relatives lacking identified pathogenic variants offers obvious psychological benefits for these individuals and significant cost savings for healthcare providers. Genetic testing for HCM has been shown to be more cost effective than clinical screening alone in UK 67 and Australian 68 studies, and the Partners Laboratory for Molecular Medicine (LMM) estimated savings of USD 700 000 (based on Medicare rates) from discharging genotype-negative relatives of 2912 HCM probands. 69 Ongoing reductions in the costs of DNA sequencing, and improvements in variant analysis as described subsequently, will further support the economic case for HCM genetic testing. It has also been recognized that genetic testing in probands can increase the often low take-up in clinical screening from at-risk relatives. 70 In contrast, knowledge of the causative genetic variant does not currently affect the clinical management for most HCM patients and is not included in sudden cardiac death (SCD) risk assessment models. 71,72 The exception to this is identifying patients with pathogenic variants in a small group of genes-such as GAA, GLA, LAMP2, PRKAG2 and TTR-that are incontrovertibly associated with metabolic diseases that mimic HCM ( Figure 1 and Table 2) but have distinctive clinical profiles, inheritance patterns, and treatment options. 73 The inclusion of these genes in diagnostic panels for HCM is advantageous given their pronounced phenotypic similarity with the classic sarcomeric form 74 and the importance of a prompt differential diagnosis, enabling correct treatment decisions and optimal patient management and de facto representing a significant step toward more personalized approaches.
Despite limited evidence for a causal role in HCM, many of the non-sarcomeric genes implicated in the disease have been progressively included in panels developed by clinical laboratories and commercial providers 75 (listed in the National Center for Biotechnology Information Genetic Testing Registry: https://www.ncbi. nlm.nih.gov/gtr/). The uncertainty around the pathogenicity of these "genes of uncertain significance" in HCM has led to difficulty for clinical diagnostic laboratories in choosing which of the ambiguous genes to include in routinely used diagnostic panels. The choice ultimately lies on the spectrum between the extremes of including only the validated genes at one end (bottom left and bottom right panels in Figure 1) and systematically screening all genes implicated in HCM at the other, irrespective of the strength of evidence in support of their causal role for  Tables 1 through 3. HCM, hypertrophic cardiomyopathy. Downloaded from http://ahajournals.org by on  HCM. The former, conservative choice plausibly leads to the possibility of false negative genetic tests but the latter, inclusive option, bears greater potential for false positive results, which can be detrimental for the patients and their family members, and increases the uncertainty associated with testing through an epidemic of variants of unknown significance (VUS). Unsurprisingly, expanding HCM panels to include genes of questionable pathogenicity has had little effect on the sensitivity of genetic testing. For example, LMM reported identifying only one pathogenic variant in 632 cases (in the PLN gene) outside of the 8 core sarcomeric genes and metabolic cardiomyopathy genes. 69 Ongoing reductions in the cost of sequencing have meant that WGS (complete sequencing of the entire genome in an individual) and WES (sequencing of the 1% of the genome that encodes proteins) may become feasible options as first-line sequencing assays in the near future. However, there is limited evidence that WGS and WES are currently cost effective in clinical practice. 76 For example, a recent study that was part of the Medseq initiative to evaluate the clinical potential of WGS 77 compared the performance of a targeted gene panel and WGS in 41 HCM patients. WGS identified 19 of the 20 variants of interest detected by panel sequencing, but missed a 18 bp duplication in MYBPC3 due to low coverage and only additionally identified a causative variant in a RASopathy gene (not in the panel) as well as several VUS and secondary findings. The limited value of WGS, WES, or super-sized gene panels for standard HCM genetic testing, that is, adult-onset and/or autosomal dominant disease in the absence of The black line indicates the number of scientific publications in PubMed featuring next-generation sequencing (colored lines, right y axis). The classification criteria for evidence in favor of pathogenicity (as detailed in the legend) and the genes included (n=39) reflect the analysis and curation effort by Walsh et al. 49 HCM, hypertrophic cardiomyopathy; and NGS, next-generation sequencing. Downloaded from http://ahajournals.org by on  large family pedigrees, is clear when considering current variant interpretation guidelines for clinical testing and the requirements for defining a variant as actionable (see next section). It will be impossible to classify a variant in a novel gene as clinically actionable in the vast majority of cases, regardless of how plausible that candidate gene may be, as the guidelines require as a minimum that the gene or variant class has a proven role in the disease.
In contrast, WGS/WES can offer some advantages as diagnostic sequencing assays if testing is restricted to genes and regions that have been validated for the disease in question. In this "virtual panel testing" approach, analysis is restricted to only those genes that can yield interpretable variants, but the extensive sequencing enables reanalysis of archived data as new and validated gene-disease associations are published, as well as potentially generating comprehensive and useful data sets for research purposes. One advantage of WGS compared with approaches targeting only protein-coding regions of genes (eg, targeted gene panel and WES) is its enhanced ability to identify variants classes such as copy number variants (CNVs)-deletions or duplications of whole exons, genes, or chromosomal regions-and variants in noncoding regions of the genome such as introns. Studies exploring the role of CNVs in HCM have found such variation in ≈1% of cases, [78][79][80] although detecting, validating, and interpreting CNVs remains a challenging and nonstandardized task. A recent study used WGS to identify deep intronic variants in MYBPC3 in 4 HCM patients who were negative for standard genetic testing, and these variants were shown to affect splicing and produce truncated transcripts. 81 The contribution of these noncoding variants that affect gene dosage is likely to be particularly relevant to genes where the underlying disease mechanism is haploinsufficiency (ie, where the loss of half of the physiological amount of functional protein causes disease). This is often the case for genes in which protein-truncating variants are the most prevalent pathogenic variant class although, in the case of MYBPC3, haploinsufficiency has also been proposed as the molecular mechanism for several missense variants. 82 For most of the other major sarcomeric HCM genes where variants act in a dominant negative manner, such noncoding variants are expected to be less relevant.
In addition to these scientific pros and cons, other factors such as affordability and the quality and technical features of generated data are also important factors in choosing a sequencing platform for genetic testing. 83,84 As highlighted previously, depth of coverage for affordable WGS/WES in the diagnostic setting may currently be inferior to panel sequencing and lead to false negative findings-this is particularly relevant for WES assays as some exons, especially in GC-rich regions, are not adequately captured. 85 It is also important to recognize the extra burden of data processing, analysis, and storage associated with WGS/WES, as well as ethical issues relating to secondary and incidental findings if analysis is not restricted to valid disease genes for the primary condition.
At present, the primary role for unbiased WGS/WES lies in the research setting in order to identify putative  56 Taking all of these issues into consideration, an effective tiered approach to genetic testing in HCM is shown in Figure 3. Analysis should be initially focused on validated HCM genes, regardless of the sequencing assay applied (gene panel or focused analysis of WGS/WES). If negative, the decision to expand analysis to novel genes (with WGS or WES) would depend on the profile of the patient being tested and the likelihood of isolating the causative variant from genome-wide sequencing. For most adult probands, it is unlikely that pathogenic variants will be identified in genes not previously associated with HCM. In any case, as described subsequently, it is increasingly likely that most such patients represent non-Mendelian and nonfamilial HCM.

INTERPRETATION OF GENETIC FINDINGS
Determination of the clinical significance of genetic variants is a complex process involving the evaluation and weighting of several lines of evidence. Accurate interpretation is key, given the potentially detrimental consequences that an erroneous interpretation can have on the proband and their relatives. 86,87 The American College of Medical Genetics and the Association of Molecular Pathologists (ACMG/AMP) first issued standardized recommendations on the interpretation of sequence variation almost 2 decades ago, 88 and a series of updated and increasingly articulated guidelines have been issued in the following years, in parallel with the establishment and the exponential growth of clinical and population sequence variation databases ( Figure 4).
The increasing complexity of variant interpretation and the extent of misclassification in previous years underscored the need for more stringent and comprehensive approaches to variant classification, prompting the release of the current ACMG/AMP guidelines in 2015 89 (Figure 4). These guidelines, considerably more complex than previous recommendations, are Figure 3. Proposal for a tiered approach to genetic testing in HCM. Initial testing is restricted to validated HCM genes using either a targeted gene panel or a virtual panel by focused analysis of WGS or WES data. The latter allows for reanalysis of genes subsequently linked to HCM, and WGS can detect additional variant classes including CNVs and deep intronic splice variants, though sequencing coverage is likely to be inferior to targeted panel sequencing. In the case of negative results, variant prioritization in the research setting can aid identification of the causative variant, directly translating into a diagnostic finding. The likelihood of detecting variation in novel genes with broader WGS/WES analysis depends on the profile of the patient and family being tested. In general, large pedigrees informative for segregation and trios (affected child with unaffected parents, indicating likely de novo variant occurrence or recessive inheritance) can enable filtering to a manageably small number of potentially causative variants for detailed evaluation. CNV, copy number variant; HCM, hypertrophic cardiomyopathy; WES, wholeexome sequencing; and WGS, whole-genome sequencing.

Mazzarotto et al
The Emerging Clarity of HCM Genetics designed to be applicable to variants in all Mendelian genes, and to serve as a basis for more detailed and focused implementation for specific genes and diseases. As a result, several rules embedded in the current interpretation framework avoid providing specific instructions on how to weight certain classes of evidence, inevitably giving rise to differences in approaches adopted by different laboratories and discordant variant classification. 90,91 For this reason, besides the development of automated, guideline-based variant interpretation tools, 92,93 the implementation of consensus guideline adaptations is now an active area of research, both for different evidence categories and for specific disorders or gene-disease pairs. Examples include proposals for weighting the evidence of cosegregation of variants with disease in pedigrees 94 and for a more finely tuned interpretation of loss-of-function variants, 95 quantitative scores estimating the likelihood of pathogenicity based on the variant location within the protein sequence, 96 an algorithm to define a diseasespecific population frequency cut-off for potentially pathogenic variants. 97 More complex approaches taking into account the variant position within the threedimensional structure of the protein (as opposed to its one-dimensional sequence) have also been used to identify protein regions enriched in HCM-causing variants, 98 and could augment variant interpretation accuracy if embedded in the variant interpretation framework. For specific genes/diseases, tailored recommendations to interpret genetic variation have been developed for RASopathies 99 and for MYH7 variants in HCM, 100 with ongoing efforts for other genes and diseases by the ClinGen initiative and other groups. A description of ACMG/AMP guidelines' rule-specific application to variants detected in HCM patients, current issues, and ongoing/future developments toward an improved application are provided in Table 4. The current guidelines are designed to minimize false-positive results (ie, over-calling of variant pathogenicity) as it is usually less detrimental for patients and their relatives to receive an inconclusive result in presence of a pathogenic variant than it is to receive a positive result in absence of a real genetic cause. While assigning variants a categorical classification, guidelines essentially follow an underlying Bayesian framework, with a variant considered "likely pathogenic" if its probability of pathogenicity is ≥90%, and "pathogenic" if ≥99%. 101 However, the lack of variant-specific information available to diagnostic laboratories for many rare novel variants often translates into the inability to correctly identify those with a probability of pathogenicity ≥90%, de facto rendering the application of the guidelines very stringent and causing a general undercalling of pathogenicity. Although this conservative approach offers the advantage of minimizing false-positive variant classifications, this inevitably comes at the cost of an increased rate of false negative results (ie, where pathogenic variants are classified as VUS). This drawback is particularly relevant in HCM because of its marked genetic heterogeneity with many variants private to affected families, the fact that most causative variants are missense substitutions which are inherently more difficult to interpret than protein-truncating variants and occur with higher background rates in controls, and the limitations of functional assays to effectively evaluate their molecular effects unlike, for example, those in genes encoding ion channels. Recent evidence has quantified this false negative rate for HCM, suggesting that ≈8% of HCM patients who undergo clinical genetic testing will receive a reported VUS that is likely to be a pathogenic variant 63 (ie, a VUS in one of the 8 sarcomeric PM2-variant is rare enough in the population to be plausibly pathogenic (also BA1, BS1, BS2).
Population frequency data from gnomAD and disease-specific threshold based on HCM characteristics provide stringent variant rarity threshold. 97 1. Low penetrance variants or modifiers may be less rare in the population than standard pathogenic variants, requiring evidence from other rules to achieve (likely) pathogenic classification. 2. Some population groups are still not well represented in databases like gnomAD (eg, North Africa and West Asia), variants from patients from these groups need to be analyzed with caution. genes was detected in 12.4% of HCM patients compared with a background rate of rare variation of 4.3% in the ExAC population data set). In support of this, analysis of the SHaRe registry (https://thesh arere gistry.org/), comprising genetic data and cardiac morphofunctional parameters for >9000 HCM patients), showed that variants classified as VUS have a measurable clinical impact on outcome, 102 highlighting how these include misclassified pathogenic variants. In order to overcome these obstacles to variant interpretation, knowledge sharing between different centers through databases such as ClinVar (https:// www.ncbi.nlm.nih.gov/clinv ar/) is critical. As an example, the MYH7 p.Arg869His variant had conflicting interpretations in ClinVar (VUS or likely pathogenic) based on the variant characteristics and some published reports. However, this variant is observed at locally high frequencies in northern Tuscany (detected in 30 of 1198 unrelated HCM probands tested at Careggi University Hospital compared with 0 of 356 non-HCM patients affected with other cardiovascular conditions, P=6.22×10 −4 ), including co-segregation with disease in 7 HCM patients in a three-generation family. Based on this evidence (detailed in ClinVar accession SCV001147002.1), the MYH7 p.Arg869His variant can now be classified as definitively pathogenic for HCM and highlights the benefits of sharing findings from different populations and geographic regions.
The role of both public and private diagnostic laboratories is key, as they represent the "hubs" where many novel variants are observed and where patients and their families are referred when a genetic cause of overt or suspected disease is investigated. Many diagnostic laboratories (such as Invitae and GeneDX) comprehensively publish variant interpretation summaries in ClinVar, based on published data and inhouse findings, which helped ClinVar recently reach the milestone of 1 million submissions. Nevertheless, some diagnostic companies still refuse to share data in this manner and instead produce proprietary variant interpretation that takes advantage of public resources such as ClinVar and gnomAD in addition to their own databases. Given the arduous challenge of variant interpretation for genetically heterogeneous diseases like HCM, and the importance of accurate classification in clinical practice, the ethics of such practices are dubious. One possible solution would be for healthcare providers and cardiologists/clinical geneticists to insist on submission to ClinVar when engaging the services of diagnostic laboratories or only employ companies committed to knowledge sharing.
Data sharing initiatives are at the basis of several approaches that aim to tackle the high rate of inconclusive findings in clinical genetic testing. The availability of increasingly large patient and control cohorts may enable the development of quantitative statistical approaches to identify variant classes with high likelihood of pathogenicity. For example, casecontrol analyses can estimate these prior probabilities of pathogenicity through metrics such as the odds ratio and etiological fraction. 63 A recent study using this approach, which defined regions of genes where case variants are significantly enriched and incorporated this evidence to the ACMG/AMP framework, increased the relative yield of clinically actionable variants in sarcomeric genes by up to 20% in HCM. 96 More comprehensive classification algorithms based on ensemble-or machine-learning techniques that integrate different classes of evidence may also plausibly improve the accuracy of variant classification and contribute to lowering the proportion of inconclusive findings in HCM and other diseases. 103,104 Advanced computational techniques like these may face considerable hurdles though with integration into current guidelines as well as the workflows of clinical genetics laboratories.
Assays that accurately assess the functional effects of specific genetic variants, through cell-based or in vivo models, also offer the possibility of informing the pathogenicity of variants without the need for segregation evidence from large families. However, the lack of effective and standardized translational assays has limited the role for this class of evidence for cardiomyopathies, with the cost and time to implement such assays also rendering them impractical in the context of clinical genetic screening. Novel techniques, in particular those using the CRISPR/ Cas9 genome editing technology, are now being applied for distinct alleles in HCM 105-108 and DCM. 109 The further development of such approaches into more comprehensive and high-throughput assays, such as a recent study that functionally evaluated thousands of alleles in BRCA1 110 simultaneously, may be required for such evidence to be routinely incorporated into clinical genetic testing.
Improvements in variant classification are especially relevant to non-white populations, which are characterized by significantly lower detection rates of (likely) pathogenic variants in cardiomyopathy patients compared with white patients. 111 This is likely to result from the fact that the vast majority of research efforts and genetic tests for cardiomyopathies have been carried out on patients and controls of European descent (and therefore recurrent and founder variants occurring in whites are more likely to have been previously observed and characterized). This is emblematic of a wider problem in genetics research, exemplified by the proportion of genomewide association studies (GWAS) performed on individuals of European descent (≈85% of the total) being vastly disproportionate to their proportion of the world population (≈15%). 112 These observations

Mazzarotto et al
The Emerging Clarity of HCM Genetics underscore the urgent need for more genetic research to be undertaken in non-European populations as well as the further development of functional and quantitative approaches that reduce reliance on the prior observation and characterization of rare pathogenic variants. The accurate interpretation of variants in HCM genes is equally critical when they have been detected as secondary or incidental findings by WGS or WES in individuals undergoing genetic testing for noncardiac conditions. Each of the 8 core sarcomeric HCM genes are on the list of genes recommended by the ACMG to be analyzed as secondary findings. 113 Given the relatively high rate of rare variation in these genes in the population, only those variants with strong evidence of pathogenicity should be considered actionable. However, as many HCM variants display reduced penetrance, clinical evaluation to detect the cardiomyopathy phenotype is essential, including ongoing surveillance screening where appropriate. 114

INTERPRETATION OF GENOTYPE-NEGATIVE HCM
When variants in sarcomeric genes were first implicated in HCM, there were indications that knowledge of the causative variant might be prognostic for patients and potentially useful for clinical decisionmaking. For example, TNNT2 variants were initially shown to be associated with poor prognosis and high incidence of sudden cardiac death (SCD), 115 whereas variants in MYBPC3 were related with delayed onset of disease and more favorable outcomes. 116 Subsequently, genotype-phenotype analyses on ever-larger cohorts of HCM patients have been undertaken to clarify these initial findings and attempt to identify clear and consistent correlations between the pathogenic variant class and the clinical profile and outcomes in patients.
Recent cohort studies have identified some statistically significant differences in phenotype between patients carrying different variant classes. In particular, HCM patients with variants in thin filament genes (TNNT2, TNNI3, TPM1, ACTC1) were shown to have distinctive patterns of left-ventricular hypertrophy and diastolic profile, as well as lower prevalence of leftventricular outflow tract obstruction (LVOTO) and late gadolinium enhancement (a marker of myocardial fibrosis) compared with those with variants in thick filament genes (MYH7, MYBPC3). 117 Although patients with variants in thin filament genes were more likely to develop advanced LV dysfunction, no differences were observed in mortality or SCD between thin and thick filament variant-positive patient groups in this study. 117 Studies comparing outcomes between patients with MYH7 and MYBPC3 pathogenic variants (the predominant genes associated with HCM) have yielded conflicting results-no significant differences were observed for 2 moderately sized cohorts 118,119 but in the larger SHaRe registry those with MYH7 variants had a higher risk of advanced heart failure and overall composite outcome. 102 Furthermore, patients with complex genotypes carrying 2 pathogenic mutations were at greater risk of adverse outcome and sudden cardiac death, irrespective of the genes involved. Despite these broad correlations, all of these studies have revealed extensive heterogeneity in symptoms, disease severity, and outcome, even among patients with the same or similar pathogenic variants. As a result, the individual patient's genetic status has limited impact on clinical management and is largely limited to cascade screening and to the small subsets of patients with rare storage or metabolic diseases mimicking HCM.
Genotype-positive patients present approximately a decade earlier and unsurprisingly have a greater proportion of family history of both HCM and SCD. Maximum left-ventricular wall thickness (LVWT) is significantly greater in genotype-positive patients, who are characterized by more asymmetrical hypertrophy than genotype-negative cases. Most important, these differences translate into poorer outcomes for genotypepositive patients across a range of measures and composites in a number of studies, including the 4591 cases of the SHaRe registry. 102 In these studies, genotype-negative cases were defined as lacking pathogenic (or potentially pathogenic, ie, rare and protein-altering) variants in the tested HCM genes (for the most part the 8 validated sarcomeric genes). The expectation that at least some of these cases may have novel genetic causes drove the many attempts to discover novel HCM genes, as described previously. However, the non-sarcomeric genes whose role in HCM has been validated thus far account for only a small proportion of HCM patients, with cumulative weight of evidence now suggesting that genotypenegative HCM is often nonfamilial and most likely to represent non-Mendelian disease.
Ingles et al 123 recently defined "non-familial HCM" as the condition that lacks both a potentially pathogenic sarcomeric variant (whether classified as pathogenic or VUS) and any prior family history for HCM. Nonfamilial cases (which accounted for 40% of the cohort in the study) were phenotypically distinct Downloaded from http://ahajournals.org by on May 21, 2020 Genotype-positive status is defined by the presence of a putatively pathogenic variant-the vast majority of these occur in one of the 8 core sarcomeric genes (MYH7, MYBPC3, TNNT2, TNNI3, TPM1,  from genotype-positive cases (though in a similar manner to overall genotype-negative patients) and with very low yield from cardiac clinical screening on first-degree relatives. Consistently, Ko et al 127 found that only 3% of first-degree relatives of "nonfamilial" probands were diagnosed with HCM by clinical screening, compared with 17% of relatives of probands with a positive genetic test and/or prior family history of HCM. As familial occurrence of HCM is still nonnegligible in pedigrees without an identified causative variant, current clinical recommendations state that regular clinical screening of family members should be considered for relatives in such families. However, the actual risk of disease in families without prior family history of HCM, and the consequent necessity for ongoing clinical screening in phenotypenegative relatives, remains uncertain. In this respect, it is worth emphasizing that history taking in HCM is a complex task due to the high variability in terms of age of onset and phenotype severity characterizing this condition. In contrast to other conditions in which the phenotype usually manifests clearly and penetrance is not age related, in the case of HCM the adoption of specific strategies to maximize precision and completeness of the information gathered from the proband and his/her relatives may help to reconstruct a reliable clinical history of the family. As an example, in addition to applying the principles detailed in internationally accepted guidelines, 114 at first clinical appointment the cardiologist could also ask for information regarding family history, so to complement and confirm those gathered by the genetic counsellor. Furthermore, it may be good practice to ask patients to share a copy of all clinical documents regarding important clinical events such as hospitalizations or cardiovascular events, due to the fact that the knowledge of family members on the cause underlying such events is generally limited.
Further large-scale studies are required to definitively assess the risk in relatives of index cases with apparently nonfamilial HCM and to accurately define the criteria to establish family history of the disease. Such studies may lead to modification of current guidelines on the extent and frequency of clinical screening recommended for relatives of HCM patients, based on more accurate assessments of their risk of developing disease depending on genetic status and family history. This may offer the opportunity for additional significant savings in healthcare provision and will further underscore the utility and cost-effectiveness of genetic testing for HCM.

FUTURE DIRECTIONS
Almost 30 years after the first definitive evidence of the genetic etiology of HCM, 7 great progress has been achieved and the genes underlying the majority of Mendelian monogenic cases of HCM have now been identified. It is clear that HCM remains essentially a disease of the sarcomere, characterized by reduced penetrance and highly variable expressivity. Although additional causative genes may be found, these will most likely explain only a minor proportion of cases, as demonstrated by the few non-sarcomeric genes now deemed validated for HCM.
Recent findings and recommendations now point to a more focused application of genetic testing in HCM. Genes of unknown significance, whose association with the disease remains unproven, should be excluded from diagnostic testing as they offer a negligible increase in diagnostic yield but considerably increase the background noise of VUS. 25,29,33,103 Conversely, more comprehensive and focused analysis of known disease genes is likely to yield greater results than attempts to discover novel disease genes. This will include improved classification techniques for coding variants and detection of variant classes historically considered to be of low pathogenic potential, including intronic variants affect splicing, which recent research has highlighted are sometimes irrefutably pathogenic and could therefore contribute to resolving an additional proportion of unexplained disease cases. 81,128 A large proportion of HCM cases still remain without an identified causative variant, with 40% to 60% of screened individuals receiving an inconclusive or a negative test result. 49,65,102,129,130 More complex genetic models for HCM are increasingly believed to hold the answer to the majority of these genetically elusive cases. These are still largely unexplored and could include a polygenic model, where multiple rare variants cause disease when carried by the same individual but are benign in isolation, and complex disease model combining the effects of common variants of low effect size with nongenetic factors such as hypertension and obesity. However, investigations into such models have so far been limited by the lack of availability of case and control cohorts large enough to reach satisfactory statistical power. To date, the only GWAS reported for HCM comprised 174 patients and 823 controls, and detected a significant association between a common intronic variant in the FHOD3 gene on chromosome 18 and the presence of disease. 131 Such efforts will be increasingly possible as single centers are more easily able to sequence large numbers of patients, and through publicly accessible (though with variable data access policies) phenotyped and sequenced population cohorts such as those by the National Institutes of Health (the ongoing "All of us" research program that aims at 1 million genomes, and the dbGap database 132 ), the UK Biobank, 133 Genomics England 134 and the Cooperative Health Research in South Tyrol. 135 Downloaded from http://ahajournals.org by on  Such data sets, coupled with data sharing efforts such as the SHaRe Registry, will plausibly enable investigations on multi-genic and other complex disease models.
The lack of clear correlation between the genetic and clinical status of patients has until now dented hopes for the development of personalized medicine approaches, with genetic testing results still being used in a strictly diagnostic (rather than predictive) fashion. The incomplete penetrance of pathogenic sarcomeric variants, and the variable phenotypic expression in variant carriers, indicates that other genetic and non-genetics factors are likely to influence the development of disease. It has been observed that the presence of multiple rare sarcomeric variants has a cumulative effect on disease onset and incidence of events, regardless of the variants' diagnostic classification (ie, even when the variants could not be classified as disease-causing), 136 even though previous studies on multiple co-occurring sarcomeric mutations in HCM likely overestimated the pathogenicity of the individual variants. 137,138 This suggests that variants that are not pathogenic in isolation may act to modify and exacerbate the disease phenotype in presence of a disease-causing allele. Research into the full range of genetic factors that may interact with primary sarcomeric mutations to influence phenotype is at an early stage but is promising-such factors could include common, low frequency, and rare variants that add to the mutational burden, act as protective variants, or regulate the ratio of expression of mutated and wild-type alleles. 139 Identifying these factors through large-scale genomewide association studies and WGS, and integrating such a genetic profile of common and rare variants with known nongenetic modifying factors such as hypertension and obesity, offers the promise in the near future of improved risk prediction for HCM patients and their relatives and a more personalized approach to clinical care.

CONCLUSIONS
The path to a complete and detailed understanding of how genetic variants cause HCM and determine its severity will be long and challenging. Three major challenges remain unresolved and need to be addressed: 1. The still suboptimal accuracy of variant interpretation strategies, 2. The scarcity of clear genotype-phenotype correlations, and 3. The necessity to expand research on genetic causes of HCM beyond the monogenic Mendelian inheritance model.
Advances in our understanding of population genetics, the development of new functional genomics techniques and quantitative approaches to variant interpretation, data sharing of cohort data and evidence from clinical genetics laboratories, and the introduction of new standards for assessing gene and variant pathogenicity have all greatly improved our understanding of the Mendelian genetics of HCM.
Genetic testing can now be more rigorously and consistently applied for HCM patients, enabling the accurate identification of primary causative variants and at-risk individuals.
Our understanding of the more complex genetics that may underlie both the highly variable phenotypes observed in sarcomeric variant carriers and the genetic basis of currently genotype negative and likely non-Mendelian HCM is still at an early stage. It is likely that significant investment in sequencing and phenotyping, and further initiatives in developing international registries and consortia, will be essential for allowing us to make similar progress in understanding the complex genetics underlying HCM.