Genetic testing in ambulatory cardiology clinics reveals high rate of findings with clinical management implications

Abstract

Purpose:

Cardiovascular disease (CVD) is the leading cause of death in adults in the United States, yet the benefits of genetic testing are not universally accepted.

Methods:

We developed the “HeartCare” panel of genes associated with CVD, evaluating high-penetrance Mendelian conditions, coronary artery disease (CAD) polygenic risk, LPA gene polymorphisms, and specific pharmacogenetic (PGx) variants. We enrolled 709 individuals from cardiology clinics at Baylor College of Medicine, and samples were analyzed in a CAP/CLIA-certified laboratory. Results were returned to the ordering physician and uploaded to the electronic medical record.

Results:

Notably, 32% of patients had a genetic finding with clinical management implications, even after excluding PGx results, including 9% who were molecularly diagnosed with a Mendelian condition. Among surveyed physicians, 84% reported medical management changes based on these results, including specialist referrals, cardiac tests, and medication changes. LPA polymorphisms and high polygenic risk of CAD were found in 20% and 9% of patients, respectively, leading to diet, lifestyle, and other changes. Warfarin and simvastatin pharmacogenetic variants were present in roughly half of the cohort.

Conclusion:

Our results support the use of genetic information in routine cardiovascular health management and provide a roadmap for accompanying research.

INTRODUCTION

Cardiovascular disease (CVD) is the leading cause of mortality in the United States, leading to nearly 1 in every 4 deaths¹. The role of inherited susceptibility to CVD is well established, from rare monogenic disorders to polygenic traits. More than half of the genes on the American College of Medical Genetics and Genomics (ACMG) secondary finding (SF) list are associated with cardiovascular phenotypes². Although the utility of genetic testing in pediatrics is well established and precision medicine approaches can shift reactive care to stratified risk assessment with the potential to mitigate risk through adjustments of diet, lifestyle, or medications³, few parallel applications exist in the adult clinical care system.

Among the many obstacles to widespread adoption of adult genetic testing is the perceived lack of clinical utility and availability of point-of-care genetic testing during physician-patient interactions. Rare alleles with large effect are often considered too infrequent or else only able to provide information superseded by family history. In selected families, individuals will be referred to genetic testing as an adjunct to standard care, reflecting a division between clinical genetics care and other service branches. Conversely, common alleles are regarded as only able to contribute minor risk compared to etiological factors. The latter has been somewhat addressed by the development of polygenic risk scores (PRS) based upon the summation of overall genotype risk and the recognition that extreme scores can approach the predictive value of single alleles with high penetrance. Nevertheless, genetic testing’s practicalities remain limiting, and the integration of a genetic screen in a standard CVD care program is not commonplace.

We, therefore, developed “HeartCare”, a next-generation sequence-based DNA capture panel, to provide a genetic screen and to serve as the implementation tool to identify and address the practical issues involved in the delivery of genetic test results for CVD and related disease risk. The test consists of a 158 gene oligonucleotide capture-based panel, evaluating 1) Mendelian, actionable conditions including cardiomyopathies, aortopathies, arrhythmias, and dyslipidemias, 2) a coronary artery disease polygenic risk score (PRS), 3) variants in the LPA gene encoding Lipoprotein(a) that are an independent risk factor for atherosclerotic CVD events, and 4) pharmacogenetic variants contributing to simvastatin-induced myopathy and warfarin metabolism.

A long-term aim of this program is to bring together the clinical translation of medicine and the accompanying gathering of detailed phenotype records from many patients, using an infrastructure that is compliant for research. The HeartCare genetic screen has enabled the identification of patients with elevated risk and offers a model for actionable return of results utilizing established clinical service delivery lines in adult cardiovascular healthcare. Further, HeartCare data management has enabled data mining, analysis, and follow up in the well-characterized participant population.

MATERIALS AND METHODS

Participants

The study was approved by the Baylor College of Medicine (BCM) Institutional Review Board (protocol number: H-43884) and was open to all adult patients between the ages of 18–85 years at participating outpatient Baylor clinics (Table 1) at no cost to them, and insurance was not billed. Participants were informed of the study at the time of a clinical visit via posters and brochures in the waiting areas, study coordinators onsite, or physician referrals (workflow outlined in Fig. 1a). Interested patients were asked to give a blood or saliva sample and provide written consent to participate after undergoing pretest counseling. As the gene panel was offered to all patients, there were no required indications and there was no targeting of patient groups within a particular clinic. Family history for cardiac disease was also not required and was collected if voluntarily provided by participants. Indications for testing and race/ethnicity were self-reported, the latter based on US Census-standardized terms. The HeartCare panel was added to the local EMR (Epic) test menu for facile ordering and ease of incorporation of demographic and clinical data elements required to interpret results.

Table 1.

Demographics, referring clinics, and top 5 indications with positive result (pathogenic/likely pathogenic variant).

Characteristic		Participants (N=709)
Demographic Characteristics
Male sex - Number (%)		397 (56)
Median age at enrollment (range) - yr		58 (19–86)
Self-reported race/ethnicity – Number (%)
White		421 (59)
Black or African American		101 (14)
Hispanic or Latinx		69 (10)
Asian		60 (8)
American Indian, Alaska Native, or Pacific Islander		5 (1)
Unknown/not reported		53 (7)
Referring Clinics – Number (%)
General Cardiology Clinic		292 (41)
Advanced Heart Failure Clinic		244 (34)
Lipid Clinic		137 (19)
Electrophysiology Clinic		18 (3)
Cardiothoracic Surgery		18 (3)
Indicationa	Number of Patients (% of Cohort)	Number of Positives (% of Indication)
Cardiomyopathy	87 (12)	11 (13)
Aortopathy	33 (5)	4 (12)
Dyslipidemia	325 (46)	32 (10)
Arrhythmia	141 (20)	13 (9)
Family History of Cardiovascular Disease	11 (2)	1 (9)

^aPatients frequently had multiple indications for testing.

Fig. 1: HeartCare workflow and clinical report.

(a) Workflow from enrollment to return of results. Final report was added to local EMR for physician determination if any additional studies, medications, or referrals were required.

(b) Example clinical report with four sections corresponding to 1) rare Mendelian conditions 2) coronary artery disease polygenic risk score, 3) LPA gene risk variants 4) pharmacogenetic variants related to simvastatin and warfarin. Subsequent pages conveyed additional information specific to each section, including guideline-based recommendations.

HeartCare Web Portal

The BCM Human Genome Sequencing Center (HGSC) developed the HeartCare Web Portal to receive test orders and facilitate data management for this project. The HeartCare Web Portal provides a secure, HIPAA compliant environment that can receive sensitive patient/participant information at enrollment and deliver finished reports when required. The Portal, therefore, serves as both the entry point for receiving test orders from clinical providers and the mechanism for delivering test results back to the providers. The service generates a local identifier for each sample and uses that to track all genetic studies in a blinded fashion throughout the sequencing workflow. When reporting data back to participants, the HeartCare Portal allows the reuniting of personal data to the genetic data and transports that back to the referring physician and the patient’s medical record. The HeartCare Portal interfaced with the Epic Health Care management system and is adaptable to similar other systems.

Panel Design and Content

The HeartCare genetic screen tests 158 genes related to inherited Mendelian cardiovascular conditions associated with increased risk of aortic aneurysms, cardiomyopathies, arrhythmias, and dyslipidemia (Table S1). The panel includes cardiac-related ACMG SF^™ genes² and additional genes deemed actionable by the internal HeartCare group (“ACMG^™, ACMG SF^™, ACMG 59^™, ACMG 56^™, and related words and designs incorporating ACMG^™, are trademarks of the American College of Medical Genetics and Genomics and may not be used without permission.”). For the latter, a review of scientific literature, OMIM-morbid genes⁴, and other laboratory panels was done to make a comprehensive list of genes linked to inherited CV phenotypes. We also included 50 single nucleotide polymorphism (SNP) sites that had achieved genome-wide significance for association with coronary artery disease (CAD) as part of a polygenic risk score (PRS) as previously described by Khera et al (Table S2)⁵. Two LPA risk alleles associated with high Lp(a) levels and a corresponding increased risk of cardiovascular disease in Caucasian populations⁶ were also analyzed (Table S3). Lastly, specific Clinical Pharmacogenetics Implementation Consortium (CPIC)-curated gene variants in SLCO1B1 and CYP2C9/VKORC1 affecting the metabolism of simvastatin and warfarin, respectively, were also included (Table S4)^7,8.

Sample collection and DNA isolation

Blood samples were collected in PAXgene Blood DNA tube (PreAnalytiX, Becton-Dickinson) or Hemogard K2 EDTA tubes (Becton-Dickinson). DNA was isolated with PAXgene blood DNA kit (PreAnalytix, QIAGEN) or Gentra Puregene Blood kit (QIAGEN), respectively. The saliva samples were collected in Oragene Dx OGR-500 tubes (DNA Genotek) and isolated with PrepIT L2P (DNA Genotek). Extractions were done following manufacturer instruction in the HGSC CLIA certified laboratory. DNA was quantified using Quant-iT PicoGreen dsDNA Assay kit (ThermoFisher) and its quality estimated by electrophoresis.

Capture, Sequencing, and Primary Analysis

The assay was a NimbleGen 641 Kb oligonucleotide capture panel, including the 158 genes described above. An additional 5.8 Kb and 23.4 Kb of IDT lockdown probes were added to improve capture of low coverage regions and include pharmacogenetic and PRS variants, respectively. Each sample was tagged with a DNA barcode and capture pool contained a total of 35 samples, and 70 samples were run per lane. Average DNA sequence coverage across the captured region was ~400X with >99% of bases covered above 20X (Fig. S1,S2). Sequence alignment with BWA-MEM⁹ and variant identification utilized Atlas and the Mercury Pipeline¹⁰, followed by Sanger confirmation for reportable variants. Copy number variant (CNV) calls were made via Atlas-CNV¹¹ and confirmed with Multiplex Ligation-dependent Probe Amplification (MRC-Holland). Sequencing was performed using the Illumina HiSeq 2500 platform in the BCM HGSC CAP/CLIA-certified clinical laboratory (HGSC-CL).

Variant Interpretation and Clinical Reporting

A HeartCare Variant Interpretation Group was formed that included clinical geneticists and cardiologists with expertise in cardiovascular genetics. Variants from the 158 Mendelian genes were classified according to ACMG interpretation guidelines¹² with gene specific recommendations used if applicable. For variants of uncertain significance (VUS) potentially related to the patient’s phenotype, additional clinical and family information was requested from the ordering clinician to provide evidence to support reclassification. In the event that reclassification was not possible, referral to adult genetics for additional evaluation, including family segregation studies, was made when appropriate. Thus, the final clinical report included pathogenic and likely pathogenic variants in accordance with ACMG guidelines that recommend variants of uncertain significance not be used in clinical decision making¹². Such VUSs remained available in the portal for clinician review. As per ACMG recommendations, reanalysis of primary data is ongoing and may also lead to reclassification as more evidence becomes available¹³.

The PRS was calculated as previously described⁵, and a positive result was returned if a patient was found to be above the 95^th percentile of risk. Patients were informed of their high PRS test result by their physician, who subsequently counseled them on recommended lifestyle modifications AHA life’s simple seven¹⁴. A positive LPA result was reported if an LPA risk allele was detected. A disclaimer was present on the report describing the uncertain applicability of the PRS and LPA findings in individuals of non-European ancestries. For pharmacogenetic findings, star alleles were determined based on the variants detected and interpreted using CPIC guidelines^7,8. Specific recommendations related to that finding were also included based on accepted society guidelines.

Final reports (Fig. 1b) appeared in the HeartCare Portal, where the laboratory director approved them, followed by upload into the Epic as a Portable Document Format (PDF) document attached to the test order for physician review. Specific parts of the LPA results were also structured as HL7 V2 messages to Epic as a first step for clinical decision support of best practice advisories¹⁵. The HGSC-developed Neptune platform was used for managing the connection between the clinical laboratory and the EMR¹⁶.

Return of Results

Clinical reports were returned to the physician for review and determination if any changes to the care plan were indicated. Negative results and pharmacogenetic-only reports were sent directly to patients with a physician letter (mail or MyChart, Epic System Corporation). Positive results required an office visit or telephone follow up for full disclosure. Complimentary genetic counseling services were available through Consultagene™, and referral to adult genetics and other specialists for additional workup and family cascade testing was streamlined. Clinician surveys were conducted for positive cases to determine the usefulness of the result in managing a patient’s health and assess the overall benefit of the test.

Statistical Analysis

The 95% confidence intervals (CI) were calculated for the reported diagnostic using the Wald method. The Pearson’s chi-squared test was used to evaluate statistical differences in medical management between groups.

RESULTS

From December 2018 through July 2020, a total of 878 patients were approached and 713 initially enrolled in the study, resulting in a refusal rate of ~19% (95% CI, 16–22%). The most common reason for declining participation was impact on life insurance. Four individuals were withdrawn either by request or death during the study, leaving 709 who completed analysis with the HeartCare gene panel. Race/ethnicity of participating patients was consistent with overall demographics within the participating clinics (Table 1). While patients frequently had multiple indications, the most common reasons for testing were dyslipidemia at 46% (n=325; 95% CI, 42–50%), arrhythmia at 20% (n=141; 95% CI, 17–23%), and cardiomyopathy at 12% (n=87; 95% CI, 10–215%) (Table 1). Overall, a genetic finding with potential clinical implications was found in 32% of participants (n=225; 95% CI, 28–35%) when considering rare, monogenic conditions, elevated polygenic risk, and LPA risk alleles alone (Fig. 2a). Among Mendelian conditions, the yield was highest among patients with a cardiomyopathy indication, where 13% (n=11; 95% CI, 7–21%) harbored a likely pathogenic or pathogenic variant in a related gene (Table 1). This was followed by aortopathy, dyslipidemia, and arrhythmia indications, in which 12% (n=4; 95% CI, 4–28%), 10% (n=32; 95% CI, 7–14%), and 9% (n=13; 95% CI, 5–15%) of such patients carried a disease-associated variant, respectively (Table 1). Overall, some 9% (n=64; 95% CI, 7–11%) of patients were molecularly diagnosed with a cardiac-related Mendelian disorder (Table 2, Fig. 2b).

Fig. 2: HeartCare results and clinical impact.

(a) HeartCare positive finding breakdown by category among the 709 study participants. LP/P = Likely pathogenic or pathogenic variants in 158 genes associated with rare, Mendelian conditions; LPA = Lipoprotein(a) gene risk allele; PRS = Polygenic Risk Score; PGx = Pharmacogenetic allele.

(b) Category breakdown among actionable findings in the patients testing positive (n=64) for an actionable (pathogenic/likely pathogenic) finding in the 158 gene panel.

Table 2:

Pathogenic/Likely pathogenic variants detected from 158 genes associated with rare, Mendelian conditions. A total of 65 variants were detected in 64 individuals.

Gene (Transcript)	Disorder	DNA	Protein	Interpretation	Age, Sex, Phenotype	Management Change
ABCA1 (NM_005502.4)	High-density lipoprotein (HDL) Deficiency	c.5398A>C	p.Asn1800His	LP	58F, pulmonary hypertension, stroke	Referral to New Specialist
ABCA1 (NM_005502.4)	High-density lipoprotein (HDL) Deficiency	c.5398A>C	p.Asn1800His	LP	67F, dyslipidemia, family history of CAD	Medication change
APOB (NM_000384.2)	Familial Hypercholesterolemia (FH)	c.10580G>A	p.Arg3527Gln	P	54F, dyslipidemia, family history of dyslipidemia	Medication change
APOB (NM_000384.2)	Familial Hypercholesterolemia (FH)	c.10519C>T	p.Arg3507Trp	LP	56M, premature CAD s/p PCI	Medication change
APOB (NM_000384.2)	Familial Hypercholesterolemia (FH)	c.10580G>A	p.Arg3527Gln	P	55F, dyslipidemia	Follow-Up with Existing Specialist
APOB (NM_000384.2)	Familial Hypercholesterolemia (FH)	c.10580G>A	p.Arg3527Gln	P	75F, dyslipidemia	Follow-Up with Existing Specialist
BAG3 (NM_004281.3)	Dilated Cardiomyopathy (DCM)	exon 2–4 deletion		P	24F, non-ischemic postpartum cardiomyopathy	Referral to New Specialist
DSG2 (NM_001943.3)	Arrhythmogenic Right Ventricular Cardiomyopathy (ARVC)	c.2857delC	p.Leu953TrpfsTer11	LP	55F, cardiomyopathy, palpitations	Follow-Up with Existing Specialist
ELN (NM_001278939.1)	Supravalvular Aortic Stenosis	c.1150+1G>A		LP	50F, family history of cardiomyopathy	Cardiac test (echo)
FBN1 (NM_000138.4)	Marfan syndrome	c.4337A>T	p.Asp1446Val	LP	44F, ascending aortic dissection at age 28, family history of aortic aneurysms	Referral to New Specialist
FBN1 (NM_000138.4)	Marfan syndrome	c.3428G>A	p.Gly1143Asp	LP	52M, aortic root aneurysm	Referral to New Specialist
FBN1 (NM_000138.4)	Marfan syndrome	c.1462T>G	p.Cys488Gly	LP	40M, ascending aortic dissection at age 30	Referral to New Specialist
GLA (NM_000169.2)	Fabry disease	c.337_354del	p.Phe113_Arg118del	P	73F, dyslipidemia	Referral to New Specialist
KCNE1 (NM_000219.5)	Long QT syndrome	c.226G>A	p.Asp76Asn	LP	42F, history of cardiac arrest s/p ICD	Referral to New Specialist
KCNH2 (NM_000238.3)	Brugada syndrome	c.1888G>A	p.Val630Ile	LP	43M, Brugada syndrome	Referral to New Specialist
LDLR (NM_000527.4)	Familial Hypercholesterolemia (FH)	c.313+1G>A		P	54F, dyslipidemia, family history of premature CAD	Medication change
LDLR (NM_000527.4)	Familial Hypercholesterolemia (FH)	c.654_656del	p.Gly219del	P	63F, dyslipidemia, family history of premature CAD	Follow-Up with Existing Specialist
LDLR (NM_000527.4)	Familial Hypercholesterolemia (FH)	c.682G>T	p.Glu228Ter	P	21M, dyslipidemia, family history of premature CAD	Lab test
LDLR (NM_000527.4)	Familial Hypercholesterolemia (FH)	c.1327T>C	p.Trp443Arg	LP	26F, dyslipidemia, family history of dyslipidemia	Medication change
LDLR (NM_000527.4)	Familial Hypercholesterolemia (FH)	c.682G>T	p.Glu228Ter	P	46M, premature CAD s/p CABG	Follow-Up with Existing Specialist
LDLR (NM_000527.4)	Familial Hypercholesterolemia (FH)	c.227_233del	p.Gly76ValfsTer128	LP	28M, dyslipidemia	Medication change
LDLR (NM_000527.4)	Familial Hypercholesterolemia (FH)	c.761A>C	p.Gln254Pro	LP	53M, dyslipidemia	Follow-Up with Existing Specialist
LDLR (NM_000527.4)	Familial Hypercholesterolemia (FH)	c.1056_1060+3del		P	48M, dyslipidemia, premature s/p CABG	Medication change
LDLR (NM_000527.4)	Familial Hypercholesterolemia (FH)	c.1118_1121dupGTGG	p.Tyr375Trpfs*7	P	79F, dyslipidemia, CAD	Follow-Up with Existing Specialist
LDLR (NM_000527.4)	Familial Hypercholesterolemia (FH)	c.2096C>T	p.Pro699Leu	LP	72F, dyslipidemia	Medication change
LDLR (NM_000527.4)	Familial Hypercholesterolemia (FH)	c.632A>T	p.His211Leu	LP	33F, dyslipidemia	Referral to New Specialist
LDLR (NM_000527.4)	Familial Hypercholesterolemia (FH)	c.1599G>A	p.Trp533Ter	P	35M, dyslipidemia, premature CAD, family history of CAD	Follow-Up with Existing Specialist
LDLR (NM_000527.4)	Familial Hypercholesterolemia (FH)	c.2096C>T	p.Pro699Leu	LP	64M, CAD, ischemic cardiomyopathy	Medication change
LDLR (NM_000527.4)	Familial Hypercholesterolemia (FH)	c.551G>Aa	p.Cys184Tyr	P	57F, premature CAD s/p CABG	Medication change
LDLR (NM_000527.4)	Familial Hypercholesterolemia (FH)	c.1103G>Aa	p.Cys368Tyr	P	57F, premature CAD s/p CABG	Medication change
LDLR (NM_000527.4)	Familial Hypercholesterolemia (FH)	c.259T>G	p.Trp87Gly	P	45F, dyslipidemia	Follow-Up with Existing Specialist
LDLR (NM_000527.4)	Familial Hypercholesterolemia (FH)	c.622G>A	p.Glu208Lys	P	35M, dyslipidemia, family history of premature CAD	Medication change
LDLR (NM_000527.4)	Familial Hypercholesterolemia (FH)	c.1329G>C	p.Trp443Cys	P	76F, CAD s/p CABG	Medication change
LDLR (NM_000527.4)	Familial Hypercholesterolemia (FH)	c.862G>A	p.Glu288Lys	P	68F, dyslipidemia, CAD	Medication change
LDLR (NM_000527.4)	Familial Hypercholesterolemia (FH)	c.682G>T	p.Glu228Ter	P	75F, dyslipidemia, CAD s/p CABG	Medication change
LDLR (NM_000527.4)	Familial Hypercholesterolemia (FH)	c.1284delC	p.Asn428Lys	LP	55F, dyslipidemia, family history of premature CAD	Behavior change recommendation
LDLR (NM_000527.4)	Familial Hypercholesterolemia (FH)	c.482T>A	p.Ile161Asn	LP	71F, dyslipidemia, family history of premature CAD	Medication change
LMNA (NM_170707.3)	Dilated Cardiomyopathy (DCM)	c.725C>T	p.Ala242Val	LP	45M, non-ischemic cardiomyopathy	Follow-Up with Existing Specialist
LMNA (NM_170707.3)	Dilated Cardiomyopathy (DCM)	c.1129C>T	p.Arg377Cys	LP	66M, heart failure	Referral to New Specialist
LPL (NM_000237.3)	Familial Combined Hyperlipidemia	c.644G>A	p.Gly215Glu	P	48F, hypertriglyceridemia, pancreatitis	Follow-Up with Existing Specialist
LPL (NM_000237.3)	Familial Combined Hyperlipidemia	c.644G>A	p.Gly215Glu	P	56F, palpitations, hypertension	Lab test
LPL (NM_000237.3)	Familial Combined Hyperlipidemia	c.644G>A	p.Gly215Glu	P	56M, hypertriglyceridemia	Lab test, Behavior change recommendation
LPL (NM_000237.3)	Familial Combined Hyperlipidemia	c.249+1G>A		P	43M, hypertriglyceridemia	Medication change
LPL (NM_000237.3)	Familial Combined Hyperlipidemia	c.929G>A	p.Cys310Tyr	LP	36M, hypertriglyceridemia	Behavior change recommendation
MYBPC3 (NM_000256.3)	Hypertrophic Cardiomyopathy (HCM)	c.3330+2T>G		P	69M, heart failure, atrial fibrillation	Referral to New Specialist
MYBPC3 (NM_000256.3)	Hypertrophic Cardiomyopathy (HCM)	c.655G>C	p.Val219Leu	P	52M, atrial fibrillation	Referral to New Specialist, Cardiac test
MYH7 (NM_000257.3)	Hypertrophic Cardiomyopathy (HCM)	c.2710C>T	p.Arg904Cys	P	38F, non-ischemic cardiomyopathy	Follow-Up with Existing Specialist
MYH7 (NM_000257.3)	Hypertrophic Cardiomyopathy (HCM)	c.5655G>A	p.Ala1885=	LP	67M, CAD s/p PCI	Referral to New Specialist
MYH7 (NM_000257.3)	Hypertrophic Cardiomyopathy (HCM)	c.2207T>C	p.Ile736Thr	P	46M, HCM	Referral to New Specialist
MYH7 (NM_000257.3)	Hypertrophic Cardiomyopathy (HCM)	c.3133C>T	p.Arg1045Cys	LP	63M, dyslipidemia	Cardiac test (echo)
NEXN (NM_144573.3)	Dilated Cardiomyopathy (DCM)	c.1955A>G	p.Tyr652Cys	LP	72M, personal and family history of ascending aortic aneurysm and dissection	Referral to New Specialist, Lab test
NEXN (NM_144573.3)	Dilated Cardiomyopathy (DCM)	c.1955A>G	p.Tyr652Cys	LP	46M, dyslipidemia, family history of heart disease	Cardiac test (echo)
PKP2 (NM_004572.3)	Arrhythmogenic Right Ventricular Cardiomyopathy (ARVC)	c.623delC	p.Thr208LysfsTer55	P	55F, palpitations, family history of ARVC	Referral to New Specialist, Medication change
PRKAR1A (NM_002734.4)	Carney Complex	c.565_566insC	p.Glu189AlafsTer44	LP	33M, Carney complex	Referral to New Specialist
SDHA (NM_004168.3)	Mitochondrial Cardiomyopathy	c.688delG	p.Glu230SerfsTer10	LP	43M, premature CAD, multiple strokes, type 2 diabetes	Referral to New Specialist
SDHA (NM_004168.3)	Mitochondrial Cardiomyopathy	c.944dupG	p.Gly316ArgfsTer5	LP	69M, HCM, pancreatic cancer	Referral to New Specialist, Non-cardiac imaging test, Lab test
TGFB2 (NM_001135599.3)	Loeys-Dietz syndrome, type 4	c.979C>T	p.Arg327Trp	P	33F, ascending aortic aneurysm	Referral to New Specialist, Cardiac test (CTA), Medication change
TNNI3 (NM_000363.4)	Hypertrophic Cardiomyopathy (HCM)	c.586G>A	p.Asp196Asn	LP	45F, CP/SOB, family history of HCM	Cardiac test (echo), Referral to New Specialist
TTN (NM_001267550.1)	Dilated Cardiomyopathy (DCM)	c.93376_93377delAG	p.Arg31126GlyfsTer2	P	59M, non-ischemic cardiomyopathy	Medication change
TTN (NM_001267550.1)	Dilated Cardiomyopathy (DCM)	c.61876C>T	p.Arg20626Ter	LP	61M, non-ischemic cardiomyopathy	Referral to New Specialist, Medication change
TTN (NM_001267550.1)	Dilated Cardiomyopathy (DCM)	c.102352C>T	p.Arg34118Ter	LP	52M, non-ischemic cardiomyopathy	Referral to New Specialist, Medication change
TTN (NM_001267550.1)	Dilated Cardiomyopathy (DCM)	c.54000G>A	p.Trp16359Ter	P	61M, family history of cardiomyopathy	Cardiac test (echo)
TTN (NM_001267550.1)	Dilated Cardiomyopathy (DCM)	c.100172–1G>C		LP	55M, non-ischemic cardiomyopathy	Referral to New Specialist
TTR (NM_000371.4)	Hereditary transthyretin (ATTR) amyloidosis	c.424G>A	p.Val142Ile	P	77M, amyloidosis	Medication change
TTR (NM_000371.4)	Hereditary transthyretin (ATTR) amyloidosis	c.424G>A	p.Val142Ile	P	64M, HCM	Cardiac test (MRI), Lab test

^aOne individual was compound heterozygous for LDLR variants. LP, likely pathogenic; P, pathogenic; M, male; F, female; CAD, Coronary artery disease; PCI, Percutaneous coronary intervention; CABG, Coronary artery bypass grafting; ICD, implantable cardioverter-defibrillator; CTA, Computed tomography angiography; MRI, magnetic resonance imaging; CP, chest pain; SOB, shortness of breath

Twenty percent (n=144; 95% CI, 18–23%) carried an LPA risk allele, and 9.3% (n=66; 95% CI, 7–12%) belonged to the high-risk group according to the PRS (Fig. 2a). Nearly half of the cohort had a positive pharmacogenetic result, with 25% (n=182; 95% CI, 46–53%) for simvastatin, 15% (n=105) for warfarin, and 9% (n=62) for both medications (Fig. 2a, Fig. S3). Notably, 30% and 9% of individuals prescribed either warfarin or simvastatin had a genetic variant impacting initial warfarin dosing or risk of simvastatin-induced myopathy, respectively (Fig. S4). Only 20% (n=29) of patients had known Lp(a) levels prior to HeartCare (Fig. S5a). As of submission of this manuscript, Lp(a) levels had been measured in 59% (n=85) of patients with a newly found LPA risk allele, with 82% (n=70) found to have elevated levels and the corresponding risk for cardiovascular disease (Fig. S5b). Reflecting the limited applicability of these known LPA alleles to non-European ancestries, Hispanic individuals made up two-thirds of cases (n=10) with normal Lp(a) levels and a positive LPA risk allele despite representing only 10% of the entire cohort (Fig. S5c).

Clinician feedback was positive among those surveyed (n=13) regarding the usefulness of HeartCare for the care of a subset of patients who tested positive on the 158 gene panel, PRS, or LPA risk allele (n=122). Clinicians reported that the HeartCare results were beneficial to the patient overall (median 4 on a scale of 1=“not at all beneficial” to 6=“extremely beneficial”), with the 158 gene panel receiving the highest usefulness rating (median 5 on a scale of 1=“not at all useful” to 6=“extremely useful”) for managing a patient’s health (Table 3). In the majority (60.0%) of cases, the clinician agreed or strongly agreed that the patient’s HeartCare results improved overall clinical care beyond what clinical evaluation and non-genetic diagnostics alone could have achieved. Notably, clinicians reported recommending management changes for 84.4% (n=103/122) of patients based on their HeartCare result (Table 3). Such interventions included but were not limited to medication changes, cascade testing, specialist referrals, and additional cardiac tests. Management changes were made more often in patients with a new diagnosis established via HeartCare (93.3%) than in those whose existing clinical diagnosis was confirmed (72.6%; p= 0.003).

Table 3:

Clinician survey results for patients testing positive for the 158 gene panel, PRS, or LPA risk allele. A total of 13 clinicians completed surveys about 122 patients with positive results.

Usefulness of each HeartCare Study component for managing the patient’s health	Not at all useful=1, Extremely useful=6; Median (Q1, Q3)
Usefulness of 158 Cardiac Gene Panel, n=119	5 (3, 6)
Usefulness of LPA Risk Variant(s), n=113	4 (3, 5)
Usefulness of Polygenic Risk Score, n=112	4 (3, 5)
Usefulness of Pharmacogenomic Findings, n=111	3 (3, 5)
Overall benefit of HeartCare Study results for patient, n=119	Not at all beneficial=1, Extremely beneficial=6; Median (Q1, Q3) 4 (4, 5)
HeartCare Study results improved overall clinical care of the patient beyond usual care, n=120	n (%)
Strongly Disagree or Disagree	4 (3.3%)
Neither Agree nor Disagree	44 (36.7%)
Strongly Agree or Agree	72 (60.0%)
Clinician-reported medical management changes based on HeartCare results
	Patient’s existing diagnosis confirmed by HeartCare results, n=51	Patient did not have existing diagnosis matching HeartCare results, n=60	Total, n=122a
Any Management Change for a Given Patient, n (%)	37 (72.5%)	56 (93.3%)	103 (84.4%)
Type of Management Change(s)b, n
Lab test (genetic test, lipid levels, and LPA level)	17	37	62
Cardiac test	10	13	24
Non-cardiac imaging test	1	2	3
Follow-Up with Existing Specialist	9	11	22
Referral to New Specialist	6	12	18
Medication change (start, stop, or change dose)	8	11	21
Behavior change recommendation	11	29	45
Family member testing recommended or performed	25	13	41

PRS, polygenic risk score; Q1, first quartile; Q3, third quartile.

^aThe total column includes reported management changes for 10 patients with missing responses to the question about whether a diagnosis was confirmed.

^bMore than one management change could be reported per patient.

DISCUSSION

In this study, we deployed a comprehensive molecular assay combining both rare, high-impact genetic variation and common susceptibility markers in the form of a PRS, LPA polymorphisms, and pharmacogenetic alleles. More than 30% of our cohort patients were found to have a positive finding with potential clinical impact, even after excluding pharmacogenetic findings present in nearly half of all patients. Physicians also found these results to benefit patient health and improve clinical care with 84% reporting medical management changes in those testing positive.

The study’s high overall yield is unheralded but reflects the manifest applicability of genetic testing to cardiovascular disease in adults. In fact, cardiology has led the way in establishing the role of molecular testing in diagnosis and management, with numerous societies having published guidelines recommending genetic testing for patients with dyslipidemias, arrhythmias, cardiomyopathies, and thoracic aortic disease^17–20. Nevertheless, uptake has been slow, due in part to lack of comprehensive testing coupled with challenges integrating it into the clinical workflow, both of which we addressed with HeartCare. Thus, our overall yield of 32% is an accurate reflection of heritable cardiovascular disease among adults and represents an opportunity to individualize patient care.

Another major barrier to the adoption of genetic testing into clinical care is payer reimbursement policies that frequently consider such testing investigational and not accepted practice²¹. Though we did not bill payers in this study, third party insurance reimbursement will be important for wider deployment of the HeartCare screen. To establish validity and utility of a diagnostic test, payers require evidence of an effect on medical outcomes and an equivalent benefit as established alternatives²². By tracking physician response in this study, we showed that incorporating both rare and common genetic variation affecting cardiovascular health does improve overall clinical management of the patient beyond standard of care. Additional economic analysis will be important in further directing reimbursement policy of payers by demonstrating cost-effectiveness of such testing.

The overall yield of 9% for identification of Mendelian disorders when testing cardiovascular disease related genes is comparable to results from studies performing comprehensive genetic testing for specific heritable cancers and chronic kidney disease (CKD). After heart disease, cancer is the second leading cause of death in the United States¹. Recent studies involving individuals with breast and colorectal cancer have shown that 9.4% and 15.5% of cases, respectively, had a pathogenic or likely pathogenic variant in a high-penetrance cancer susceptibility gene^23,24. Similarly, CKD is also a major cause of morbidity and mortality and affects more than 10% of the population²⁵, with one study reporting a diagnostic yield of 9.3% in CKD patients undergoing whole exome sequencing (WES). In contrast, some studies have shown a yield as high as 30–40% for patients meeting strict criteria for diagnoses like hypertrophic cardiomyopathy²⁶. Our study, on the other hand, was open to all cardiology clinic patients presenting with a variety of conditions and did not require any particular indications for inclusion. Thus, it offers a ‘real world’ view of the yield expected in a typical cardiology care population.

Half of the molecularly diagnosed Mendelian conditions in this study were lipid-related, the majority being familial hypercholesterolemia (FH). FH is highly treatable, with high-dose statin therapy and PCSK9 inhibitors leading to significant reductions in cardiac-related mortality^27,28, and yet 9 in 10 cases go undiagnosed²⁹. Many insurance payers require genetic testing results to cover PKSK9 inhibitors for FH patients but will not pay for the genetic tests³⁰. We identified 25 patients in our cohort with molecularly confirmed FH, of which only a third were already on a PCSK9 inhibitor, indicating a significant potential opportunity for further LDL-C reducing therapy. We also identified several individuals with pathogenic variants in LPL, a gene associated with severe hypertriglyceridemia and pancreatitis. In addition to lifestyle and diet changes, these individuals are candidates for aggressive triglyceride-reducing therapy with a combination of fibrates, niacin, and/or fish oil with newer therapies under development using anti-sense and siRNA approaches which are effective in individuals with loss of function variants in LPL³¹.

In 2018 the Heart Failure Society of America (HFSA) and ACMG published guidelines recommending genetic testing for patients with cardiomyopathies to assist in patient management¹⁷. More than 30% of our positive cases were cardiomyopathy-related and the diagnostic yield was highest for this indication (Table 1). One informative case included a 24-year-old female with a history non-ischemic postpartum cardiomyopathy. As part of her cardiology workup, the HeartCare panel was sent and identified a pathogenic multi-exon deletion in BAG3, a gene associated with autosomal dominant dilated cardiomyopathy (DCM)³². Other loss-of-function variants in BAG3 have been reported in multi-generational families with DCM with some individuals requiring cardiac transplantation^32,33. This patient was subsequently referred to adult genetics for additional genetic counseling and discussion of cascade testing for her siblings and children. With this molecular diagnosis and worsening cardiac function requiring a left ventricular assist device, she is currently undergoing evaluation for a heart transplant. Two African American males in the cohort with non-ischemic cardiomyopathy were found to carry the TTR c.424G>A (p.Val142Ile) variant associated with hereditary transthyretin (hATTR) amyloidosis. This variant, present in up to 3.5% of the African American population, is an under-recognized cause of heart failure in this population, particularly those over age 60³⁴. This genetic finding is important as it may help predict the overall disease course, and it allows for the use of two genetic therapies, inotersen and patisiran, to ameliorate amyloid-associated polyneuropathy³⁵.

Genetic testing is also important in risk stratification and management of individuals with aortic aneurysm disease²⁰. One such case involved a 33-year-old female originally thought to have hypermobile Ehlers Danlos syndrome due to long standing joint hypermobility and dislocations. Baseline echocardiography revealed mild dilation of the aortic root at the sinus of Valsalva (4.0 x 3.6 cm), and on physical exam she had a pectus carinatum defect. Family history was significant for scoliosis in her mother and a maternal grandfather with a descending aortic aneurysm at age 65 years. The HeartCare panel was sent and identified a pathogenic variant in the TGFB2 gene, c.979C>T (p.Arg327Trp). Defects in TGFB2 are the cause of Loeys-Dietz syndrome 4 (LDS4), an autosomal dominant connective tissue disorder characterized by early-onset aneurysms and dissections of the aorta³⁶. Though LDS has phenotypic overlap with Marfan syndrome, the vascular disease tends to be more aggressive in LDS compared to Marfan with aortic dissection occurring at smaller diameters and anywhere along the aorta. Consequently, surgical guidelines recommend prophylactic root replacement at smaller diameters compared to Marfan³⁶. This patient is being followed closely by cardiothoracic surgery and was referred to adult genetics for evaluation of other systemic features of LDS and additional discussion, including family cascade testing and the increased risk of dissection with pregnancy.

The LPA gene that encodes the apolipoprotein(a) component of the Lp(a) lipoprotein particle has received significant medical interest recently due to its association with atherosclerotic cardiovascular disease. Increased levels of circulating Lp(a) lipoprotein are associated with an elevated risk of coronary disease and stroke^6,37. In particular, two common LPA polymorphisms (rs3798220 and rs10455872) are strongly associated with an increased level of Lp(a) lipoprotein and increase the risk of CAD by 42%−57% in individuals of European ancestry³⁷. The variable Kringle-IV (KIV-2) repeats within the LPA gene are also thought to contribute to circulating Lp(a) levels³⁸. Current management for elevated Lp(a) includes lifestyle modifications as well as statin therapy. Clinical trials of an apo(a) antisense oligonucleotide (ASO) and small interfering RNAs that reduce Lp(a) up to 80% offer hope for future reduction of cardiovascular events in those at risk³⁹. Despite the well-established association with CVD and potentials for therapy, Lp(a) levels are rarely checked alongside standard lipid profiles and testing for the known risk alleles is not routinely done. Unsurprisingly, of the 144 patients with a LPA risk allele, only 20% had Lp(a) levels determined via a biochemical test before HeartCare. With this result, however, their Lp(a) levels are being followed and managed appropriately. Our study also demonstrates the need for additional studies of LPA genetics in diverse ethnicities since Lp(a) levels did not correlate well with the two alleles tested in individuals of non-European ancestry. Research into elucidating the number and role of LPA Kringle repeats in atherosclerotic disease in this cohort is ongoing.

Recent recommendations focus on the utility of genetic testing for high-penetrance, monogenic conditions after careful phenotyping and collection of pedigree information⁴⁰. However, within the precision medicine paradigm, pharmacogenetic results and PRS may offer an additional means to improve individual-level risk assessment and to better predict treatment response in both primary and secondary prevention for a broader population. We elected to test for pharmacogenetic variants associated with simvastatin and warfarin given the common use of these classes of medications in adults. Nearly 30% of all adults and 50% of those with atherosclerotic CVD in the US were prescribed a statin medication as of 2013⁴¹. And although newer anticoagulants that do not require monitoring or dose adjustment are gaining popularity, in 2017 more than 15 million prescriptions for warfarin were written in the US according to the Medical Expenditure Panel Survey (MEPS), due in part to its effectiveness, low cost, and ability to quickly reverse. In our cohort, nearly half of patients had a positive pharmacogenetic finding related to these two medications. Among those currently prescribed either warfarin or simvastatin, 30% and 9% carried a genetic variant associated with altered metabolism, respectively. These individuals are being monitored for related complications (e.g., bleeding, myopathy) and family members have been encouraged to consider testing themselves. Additional pharmacogenetic markers affecting other cardiac-related medications (e.g., clopidogrel, beta blockers) may be another area of opportunity for this cohort in future studies.

Polygenic risk scores (PRS) for CAD like the one used here have been proposed as a method to improve cardiovascular risk prediction by aggregating weighted associations between genetic variants and disease outcomes derived from genome-wide association studies (GWAS). Using a 50 SNP PRS that was state of the art at study onset, more than 9% of our cohort was found to be in the highest CVD risk group (top 5%). The clinical utility of CAD PRS has been investigated in a number of important areas, with a primary focus on applications to primary prevention in individuals without pre-existing CAD⁴². For example, studies have demonstrated the ability of a CAD for PRS to effectively stratify individuals according to their risk for CAD and to identify individuals with high risk for CAD who did not have high cholesterol levels or other traditional risk factors⁴³. Additional work has provided evidence that genetic risk for CAD can be mitigated by statin therapy and a healthy lifestyle^43,44. PRS construction has now expanded to scores comprised of millions of SNPs with improved risk classification and the ability to identify individuals with risk for CAD equivalent to a monogenic trait⁴³. Ultimately, further research is still needed to understand the proper role of PRS within a clinical context. Furthermore, research should be directed at the development of PRS that perform well in multiethnic populations and on defining target populations for genetic risk stratification.

In summary, we present here the compelling results of our HeartCare study, a comprehensive test targeting genes that influence risk for cardiovascular disease. Impressively, the overall positive rate was 32%, reflecting the enriched patient population seen in ambulatory cardiology clinics where patients commonly present with arrhythmias, heart failure, dyslipidemias, and aortic aneurysms. Critical management changes resulted from these findings, from new pharmacologic therapies to surgical considerations. Our 9% rate for Mendelian cardiovascular conditions was consistent with other studies evaluating genetic causes of common diseases such as cancer and kidney disease. The primary findings have implications beyond the tested patients as close family members have up to 50% chance of sharing the same genetic risk, providing practitioners the opportunity to implement cascade testing of at-risk family members and offer preventive interventions. Clinicians were supportive of this test overall, finding it benefitted patient health and improved clinical care. Data will continue to be re-analyzed as new interpretations become available, and we will leverage the research consent to perform whole-genome sequencing to assess the added clinical utility of a broader assay. There is also opportunity to utilize phenotype data in the EMR to identify additional patients who may benefit from genetic testing, thus speeding up diagnosis and improving care even further⁴⁵. In conclusion, our results demonstrate that comprehensive molecular testing can be routinely used in the ambulatory setting and that cardiovascular disease is a prime model to demonstrate precision medicine’s utility.

Data Availability

All reported variants have been submitted to ClinVar (https://www.ncbi.nlm.nih.gov/clinvar/) under Human Genome Sequencing Center Clinical Lab, Baylor College of Medicine. Please contact the corresponding author for original data access requests.