Rare Coding Variants Associated With Electrocardiographic Intervals Identify Monogenic Arrhythmia Susceptibility Genes: A Multi-Ancestry Analysis.

BACKGROUND: Alterations in electrocardiographic (ECG) intervals are well-known markers for arrhythmia and sudden cardiac death (SCD) risk. While the genetics of arrhythmia syndromes have been studied, relations between electrocardiographic intervals and rare genetic variation at a population level are poorly understood.
METHODS: Using a discovery sample of 29 000 individuals with whole-genome sequencing from Trans-Omics in Precision Medicine and replication in nearly 100 000 with whole-exome sequencing from the UK Biobank and MyCode, we examined associations between low-frequency and rare coding variants with 5 routinely measured electrocardiographic traits (RR, P-wave, PR, and QRS intervals and corrected QT interval).
RESULTS: We found that rare variants associated with population-based electrocardiographic intervals identify established monogenic SCD genes (KCNQ1, KCNH2, and SCN5A), a controversial monogenic SCD gene (KCNE1), and novel genes (PAM and MFGE8) involved in cardiac conduction. Loss-of-function and pathogenic SCN5A variants, carried by 0.1% of individuals, were associated with a nearly 6-fold increased odds of the first-degree atrioventricular block (P=8.4×10-5). Similar variants in KCNQ1 and KCNH2 (0.2% of individuals) were associated with a 23-fold increased odds of marked corrected QT interval prolongation (P=4×10-25), a marker of SCD risk. Incomplete penetrance of such deleterious variation was common as over 70% of carriers had normal electrocardiographic intervals.
CONCLUSIONS: Our findings indicate that large-scale high-depth sequence data and electrocardiographic analysis identifies monogenic arrhythmia susceptibility genes and rare variants with large effects. Known pathogenic variation in conventional arrhythmia and SCD genes exhibited incomplete penetrance and accounted for only a small fraction of marked electrocardiographic interval prolongation.

Nearly 1 in 3 adults will experience an arrhythmia and up to 10% will die of sudden cardiac death (SCD) during their lifetime.[1,2] The ECG is an inexpensive, noninvasive, and widely used screening test for abnormalities in cardiac conduction. Previous work has demonstrated that electrocardiographic intervals are quantitative markers for arrhythmias and SCD[3-8] and have a considerable heritable basis.[9]

Monogenic mutations underlying many conduction disorders and arrhythmia syndromes, such as the long QT syndrome (LQTS),[10,11] have been the focus of extensive study over the past 2 decades. Typically, these studies have focused on sequencing a modest number of affected patients or families to identify the causative genes. In contrast, large-scale genome-wide association studies have identified a multitude of loci associated with electrocardiographic traits by studying common variants in large study samples comprising thousands of individuals.[12-15] However, the effect of discovered genome-wide association studies variants is inherently small and direct implication of any particular gene tagged by an identified common variant is difficult.

To date, a missing gap in our understanding of common electrocardiographic traits and related diseases has been whether rare coding variants have a substantial contribution to variation at the population level. Rare variants may have substantial effect sizes, confer pathogenicity, and have a measurable impact on disease risk. Yet, identifying such variants has been technically challenging since this analysis requires both the sequencing of large study samples and the availability of electrocardiographic data.

To address this challenge, we used a unique resource of over 130 000 individuals with whole-exome or -genome sequencing, and a rigorous 3-stage design, to examine the associations between rare coding genetic variation and several routinely collected electrocardiographic traits, including the heart rate, maximum P-wave duration, PR interval, QRS duration, and corrected QT interval (QTc). We then specifically assessed the frequency, magnitude of association, and penetrance of clinically pathogenic variation in select SCD genes that are associated with electrocardiographic variation.

Methods

Detailed methods are provided in the Data Supplement. The data that support the findings of this study are available from the corresponding author upon reasonable request. There are restrictions to the availability of raw Trans-Omics in Precision Medicine (TOPMed) and MyCode phenotypic and genotypic data due to the identifiable nature of this data. UK Biobank raw data are available to researchers via application through the UK Biobank website. All TOPMed participants provided written informed consent, and participating studies obtained ethical approval from their local institutional review boards. The UK Biobank resource was approved by the UK Biobank Research Ethics Committee, and all participants provided written informed consent to participate. Use of UK Biobank data was performed under application number 17488 and was approved by the local Massachusetts General Brigham Institutional Review Board.

Results

An overview of the study design and sample selection flow is presented in Figure 1 and Figures I through III in the Data Supplement. Characteristics of the included studies are displayed in the Table and Tables I through III and Figure IV in the Data Supplement.

Figure 1.

Flow chart of study and analyses.Top of the figure illustrates different traits detected by the ECG. Genetic association studies were performed for 5 electrocardiographic traits in 9 studies from Trans-Omics in Precision Medicine (TOPMed) as a discovery cohort and findings from discovery analyses were replicated using UK Biobank and MyCode studies (blue). We analyzed genetic variations using both single variant association and gene-based association approaches (orange). Moreover, we calculated frequency of loss-of-function, pathogenic and likely pathogenic variants in long QT syndrome genes and performed association tests between such variants and QT interval. EUR indicates European ancestry; ME, multi-ethnic; PWD, P-wave duration; QRS, QRS duration; QTc, corrected QT interval; WES, whole-exome sequencing; and WGS, whole-genome sequencing.

Once we had aggregated electrocardiographic and genetic data for participants in TOPMed, we began by performing genome-wide association studies for common variants related to 5 electrocardiographic traits (Figure 2A and Table IV in the Data Supplement). As expected, we observed associations at 44 previously reported loci (Results in the Data Supplement). Cross-trait pleiotropy was particularly notable at the SCN5A-10A locus and the CAV1 locus, which demonstrated robust associations across multiple electrocardiographic traits.

Figure 2.

Manhattan plots for 5 electrocardiographic traits.A illustrates circular Manhattan plot illustrating genome-wide association testing results between 5 electrocardiographic traits and common variants with minor allele frequency (MAF) >1%. Loci that reached a conventional genome-wide significant threshold (P=5×10−8, red dotted lines) are annotated with the nearest genes. B shows associations between low-frequency (0.1% ≤ MAF <1%) variants and PR interval. The gray dotted line is the significant threshold (0.05/83 994 variants =6.0×10−7).

We then performed an analysis of the protein-coding regions of the genome and identified a low-frequency coding variant in PAM that was associated with PR interval duration (p.Ser539Trp, rs78408340_G, MAF=0.5%, β=8 ms, P=1.9×10−7, Figure 2B and Table V in the Data Supplement). This finding was replicated in the UK Biobank (β=2 ms, P=0.01) and in MyCode (β=3 ms, P=1.8×10−4). Variants in PAM were also associated with PR interval duration in gene-based testing in TOPMed (P=4.5×10−7, Figure 3, Table VI in the Data Supplement), an association driven by p.Ser539Trp (Figure V in the Data Supplement). We did not observe consistent associations between rare high-confidence loss-of-function (LOF) variants in PAM and PR interval duration across data sets (Results, Tables VII and VIII, and Figure VIA in the Data Supplement). Low-frequency coding variant testing also identified a synonymous variant in MFGE8 (p.Ser52=rs141997845_T, MAF=0.1%, β=19 ms, P=4.9×10−8), which was associated with marked PR interval prolongation in TOPMed (Figure 2B, Results and Table V in the Data Supplement). Only one carrier was identified in the UK Biobank, who had a first-degree atrioventricular block (PR=200 ms). In MyCode, the association was replicated (β=15 ms, P=1.6×10−3).

Figure 3.

Association results between electrocardiographic traits and predicted-deleterious variants in genes from candidate loci. Figure 3 illustrates associations between electrocardiographic traits (RR interval, P-wave duration [PWD], PR interval, QRS duration, and corrected QT interval [QTc]) and genes in candidate loci in Trans-Omics in Precision Medicine (TOPMed) using SMMAT. Genes with P>0.05 for all traits were removed from this figure. As shown in the key legend, the gradient of blue colors represents the strength of associations in this heatmap. Genes with a star (*) were significantly associated with an electrocardiographic trait (P<1.2×10−4); tests with P>0.05 have been made white. The maximum PWD was significantly associated with HAND1 (P=2.4×10−5). PR interval was significantly associated with SCN5A (P=7.6×10−7) and PAM (P=4.5×10−7). QRS duration was significantly associated with CR1L (P=1.2×10−4). QTc was significantly associated with KCNQ1 (P=2.3×10−12) and KCNH2 (P=3.2×10−8). PR indicates PR interval; QRS, QRS duration; and RR, RR interval.

Gene-based testing in TOPMed also identified 3 genes established as important for cardiac conduction, arrhythmias, and SCD (SCN5A [PR interval, P=7.6×10−7], KCNQ1 [QTc, P=2.3×10−12], KCNH2 [QTc, P=3.2×10−8], Figure 3, Table VI in the Data Supplement). Rare LOF variants in SCN5A conferred a 38 ms (P=4.3×10−32, N carriers=70) increase in PR interval duration across all data sets (Results, Figure VIA, and Tables VII and VIII in the Data Supplement). Furthermore, gene-based testing highlighted pleiotropy of SCN5A, as rare variants also associated with prolonged P-wave duration and QRS duration (Results and Figure VII in the Data Supplement). Similarly, rare LOF variants in KCNQ1 were associated with marked prolongation of the QTc (β=42 ms, P=3.9×10−44, N carriers=78), as were rare LOF variants in KCNH2 (β=38 ms, P=8.5×10−12, N carriers=21, Results, Figure VIB, and Tables VII and VIII in the Data Supplement). In exploratory exome-wide analyses, a number of genes reached significance (Results and Table IX in the Data Supplement). Of these genes, none reached P<0.05 in replication among UK Biobank participants, except for SCN5A (PR interval), PAM (PR interval), KCNQ1 (QTc), and KCNH2 (QTc).

In an analysis of 17 genes included on a typical clinical sequencing panel for long QT syndrome,[16] in addition to KCNQ1 and KCNH2, we also identified an association between QTc and predicted-deleterious variants in KCNE1 in TOPMed (P=1.2×10−4), which was replicated in the UK Biobank (P=9.0×10−5; Results and Table X in the Data Supplement). No KCNE1 LOF variants were identified in TOPMed or the UK Biobank, although rare predicted-deleterious missense variants in these data sets were associated with a 16 ms (P=3.3×10−7) prolongation of the QTc (Results and Tables XI and XII in the Data Supplement). In contrast to the large effect sizes observed for deleterious variation in the aforementioned genes, the top variants in common variant analyses for PR interval and QTc conferred effect sizes of 4 and 3 ms, respectively (Table IV in the Data Supplement).

We observed similar findings for pathogenic or likely pathogenic variants adjudicated by clinical testing laboratories and submitted to ClinVar. Among 54 355 TOPMed and UK Biobank participants, 239 (0.44%) carried such a variant in a LQTS gene from the panel cited above. Half were located in one of the 3 most validated susceptibility genes, KCNQ1, KCNH2, and SCN5A (Table XIII in the Data Supplement).[10] Pathogenic or likely pathogenic variants in KCNQ1 and KCNH2 were associated with substantially prolonged QTc values across all studies (Figure VIII and Table VII in the Data Supplement). Pathogenic and likely pathogenic variants in SCN5A were not associated with QTc prolongation, likely owing to heterogeneity of allele effects, although were associated with substantially prolonged PR intervals (Results and Figure IXA in the Data Supplement). When aggregated together, LOF, pathogenic, or likely pathogenic variants in KCNQ1 and KCNH2 were associated with a 30 ms (P=1.1×10−67) and 27 ms (P=1.0×10−16) prolongation of the QTc, respectively (Figure 4).

Figure 4.

Forest plots for loss-of-function (LOF), pathogenic or likely pathogenic variants in Across Trans-Omics in Precision Medicine (TOPMed), UK Biobank and MyCode datasets, KCNQ1 and KCNH2 LOF and pathogenic or likely pathogenic variants significantly and markedly prolonged the QTc, with inverse-variance weighted fixed-effects meta-analyzed effect estimates of 30 ms (P=1.1×10−67) and 27 ms (P=1.0×10−16) prolongation, respectively.

Despite the marked effect of deleterious variation on electrocardiographic intervals, incomplete penetrance was common (Table XIV in the Data Supplement). LOF, pathogenic, or likely pathogenic variants in SCN5A, which were carried by 0.1% of individuals, and were associated with an ≈6-fold increased odds of first-degree atrioventricular block in TOPMed and the UK Biobank (P=8.4×10−5), and 12-fold increased odds in MyCode (P=2.7×10−12; Tables XV in the Data Supplement). Nevertheless, about 70% of carriers had a PR interval of <200 ms, indicating absence of first-degree atrioventricular block (Results and Figure IX in the Data Supplement).

Similarly, deleterious KCNQ1 and KCNH2 variants were carried by 0.2% of individuals and were associated with an almost 23-fold increased odds of a QTc duration of at least 480 ms in TOPMed and UK Biobank (P=4.5×10−25) and 9-fold increased odds in MyCode, an interval length that is suggestive of LQTS (Figure 5 and Table XV and Figure X in the Data Supplement).[17] Yet over 75% of individuals carrying an LOF, pathogenic, or likely pathogenic variant in KCNQ1 or KCNH2 had a QTc below 480 ms (95/110 in TOPMed and UK Biobank, 137/181 in MyCode).

Figure 5.

Effect of loss-of-function (LOF), pathogenic or likely pathogenic variants in A illustrates distributions for carriers (red, N=110) of a LOF or pathogenic or likely pathogenic variant in KCNQ1, KCNH2, and noncarriers (gray, N=54 245) in Trans-Omics in Precision Medicine (TOPMed) and UK Biobank. The dotted lines represent QTc cutoffs of 460, 480, and 500 ms. Of the carriers, 15 (13.6%) individuals had QTc interval ≥480 ms while 662 (1.2%) of noncarriers revealed QT prolongation. B illustrates the odds ratio for QTc prolongation at different cutoffs (460, 480, and 500 ms) conferred by LOF, pathogenic or likely pathogenic variants in KCNQ1 and KCNH2 in TOPMed and UK Biobank.

Few individuals with prolonged intervals carried known deleterious variation. For example, among individuals with first-degree atrioventricular block, only 0.3% carried a LOF, pathogenic, or likely pathogenic variant in SCN5A in TOPMed and the UK Biobank, and 0.5% in MyCode. Similarly, among individuals with marked QTc prolongation (eg, ≥480ms), only 2.4% carried a LOF or known deleterious variant in KCNQ1 or KCNH2 in TOPMed and UK Biobank, and 1.2% in MyCode (Figure 5). Extended analyses summarizing the frequency of additional rare protein-coding variation are displayed in Table XVII in the Data Supplement. Notably, fewer than 11% of individuals with QTc≥480ms or QTc≥500 ms carried any rare KCNQ1 or KCNH2 protein-altering variant across TOPMed and UK Biobank.

Discussion

Using a unique resource of high-depth genomic sequence data from over 130 000 individuals, we identified low-frequency and rare genetic variants underlying variability in 5 routinely collected electrocardiographic traits. Our findings indicate that pathogenic variation in arrhythmia and SCD genes are associated with marked PR (SCN5A) and QTc (KCNQ1, KCNH2, and KCNE1) prolongation in the general population. Nevertheless, over 70% of individuals with deleterious variation had normal electrocardiographic intervals, indicating that routinely measured electrocardiographic intervals may be insensitive for the detection of such carriers. Moreover, <3% of individuals with marked PR interval or QTc variation carried a known deleterious variant, and <10% with marked QTc prolongation carried any rare protein-changing variant in a predominant LQTS gene. The causes of such electrocardiographic interval prolongation remain unclear and warrant further examination. Last, our findings highlight the value of large-scale sequencing efforts to identify novel genes, such as PAM and MFGE8, which we implicated in atrioventricular conduction.

Our study complements and extends prior literature. To date, most genome-wide association studies of electrocardiographic traits have largely been focused on common genetic variants, have studied these traits individually,[12,13,15,18-20] or have relied on sequencing in smaller samples with imputation of low-frequency variants.[14] In contrast, the recent and rapid innovation in sequencing technology has enabled the analysis of very rare and potentially deleterious coding variation in relation to cardiac traits. Such rare variants are likely to directly implicate genes in cardiac physiology and may confer large effect sizes which could be of clinical relevance.

Our findings have 3 important implications. First, rare variants in arrhythmia and SCD susceptibility genes are associated with large effects on the ECG in the population, yet routinely measured electrocardiographic intervals may not be a reliable method for identifying most carriers. Incomplete electrocardiographic penetrance was common among individuals carrying deleterious variation. Indeed, it is likely that pathogenic variation in arrhythmia and SCD genes exhibits lower penetrance than was reported previously in family-based analyses,[21-23] consistent with reports for arrhythmogenic cardiomyopathy genes.[24] Considering the frequency of second hits that can predispose to QTc prolongation and lethal ventricular arrhythmias, the consequences of subclinical genetic predisposition to arrhythmias requires prospective evaluation, particularly given the adoption of genome-first approaches for which return of incidental findings in several arrhythmia and SCD genes is encouraged.[25] Since electrocardiographic intervals may vary over time, future analyses that leverage repeated electrocardiographic measures may provide more accurate estimates of rare variant penetrance. Moreover, analysis of additional electrocardiographic features beyond standard intervals is warranted. Importantly, whether a genome-first approach to identifying pathogenic variant carriers will have a material impact on SCD risk requires prospective evaluation.

Second, our observations suggest that quantitative traits measured in population-based studies are endophenotypes for pathogenic variation. Analyzing low-frequency and rare genetic variation in relation to commonly ascertained electrocardiographic traits is an efficient approach for identifying important genes, both established and novel, that are related to cardiac conduction and arrhythmia risk. For example, variation in SCN5A was associated with PR duration. SCN5A encodes the α-subunit of the cardiac sodium channel and comprises the major inward sodium current responsible for cardiomyocyte depolarization during phase 0. Mutations in SCN5A are responsible for several conditions, including atrial fibrillation, bradyarrhythmias, cardiomyopathy, Brugada syndrome, LQTS, and SCD.[11] Additionally, low-frequency and rare coding variants in KCNQ1, KCNH2, and KCNE1 were associated with QTc. KCNQ1 and KCNH2 encode voltage-gated potassium channel subunits responsible for the outward rectifier currents I and I, respectively, which govern cardiomyocyte repolarization during phases 2 and 3 of the action potential. Mutations in KCNQ1 and KCNH2 represent the most common forms of LQTS.[10] KCNE1 encodes a β-subunit that interacts with KCNQ1 to form the I current. Notably, KCNE1 has recently been considered a controversial susceptibility gene for typical monogenic LQTS.[10,26]

Examination of sequence data in relation to ECGs can also identify novel pathways involved in cardiac physiology. PAM encodes peptidylglycine-alpha-amidating-monooxygenase, an enzyme expressed in atrial cardiomyocytes, where it colocalizes with atrial natriuretic peptide.[27-29] MFGE8 encodes the milk fat globule-epidermal growth factor 8 that is involved in phagocytic signaling and has been implicated in neovascularization,[30] cardiac hypertrophy,[31] and atrial fibrosis.[32] These novel conduction genes have not been examined extensively in relation to cardiovascular disease. Future work is necessary to characterize the relations between genetic variation in these genes and disease, as well as their mechanistic roles in cardiovascular biology. Larger discovery samples anticipated in the near future are likely to identify additional arrhythmia and SCD susceptibility genes, emphasizing the importance of high-throughput functional characterization of new genes.

Third, few individuals with markedly abnormal electrocardiographic intervals had known deleterious variation in classic arrhythmia and SCD genes, indicating that the causes of such electrocardiographic variability remain unclear. We suspect that multiple factors may account for prolonged electrocardiographic intervals, including rare variants in genes not traditionally implicated in arrhythmias and SCD, polygenic susceptibility to electrocardiographic interval prolongation, and other factors, such as electrolyte abnormalities or medication exposures, that were not accounted for in our analysis. We excluded individuals on antiarrhythmic medications, individuals with paced rhythms, and pathological QRS prolongation which may confound some intervals (eg, QTc). The yield of contemporary panel-based genetic evaluations in individuals with isolated electrocardiographic interval prolongation is, therefore, likely low. The cause, prognosis, and optimal management of these genotype-negative individuals in the community warrants evaluation considering the adverse prognosis traditionally associated with both prolonged PR interval and QTc.[3,8,33-36]

Our results should be evaluated in the context of the study design. The sample consisted mainly of middle-aged individuals of European ancestry, limiting generalizability of results beyond the ancestral groups and age strata represented. We used single time point electrocardiographic analyses and intervals may vary over time, although for the studied traits 10 second ECGs have been determined to be reliable.[37,38] We did not account for all medications that may affect cardiac conduction, which requires future investigation. Finally, we cannot exclude a survival bias since included participants were adults at the time of enrollment.

In conclusion, we demonstrate the value of large-scale high-depth sequence analysis for interrogating the genetic basis of the ECG. As biorepositories grow in the near future, similar approaches will undoubtedly uncover additional rare variants with other high-impact on cardiovascular diseases.

Acknowledgments

We gratefully acknowledge the studies and participants who provided biological samples and data for TOPMed and the DiscovEHR.

Sources of Funding

The whole-genome sequencing for the Trans-Omics in Precision Medicine (TOPMed) program was supported by the National Heart, Lung, and Blood Institute (NHLBI). Whole-genome sequencing for NHLBI TOPMed: the Atherosclerosis Risk in Communities (phs001211.v1.p1) was performed at the Broad Institute of MIT and Harvard (3R01HL092577-06S1) and Baylor Human Genome Sequencing Center (3U54HG003273-12S2 and HHSN268201500015C). Whole-genome sequencing for NHLBI TOPMed: Genetics of Cardiometabolic Health in the Amish (phs000956.v1.p1) was performed at the Broad Institute of MIT and Harvard (3R01HL121007-01S1). Whole-genome sequencing for NHLBI TOPMed: Mount Sinai BioMe Biobank (phs001644.v1.p1) was performed at the Baylor Human Genome Sequencing Center (HHSN268201600033I) and McDonnell genome institute (HHSN268201600037I). Whole-genome sequencing for NHLBI TOPMed: Cleveland Family Study—Whole-genome sequencing collaboration (phs000954.v1.p1) was performed at the University of Washington northwest genomics center (3R01HL098433-05S1 and HHSN268201600032I). Whole-genome sequencing for NHLBI TOPMed: Cardiovascular Health Study (phs001368.v1.p1) was performed at the Baylor Human Genome Sequencing Center (3U54HG003273-12S2, HHSN268201500015C, and HHSN268201600033I). Whole-genome sequencing for NHLBI TOPMed: Framingham Heart Study (phs000974.v1.p1) was performed at the Broad Institute of MIT and Harvard (3R01HL092577-06S1 and 3U54HG003067-12S2). Whole-genome sequencing for NHLBI TOPMed: The Jackson Heart Study (phs000964.v1.p1) was performed at the University of Washington Northwest Genomics Center (HHSN268201100037C). Whole-genome sequencing for NHLBI TOPMed: Multi-Ethnic Study of Atherosclerosis (phs001416.v1.p1) was performed at the Broad Institute of MIT and Harvard (3U54HG003067-13S1). Whole-genome sequencing for NHLBI TOPMed: Womens Health Initiative (phs001237.v1.p1) was performed at the Broad Institute of MIT and Harvard (HHSN268201500014C). Centralized read mapping and genotype calling, along with variant quality metrics and filtering were provided by the TOPMed Informatics Research Center (3R01HL-117626-02S1; contract HHSN268201800002I). Phenotype harmonization, data management, sample-identity QC, and general study coordination were provided by the TOPMed Data Coordinating Center (3R01HL-120393-02S1; contract HHSN268201800001I). The TOPMed component of the Amish Research Program was supported by the National Institutes of Health (NIH) grants R01 HL121007, U01 HL072515, and R01AG18728. The Mount Sinai BioMe Biobank has been supported by The Andrea and Charles Bronfman Philanthropies and, in part, by Federal funds from the NHLBI and NHGRI (U01HG00638001; U01HG007417; and X01HL134588). The ARIC study (Atherosclerosis Risk in Communities) has been funded in whole or in part with Federal funds from the NHLBI, NIH, Department of Health and Human Services (contract numbers HHSN268201700001I, HHSN268201700002I, HHSN268201700003I, HHSN268201700004I, and HHSN268201700005I). The CHS (Cardiovascular Health Study) was supported by contracts HHSN268201200036C, HHSN268200800007C, HHSN268201800001C, N01HC55222, N01HC85079, N01HC85080, N01HC85081, N01HC85082, N01HC85083, N01HC85086, 75N92021D00006, and grants U01HL080295 and U01HL130114 from the NHLBI, with additional contribution from the National Institute of Neurological Disorders and Stroke (NINDS). Additional support was provided by R01AG023629 from the National Institute on Aging (NIA). A full list of principal CHS investigators and institutions can be found at CHS-NHLBI.org. The content is solely the responsibility of the authors and does not necessarily represent the official views of the NIH. The CFS (Cleveland Family Study) has been supported, in part, by NIH grants (R01-HL046380, KL2-RR024990, R35-HL135818, and R01-HL113338). The FHS (Framingham Heart Study) acknowledges the support of contracts NO1-HC-25195 and HHSN268201500001 from the NHLBI and grant supplement R01 HL092577-06S1, and 18SFRN34110082 from American Heart Association for this research. We also acknowledge the dedication of the FHS study participants without whom this research would not be possible. Dr Vasan is supported, in part, by the Evans Medical Foundation and the Jay and Louis Coffman Endowment from the Department of Medicine, Boston University School of Medicine. Dr Kornej has received funding from the Marie Sklodowska-Curie Actions under the European Union’s Horizon 2020 research and innovation programme (grant agreement no. 838259). Dr Lunetta acknowledges support under contract R01 HL092577-10. Dr Benjamin is supported under contracts R01HL128914, 2R01 HL092577 from the NHLBI and 18SFRN34110082 from the American Heart Association. The JHS (Jackson Heart Study) is supported and conducted in collaboration with Jackson State University (HHSN268201800013I), Tougaloo College (HHSN268201800014I), the Mississippi State Department of Health (HHSN268201800015I) and the University of Mississippi Medical Center (HHSN268201800010I, HHSN268201800011I, and HHSN268201800012I) contracts from the NHLBI and the National Institute on Minority Health and Health Disparities (NIMHD). MESA (Multi-Ethnic Study of Atherosclerosis) and the MESA SHARe project are conducted and supported by the NHLBI in collaboration with MESA investigators. Support for MESA is provided by contracts HHSN268201500003I, N01-HC-95159, N01-HC-95160, N01-HC-95161, N01-HC-95162, N01-HC-95163, N01-HC-95164, N01-HC-95165, N01-HC-95166, N01-HC-95167, N01-HC-95168, N01-HC-95169, UL1-TR-000040, UL1-TR-001079, and UL1-TR-001420. MESA family is conducted and supported by the NHLBI in collaboration with MESA investigators. Support is provided by grants and contracts R01HL071051, R01HL071205, R01HL071250, R01HL071251, R01HL071258, and R01HL071259 and by the National Center for Research Resources, Grant UL1RR033176. The provision of genotyping data was supported, in part, by the National Center for Advancing Translational Sciences, CTSI grant UL1TR001881, and the National Institute of Diabetes and Digestive and Kidney Disease Diabetes Research Center (DRC) grant DK063491 to the Southern California Diabetes Endocrinology Research Center. The Women’s Health Initiative program is funded by the NHLBI, NIH, US Department of Health and Human Services through contracts HHSN268201600018C, HHSN268201600001C, HHSN268201600002C, HHSN268201600003C, and HHSN268201600004C. We would further like to thank Jacques Rossouw, Shari Ludlam, Joan McGowan, Leslie Ford, and Nancy Geller from the NHLBI program office, Garnet Anderson, Ross Prentice, and Andrea LaCroix from the Clinical Coordinating Center (Fred Hutchinson Cancer Research Center, Seattle, WA), Investigators JoAnn E. Manson (Brigham and Women’s Hospital, Harvard Medical School, Boston, MA), Barbara V. Howard (MedStar Health Research Institute/Howard University, Washington, DC), Marcia L. Stefanick (Stanford Prevention Research Center, Stanford, CA), Rebecca Jackson (The Ohio State University, Columbus, OH), Cynthia A. Thomson (University of Arizona, Tucson/Phoenix, AZ), Jean Wactawski-Wende (University at Buffalo, Buffalo, NY), Marian Limacher (University of Florida, Gainesville/Jacksonville, FL), Jennifer Robinson (University of Iowa, Iowa City/Davenport, IA), Lewis Kuller (University of Pittsburgh, Pittsburgh, PA), Sally Shumaker (Wake Forest University School of Medicine, Winston-Salem, NC), Robert Brunner (University of Nevada, Reno, NV), and Mark Espeland from the Women’s Health Initiative Memory Study (Wake Forest University School of Medicine, Winston-Salem, NC). The DiscovEHR (MyCode) study was supported by the NHLBI through grant number R01HL141901. Dr Lubitz is supported by NIH grant 1R01HL139731 and American Heart Association 18SFRN34250007. Dr Sotoodehnia is supported by NIH grant R01HL141989, by AHA grant 19SFRN34830063, and by the Laughlin Family.

Disclosures

Dr Lubitz receives sponsored research support from Bristol Myers Squibb/Pfizer, Bayer AG, Boehringer Ingelheim, and Fitbit and has consulted for Bristol Myers Squibb/Pfizer and Bayer AG and participates in a research collaboration with IBM. Dr Ellinor is the principal investigator on a grant at the Broad Institute from Bayer HealthCare.

Supplemental Materials

Supplemental Methods

Extended Supplemental Methods and Results

Description of TOPMed participating studies

Banner Authors

Supplemental Acknowledgments

Supplemental Figures I–XIII

Supplemental Tables I–XVII

References [39-64]

Table.

Baseline Characteristics of TOPMed Participants