Long QT Syndrome–Associated Mutations in Intrauterine Fetal Death

Importance Intrauterine fetal death or stillbirth occurs in approximately 1 out of every 160 pregnancies and accounts for 50% of all perinatal deaths. Postmortem evaluation fails to elucidate an underlying cause in many cases. Long QT syndrome (LQTS) may contribute to this problem.

Objective To determine the spectrum and prevalence of mutations in the 3 most common LQTS susceptible genes (KCNQ1, KCNH2, and SCN5A) for a cohort of unexplained cases.

Design, Setting, and Patients In this case series, retrospective postmortem genetic testing was conducted on a convenience sample of 91 unexplained intrauterine fetal deaths (mean [SD] estimated gestational age at fetal death, 26.3 [8.7] weeks) that were collected from 2006-2012 by the Mayo Clinic, Rochester, Minnesota, or the Fondazione IRCCS Policlinico San Matteo, Pavia, Italy. More than 1300 ostensibly healthy individuals served as controls. In addition, publicly available exome databases were assessed for the general population frequency of identified genetic variants.

Main Outcomes and Measures Comprehensive mutational analyses of KCNQ1 (K_V7.1, LQTS type 1), KCNH2 (HERG/K_V11.1, LQTS type 2), and SCN5A (Na_V1.5, LQTS type 3) were performed using denaturing high-performance liquid chromatography and direct DNA sequencing on genomic DNA extracted from decedent tissue. Functional analyses of novel mutations were performed using heterologous expression and patch-clamp recording.

Results The 3 putative LQTS susceptibility missense mutations (KCNQ1, p.A283T; KCNQ1, p.R397W; and KCNH2 [1b], p.R25W), with a heterozygous frequency of less than 0.05% in more than 10 000 publicly available exomes and absent in more than 1000 ethnically similar control patients, were discovered in 3 intrauterine fetal deaths (3.3% [95% CI, 0.68%-9.3%]). Both K_V7.1-A283T (16-week male) and K_V7.1-R397W (16-week female) mutations were associated with marked K_V7.1 loss-of-function consistent with in utero LQTS type 1, whereas the HERG1b-R25W mutation (33.2-week male) exhibited a loss of function consistent with in utero LQTS type 2. In addition, 5 intrauterine fetal deaths hosted SCN5A rare nonsynonymous genetic variants (p.T220I, p.R1193Q, involving 2 cases, and p.P2006A, involving 2 cases) that conferred in vitro electrophysiological characteristics consistent with potentially proarrhythmic phenotypes.

Conclusions and Relevance In this molecular genetic evaluation of 91 cases of intrauterine fetal death, missense mutations associated with LQTS susceptibility were discovered in 3 cases (3.3%) and overall, genetic variants leading to dysfunctional LQTS-associated ion channels in vitro were discovered in 8 cases (8.8%). These preliminary findings may provide insights into mechanisms of some cases of stillbirth.

Intrauterine fetal death is a major public health problem. About 1 million fetal deaths occur in the United States annually, with the vast majority occurring prior to 20 weeks' estimated gestational age.¹ Fetal death occurring after 20 weeks, defined as stillbirth, has an incidence of 6.05 per 1000 live births.¹ In fact, there were 25 972 reported fetal deaths at 20 weeks of gestation or more in the United States in 2006, a number rivaling that of all infant deaths (28 509) during this period.¹ In 2009, an estimated 2.64 million stillbirths occurred worldwide.²

Although about 50% of fetal deaths can be explained by chromosomal abnormalities, congenital anomalies, maternal or fetal infection, hemorrhage, placental or cord abnormalities, and maternal diseases, postmortem evaluations fail to identify a cause in approximately 25% to 40%, thus prompting a diagnosis of unexplained fetal death.³ However, long QT syndrome (LQTS) has been shown to be a major determinant in young sudden death individuals for which an autopsy was performed but had remained inconclusive⁴ and a determinant for as much as 10% of sudden infant death syndrome (SIDS).^5-7 Long QT syndrome may also contribute to sudden unexpected fetal mortality.⁸

Long QT syndrome is characterized by delayed myocardial repolarization that may manifest as a prolonged QT interval on a resting 12-lead surface electrocardiogram.⁹ With a prevalence of approximately 1:2000 (0.05%) in the general population,¹⁰ individuals with LQTS are at an increased risk of syncope, seizures, and sudden cardiac death, despite a structurally normal heart. Life-threatening cardiac arrhythmias can occur unexpectedly, mainly during childhood or adolescence. There have been anecdotal reports demonstrating fetal presentation of LQTS^11,12 and associating it with fetal death.^13-15

Herein, we report the spectrum and prevalence of functionally significant LQTS genetic variants in unexplained fetal death.

This case series study was of a convenience sample of unexplained intrauterine fetal deaths ascertained from 2006 to 2012 by 2 independent centers. In the United States, the fetuses were analyzed by the Mayo Clinic Windland Smith Rice Sudden Death Genomics Laboratory; in Italy, by 2 departments of Obstetrics and Gynecology (Milano and Modena) and referred to the Molecular Cardiology Laboratory of the Fondazione IRCCS Policlinico San Matteo of Pavia. All deaths remained “unexplained” following postmortem investigation including external evaluation of the fetus, placenta, and umbilical cord; circumstance of death review to exclude preterm premature rupture of membranes, preterm labor, abruption, peripartum infection, and karyotype analysis; and in some cases extensive toxicology, histologic, microbiologic, and biochemical examinations.¹⁶ Estimated gestational age at death and sex were known in all cases. Due to the anonymous nature of the US component of the study, race/ethnicity was not available for most cases. Based on the estimated gestational age, each unexplained death was classified as either a late abortion or miscarriage, occurring from the 14th to the 19th gestational week, or stillbirth, occurring the 20th week of gestation or after.

Race/ethnicity was self-reported by the participants and assessed due to potential ethnic differences in observed genotype frequencies. The Mayo Clinic cases analyzed were anonymous samples derived from fetuses referred to the Mayo Clinic for chromosomal analysis as part of a workup for unexplained fetal death and were noted to be nonanomalous and found to be karyotypically normal. The Mayo Foundation Institutional Review Board approved this anonymous autopsy study and provided waiver of consent. Only limited medical information such as sex and age at fetal death was available. For the Italian cases, the data reported are part of a larger study on intrauterine fetal death in Italy that was approved by the Italian Ministry of University and Research. Patient characteristics and autopsy data were obtained as part of routine clinical protocols, and the decedent's parent(s) signed informed consent, approved by the ethic committee of the University of Modena to perform genetic analysis. A definition of terms used in this article are listed in the Box.

• Denaturing high-performance liquid chromatography: A high-performance liquid chromatography instrument that uses temperature-dependent separation of DNA containing mismatched base pairs from polymerase chain reaction–amplified DNA fragments for chromatographic mutation analysis.

• Heterologous expression: A research technique that causes a protein to be produced in a cell that does not normally make (ie, express) that protein.

• Heterotetrameric, homotetrameric, and heteromultimeric ion channels: Ion channels made up of different combinations of protein subunits; 4 different subunits (heterotetrameric), 4 of the same subunits (homotetrameric), and 2 or more different subunits (heteromultimeric).

• Patch-clamp technique: A laboratory technique in electrophysiology that allows the study of single or multiple ion channels in cells.

For a complete list of genomic terms, see the ²⁹⁵⁰ in this issue.

Genomic DNA was extracted using standard techniques from placenta, umbilical cord, or fetal tissue. All cases underwent a comprehensive open reading frame and splice-site mutational analysis of KCNQ1 (HGNC 6294), KCNH2 (HGNC 6251), and SCN5A (HGNC 10593), the 3 major genes associated with LQTS, by means of polymerase chain reaction (PCR), denaturing high-performance liquid chromatography (Wave TM model 3500HT; Transgenomic), and DNA sequencing (3500 Dx Genetic Analyzer; Applied Biosystems). The PCR primer sequences are available upon request.

All identified genetic variants were denoted using accepted nomenclature¹⁷ and were categorized as putative pathogenic mutations, rare genetic variants of uncertain clinical significance, or common genetic variants. To be considered a putative pathogenic (ie, disease-causing) mutation, the genetic variant had to be absent in a large panel of ethnically similar controls derived from a previously investigated panel, which included data from 595 white, 319 black, 134 Asian, and 118 Hispanic individuals¹⁸; in the Helmholtz Zentrum exome database, which included data from 1414 white individuals; and in 3 publicly available databases: the 1000 Genome Project¹⁹ (http://www.1000genomes.org/ensembl -browser), which included data from 1094 individuals, 381 of whom were white, 246 were black, 286 were Asian, and 181 were Hispanic individuals; the NHLBI GO Exome Sequencing Project²⁰ (http://evs.gs.washington.edu/EVS), which included data from 5379 individuals, 3510 of whom were white and 1869 were black individuals); and the Exome Chip Design²¹ (http://genome.sph.umich.edu/wiki/Exome_Chip_Design), which included data from 12 000 individuals).

Given the previously documented prevalence (1:2000; 0.05%) of LQTS in the general population,¹⁰ variants with a heterozygous frequency less than 0.05% among the databases listed above were also considered as putative pathogenic mutations. In addition, to be considered disease-causing, the genetic variant had to exhibit an abnormal electrophysiological phenotype determined by in vitro functional analysis using patch-clamp recording.

Rare genetic variants of uncertain clinical significance were those present in ethnically similar control populations with a heterozygote frequency between 0.05% and 0.3%. Genetic variants with a frequency less than 0.05% but with a normal (wild-type) electrophysiological phenotype were also considered rare genetic variants of uncertain clinical significance.

Common genetic variants were those variants identified in ethnically similar controls with a heterozygous frequency greater than 0.3%. Synonymous and intronic variants not presumed to affect splicing were excluded from our analysis.

Binomial exact 95% confidence intervals were computed using the statistical program R,²² version 2.15.2, to assess the reliability of the estimated proportion of intrauterine fetal death victims with mutations and rare genetic variants. The data on estimated gestational age are presented as mean (SD) and compared by t test for independent samples using GraphPad Prism software, version 3. A 2-sided P value <.05 was considered statistically significant.

Genes can be spliced alternatively and produce different mRNA transcripts that contain unique amino acid coding sequences. The KCNH2 transcript is present in 2 forms, each with a unique exon 1 (Figure 1). These alternative transcripts not only can have different expression profiles among different tissue types but their level of expression can vary with developmental age (fetal tissue vs adult tissue). Because we analyzed an alternative KCNH2 exon 1 (belonging to the KCNH2 [1b] transcript) for mutations in our intrauterine fetal death cohort, we determined the level of expression of this alternative transcript in fetal heart tissue compared with adult heart tissue.

Expression of alternatively spliced KCNH2 mRNA transcripts (KCNH2 [1a] and KCNH2 [1b]) was examined in human heart tissue by real-time quantitative RT-PCR using gene-specific primers and fluorescent Taqman probes (sequences provided in eTable 1) using previously described methods and tissues.¹² Relative expression levels of the 2 transcripts were calculated using ratios of cycle threshold values. A standard curve generated from assaying cDNA standards mixed at known log₂ ratios and fitted by linear regression was used to interpolate the ratio of isoform expression. All tissues were assayed 6 times for each transcript.

Five genetic variants (KCNQ1, p.A283T; KCNQ1, p.R397W; KCNH2 [1b], p.R25W; SCN5A, p.D772N; and SCN5A, p.R1116Q), which were absent in ethnically similar controls and in which functional studies have never been reported in the literature, were characterized functionally by patch-clamp electrophysiological recording to assess their pathogenic role. Detailed methods are presented in the eAppendix. Data are presented as means (95% CIs). Comparisons were made using 1-way analysis of variance or t test where appropriate and P values <.05 were considered significant.

The study population consisted of 91 cases (51 females, 40 males) of unexplained intrauterine fetal death. The average estimated gestational age at the time of fetal demise was 26.3 (8.7) weeks (range, 14-41 weeks). Seventy-four percent of cases with known ethnicity were white. Females had a lower estimated gestational age (24.3 [8.1] weeks) than males (28.9 [8.7] weeks; P = .01). Sixty-one cases (30 females, 31 males; mean estimated gestational age, 31.1 [6.4] weeks; range, 20-41 weeks; 72% white) were classified as stillbirth (≥20 weeks' gestation) while 30 cases (21 females, 9 males; mean estimated gestational age, 16.6 [1.5] weeks; range, 14-19 weeks; 83% white) were classified as late abortion or miscarriage (<20 weeks). For the cohort demographics, see Table 1 and for individual case characteristics, eTable 2.

Excluding 2 very common genetic variants (KCNH2, p.K897T, and SCN5A, p.H558R), we identified 14 genetic variants in 18 intrauterine fetal deaths (19.8%) of 91 (95% CI, 12.2%-29.4%), 3 late abortion or miscarriages (10%) of 30 (95% CI, 2.1%-26.5%), and 15 stillbirth (24.6%) of 61 (95% CI, 14.5%-37.3%; Table 2). Three variants, (KCNQ1, p.A283T, KCNQ1, p.R397W, and KCNH2 [1b], p.R25W; Figure 1 and Figure 2) found in 3 intrauterine fetal death cases (3.3%) of 91 (95% CI, 0.68%-9.3%) were considered putative pathogenic mutations based on their absence in more than 1000 ethnically similar controls, a heterozygote frequency below the prevalence of LQTS in the general population (0.05%) as determined by analysis of more than 10 000 publicly available exomes, and an abnormal functional electrophysiological profile (see below). Due to the anonymity of the cases carrying the KCNQ1 mutations, we were unable to assess whether these mutations arose de novo or were transmitted from a parent. The p.R25W mutation in KCNH2 (1b) was inherited from the mother who exhibited borderline QTc prolongation.

Nine additional intrauterine fetal deaths (9.9% [95% CI, 4.6%-17.9%]; 5 females and 4 males) had rare nonsynonymous genetic variants (Table 2). In total, 7 rare nonsynonymous variants were identified including one KCNQ1 variant (p.K393N) and 6 SCN5A variants (p.T220I, p.S524Y, p.D772N, p.R1116Q, p.R1193Q involving 2 cases, and p.P2006A involving 2 cases). Three of these variants identified in 5 cases (SCN5A, p.T220I, p.R1193Q, and p.P2006A) have been shown previously to be functionally disruptive.^23,24 Two of these variants (KCNQ1, p.K393N and SCN5A, p.S524Y) have been shown to be functionally normal.^25,26 We determined that the functional properties of SCN5A variants p.D772N and p.R1116Q did not differ from the wild-type allele (eFigures 1-4). Furthermore, the SCN5A variant p.R1116Q (identified in a case of African ancestry) has a minor allele frequency of 0.07 in African Americans according to the National Heart, Lung, and Blood Institute Exome Sequencing Project.²¹ Together, this group of rare alleles were categorized as variants of uncertain clinical significance.

We also observed 6 common nonsynonymous variants (KCNH2: p.K897T and p.R1047L; SCN5A: p.R34C, p.H558R, p.V1951L, and p.F2004L) with an ethnically similar control population heterozygote frequency greater than 0.3% (Table 2). The 2 most common variants were p.K897T in KCNH2 (minor allele frequency, 0.23) and p.H558R in SCN5A (minor allele frequency, 0.24) detected in 37.4% (27 of 91 KT, 7 of 91 TT) and 42.9% (35 of 91 HR, 4 of 91 RR) of intrauterine fetal deaths, respectively. Three cases carrying SCN5A rare variants (p.P2006A, involving 2 cases; p.T220I, involving 1) were also heterozygous for SCN5A, p.H558R; however, the phase of these alleles could not be determined. None of the cases carrying a rare KCNH2 variant also carried the common KCNH2 variant (p. K897T). Overall, the carrier frequency observed for common variants among intrauterine fetal death cases was similar to the control populations.

Thus, we identified 3 putative pathogenic mutations (3.3% [95% CI, 0.68%-9.3%]) and 3 rare variants with a functional effect in 5 cases. In total, there were 8 cases (8.8% [95% CI, 3.9%-16.6%]) associated with dysfunctional LQTS-associated ion channels; 2 cases were classified as late abortions or miscarriages (<20 weeks' estimated gestational age) and 6 were classified as stillbirths (≥20 weeks' estimated gestational age).

The KCNQ1 gene encodes the pore forming α-subunit of the K_V7.1 voltage-gated K⁺ channel (otherwise known as K_V7.1) and is important for the repolarization of the human cardiac action potential. Mutation K_V7.1-A283T represents a novel missense mutation affecting a highly conserved alanine within the extracellular loop between transmembrane segment 5 and the K_V7.1 channel pore, whereas K_V7.1-R397W affects a conserved residue in the K_V7.1 C-terminus (Figure 1). Although KCNQ1, p.R397W was observed in 3 of 12 000 (0.025%) individuals included in the Exome Chip Design project, we still considered this variant as a possible pathogenic mutation because its allele frequency is lower than the population prevalence of LQTS (0.05%).

To determine whether these 2 KCNQ1 mutations confer functional defects with plausible contributions to intrauterine fetal death, we performed in vitro electrophysiological studies using the whole cell patch-clamp technique. The KCNQ1 mutations (p.A283T, p.R397W) were engineered in a recombinant K_V7.1 potassium channel plasmid vector and heterologously coexpressed with KCNE1 cDNA (HGNC:6240) to assess their functional consequences. Macroscopic currents (I_Ks) were recorded by applying steplike test pulses from −80 mV to 70 mV in 10-mV increments for 5 seconds, followed by a “tail” pulse for 5 seconds to −50 mV (Figure 3A). The peak step current (I_Ks) recorded at the end of the test pulse or at the initiation of the tail pulse was plotted as a function of the test-pulse potential (Figure 3, A and B). The peak tail current-voltage (I-V) relations were described with a Boltzmann equation (Figure 3C). Cells expressing p.A283T or p.R397W mutant K_V7.1 channels generated significantly smaller current densities (maximal I_Ks; I_MAX) than cells expressing wild-type K_V7.1 (Figure 3D). Additionally, cells expressing p.A283T exhibited a more positive potential of half-maximal voltage (V_½) of activation without a change in slope factor (k). Thus, K_V7.1-A283T and K_V7.1-R397W cause a loss of function and marked attenuation of I_Ks consistent with an in utero diagnosis of the type 1 LQTS (LQT1).

The KCNH2 [1b], p.R25W mutation alters a residue exclusive to an alternatively spliced transcript encoding HERG isoform 1b (HERG1b, Figure 1).^27,28 Biochemical and functional studies have demonstrated that HERG-mediated I_Kr current in native cardiac myocytes is generated by heterotetrameric potassium channels comprised of both HERG1a (encoded by the canonical KCNH2 splice isoform) and HERG1b subunits.²⁷ To enable functional studies of HERG1b-R25W, we implemented heterologous expression of homotetrameric HERG1a or HERG1b, or heteromultimeric HERG1a plus HERG1b channels in Chinese hamster ovary cells. Current density observed in cells expressing heteromultimeric wild-type HERG1a plus HERG1b was significantly larger, exhibited faster activation and deactivation kinetics, and inactivated at less negative potentials than cells expressing homotetrameric wild-type HERG1a (eTable 3). Cells expressing heterotetrameric wild-type HERG1a plus HERG1b-R25W mutant channels exhibited significantly smaller current densities, slower activation kinetics, significantly depolarized voltage dependence of activation and a delayed time course of recovery from inactivation (Figure 4; eTable 4) compared with wild-type HERG1a plus HERG1b channels. These findings indicate that heterotetrameric HERG channels containing HERG1b-R25W impair the generation of I_Kr.

Furthermore, we determined that expression of mRNA encoding HERG1b was 2-fold more abundant than HERG1a-encoding transcripts in fetal human hearts compared with adult hearts for which HERG1a transcript expression is 1.3-fold greater than the alternatively spliced isoform. These additional mRNA expression data further support the plausible contribution of HERG1b-R25W to attenuate I_Kr in fetal human hearts and confer susceptibility for in utero type 2 LQTS (LQT2).

In this study, pathogenic mutations or rare variants exhibiting abnormal functional effects involving the 3 most common LQTS-susceptibility genes were present in 8 of the 91 cases of unexplained intrauterine fetal death occurring after the 14th week of gestation. To our knowledge, this represents the first demonstration of such findings. This preliminary evidence suggests LQTS is one plausible cause of intrauterine fetal death; supports the previously proposed mechanistic link between some cases of intrauterine fetal death, SIDS, and LQTS⁷; and provides precedence for further large-scale investigations into the extent and role of cardiac channelopathies in stillbirth.

Although previous studies have provided some anecdotal evidence implicating LQTS as a cause, there has been no systematic investigation regarding this possible association.^13-15,29 Herein, we provide an estimate of the prevalence and spectrum of functionally significant LQTS gene variants in a large case series. Our results indicate that 8 of the 91 unexplained intrauterine fetal deaths occurring after the 14th week of gestation hosted either a pathogenic mutation consistent with in utero LQT1 or LQT2 (3.3%) or rare, functionally abnormal variants (5.5%) that could confer risk of life-threatening arrhythmia to the fetus. The yield of LQTS-associated genetic variants observed in intrauterine fetal death was remarkably similar to what we have reported previously about SIDS.^5-7

Approximately 75% of patients with a clinically certain diagnosis of LQTS host mutations in 1 of 3 major genes: KCNQ1 (LQT1, ≈ 35%), KCNH2 (LQT2, ≈ 30%) and SCN5A (LQT3, ≈ 10%).³⁰ In SIDS, more than half of the previously identified cardiac LQTS gene mutations involve the cardiac sodium channel macromolecular complex,^5-7 which underlies type 3 LQTS (LQT3) in which affected individuals are predisposed to cardiac events during periods of rest or sleep.³¹ By contrast, 2 of the 3 pathogenic mutations observed in the intrauterine fetal deaths were identified in KCNQ1 (LQT1); individuals with this genetic subtype often have cardiac events following activation of the sympathetic nervous system (ie, exercise or extreme emotion).³¹ The 2 KCNQ1 mutations (p.A283T and p.R397W) that we identified in late-abortion or miscarriage cases both exhibited significant (>70%) reductions in I_Ks density, consistent with the in vitro electrophysiological phenotype of an LQT1-causative mutation. The third putative disease-causing mutation was identified in the HERG 1b-encoding alternative splice isoform of KCNH2. The HERG1b-R25W mutation caused a 35% to 50% reduction in current density when expressed as heterotetramers with the canonical splice variant encoding HERG1a, also consistent with the in vitro electrophysiological phenotype of LQT2. Some of the SCN5A variants (p.T220I, p.R1193Q, and p.2006A) identified in this study have been associated previously with arrhythmia predisposition, unexplained sudden death, and an abnormal electrophysiological in vitro phenotype consistent with a cardiac channelopathy.^{6,23,24,32-35}

A recent study demonstrated that progesterone decreased I_Kr through impaired channel trafficking to the plasma membrane by disruption of intracellular cholesterol homeostasis³⁶ possibly reducing repolarization reserve in the normal fetal ventricle. Because during late pregnancy, electrocardiographic intervals including the QTc become longer^37,38 and fetal progesterone levels rise significantly,³⁹ one might anticipate that an intrauterine fetal death-related LQTS channelopathy would be more represented in advanced pregnancy. In fact, about two-thirds of the cases with a pathogenic or rare functional genetic variant had intrauterine fetal death occur during the third trimester. The presence of pathogenic, LQTS-causative channel mutations or rare variants having abnormal electrophysiological phenotypes may confer additional risk of lethal ventricular arrhythmias during this vulnerable period of intrauterine development.

Long QT syndrome has been associated with mutations in 15 genes accounting for approximately 75% to 80% of cases. We performed our molecular interrogation on the 3 most common LQTS-susceptibility genes that account for greater than 90% of all genotype-positive LQTS cases. The 12 minor LQTS-susceptibility genes were not assessed. Although this is in accordance with the current expert consensus guidelines for LQTS genetic testing in cases of autopsy-negative sudden unexplained death and SIDS,⁴⁰ we cannot exclude the presence of mutations involving these minor LQTS genes.⁹ In addition, our mutation analysis strategy, denaturing high-performance liquid chromatography followed by direct DNA sequencing, may miss homozygous mutations and large gene rearrangements including copy number variations that could be detected using alternative strategies.⁴¹ Therefore, our analysis may underestimate the true prevalence of LQTS-associated mutations in intrauterine fetal death.

In this molecular genetic evaluation of 91 cases of intrauterine fetal death, missense mutations associated with LQTS susceptibility were discovered in 3 cases (3.3%) and overall, genetic variants leading to dysfunctional LQTS-associated ion channels in vitro were discovered in 8 cases (8.8%). This preliminary evidence provided by our present study suggests that LQTS may contribute to the pathogenesis of some intrauterine fetal deaths.

Corresponding Author: Michael J. Ackerman, MD, PhD, Long QT Syndrome Clinic and the Mayo Clinic Windland Smith Rice Sudden Death Genomics Laboratory, Mayo Clinic, Guggenheim 501, 200 First St SW, Rochester, MN 55905 (ackerman.michael@mayo.edu).

Author Contributions: Drs Crotti and White and Mr Tester contributed equally to this manuscript and are considered coequal first authors and had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis. Drs George, Schwartz, and Ackerman are considered co-equal senior authors

Study concept and design: Crotti, Tester, White, Cetin, Van Dyke, Wick, Brost, George, Schwartz, Ackerman.

Acquisition of data: Tester, White, Bartos, Insolia, Besana, Kunic, Will, Velasco, Bair, Ghidoni, Cetin, Van Dyke, Brost, Facchinetti, George.

Analysis and interpretation of data: Crotti, Tester, White, Bartos, Insolia, Besana, Ghidoni, Van Dyke, Delisle, George, Schwartz, Ackerman.

Drafting of the manuscript: Crotti, Tester, White, Bartos, Insolia, Kunic, Velasco, Ghidoni, Delisle, George, Ackerman.

Critical revision of the manuscript for important intellectual content: Crotti, Tester, White, Bartos, Besana, Will, Bair, Cetin, Van Dyke, Wick, Brost, Facchinetti, George, Schwartz, Ackerman.

Statistical analysis: Bartos, Besana, Ghidoni, Delisle, George.

Obtained funding: Crotti, Cetin, Delisle, George, Ackerman.

Administrative, technical, or material support: Tester, White, Insolia, Kunic, Will, Velasco, Bair, Ghidoni, Van Dyke, Brost.

Study supervision: Crotti, Tester, White, Wick, Delisle, Facchinetti, George, Schwartz, Ackerman.

Funding/Support: Dr Spazzolini's work is supported by Fondazione IRCCS Policlinico S.Matteo, Pavia, Italy. This research is supported by the Mayo Clinic Windland Smith Rice Comprehensive Sudden Cardiac Death Program (Dr Ackerman), the Sheikh Zayed Saif Mohammed Al Nahyan Fund in Pediatric Cardiology Research (Dr Ackerman), the Dr Scholl Foundation (Dr Ackerman), the Hannah M. Wernke Memorial Foundation (Dr Ackerman), the National Institutes of Health (grants HD042569 to MJA, HL087039 to BPD, and HL083374 to Dr George), and the Italian Ministry of Health (GR-2010-2305717 to Dr Crotti).

Role of the Sponsor: The sponsors had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; and preparation, review, or approval of the manuscript.

Additional Contributions: We thank Carla Spazzolini, DVN, MS (Department of Cardiology, IRCCS Fondazione Policlinico S. Matteo, Pavia, Italy) for her statistical support of this article for which no compensation was received.

This article was corrected for errors on August 22, 2013.