ATP1A3‐Encoded Sodium‐Potassium ATPase Subunit Alpha 3 D801N Variant Is Associated With Shortened QT Interval and Predisposition to Ventricular Fibrillation Preceded by Bradycardia

Nonstandard Abbreviations and Acronyms

The alpha‐3 subunit of the sodium‐potassium ATPase (ATP1A3), encoded by the gene ATP1A3, is responsible for maintaining the resting plasma membrane electrochemical gradient in neurons.¹, ² Genetic variants in ATP1A3 are associated with several neurological disorders, namely rapid‐onset dystonia parkinsonism; alternating hemiplegia of childhood (AHC); and a rare syndrome of cerebellar ataxia, areflexia, pes cavus, optic atrophy, and sensorineural hearing loss.¹, ², ³, ⁴, ⁵, ⁶ Patients diagnosed with AHC experience a spectrum of clinical symptoms, with the disease‐defining features being sudden onset of episodes consisting of dystonia, quadriparesis, seizures/seizure‐like events, and oculomotor abnormalities before 18 months of age.¹, ⁴, ⁵, ⁶ Additionally, patients with AHC have an increased risk of mortality of 3.2 deaths per 1000 people per year, which increases to 4.5% by 29 years of age.⁷, ⁸

The increased mortality observed in patients with AHC is attributable to sudden death. Sudden unexpected death in epilepsy (SUDEP) has been observed in association with seizure activity in patients with AHC with concurrent epilepsy.⁷, ⁸, ⁹ SUDEP has also been observed in mouse models of AHC hosting the D801N and E815K variants of the ATP1A3 gene¹⁰, ¹¹; however, the mechanism remains unknown. A recent study focusing on ATP1A3‐related syndromes identified cardiac repolarization abnormalities, but the genetic underpinning of this finding remains unknown.¹² One potential explanation for sudden unexplained death in patients with AHC are cardiac arrhythmias, which may be fatal and have been associated with SUDEP in overlapping phenotypes.¹³ Although the neurologic impacts of ATP1A3 variants are well described, the role of ATP1A3 in the heart is not well described, and the cardiac manifestations of disease‐associated variants are not well delineated.

A known risk factor for the development of cardiac arrhythmias and a predisposition to sudden death is abnormal cardiac repolarization. Variants in genes encoding ion channels, which govern both depolarization and repolarization, have been shown to cause sudden death.¹⁴ The rate of repolarization can be measured using the QT interval, with a shortened QT interval reflecting accelerated repolarization. An abnormally short QT interval can predispose individuals to ventricular fibrillation, and therefore, is associated with sudden cardiac death.¹⁴ In this study, we aimed to determine if there is a unifying AHC genotype and phenotype associated with short corrected QT (QTc), and therefore, potentially an increased risk of sudden death. We demonstrate that the ATP1A3‐D801N variant is associated with markedly short QTc measurements, which suggests a possible predisposition to lethal ventricular arrhythmias preceded by bradycardia among patients with AHC. This finding may suggest a high‐risk genotype among patients with AHC and underscores a critical role for ATP1A3 in maintaining the timing of cardiac repolarization.

Methods

The data that support the findings of this study are available from the corresponding author upon reasonable request.

Participants

This study was approved by both the Baylor College of Medicine and Duke University School of Medicine institutional review boards, and all participants provided informed consent in accordance with the Declaration of Helsinki. Based on the current consensus on AHC diagnosis,⁹ we adopted minimal criteria to define our AHC cohort. Typical patients with AHC were defined as patients who fulfilled the Aicardi criteria, which consists of 4 clinical findings: episodes of alternating hemiplegia/hemiparesis, dystonia, abnormal eye movements, and epileptic seizures with events of autonomic dysfunction, all of which occur before 18 months of age.⁸, ¹⁵ Intellectual disability is also common. Patients experience dystonic attacks with distinctive notable triggers that can be identified such as heat and stress. Notably, sleep restores functionality following an attack, although patients still demonstrate abnormal movements between attacks.⁸ Atypical patients with AHC were defined as patients who did not fulfill the entire Aicardi criteria but possessed enough AHC features to be on the AHC spectrum.⁹ These patients had most of the major AHC features and did not fit into one of the other ATP1A3‐related syndromes.

Inclusion criteria for this study were (1) a diagnosis of typical or atypical AHC, (2) clinical records available at Duke University Hospitals or Texas Children’s Hospital, (3) documented informed consent, and (4) availability of clinical genetic test results for the major genes associated with AHC, including ATP1A3, ATP1A2, CACNA1A, SCN1A, and SCN2A. Patients found to have a pathogenic or likely pathogenic variant in one of these major genes were defined as genotype‐positive, whereas patients found to not have a pathogenic/likely pathogenic variant in one of these major genes were defined as genotype‐negative. Some patients included in this study did not have each of these major genes fully genotyped. Patients were excluded from analysis if (1) their diagnosis consisted of rapid‐onset dystonia parkinsonism or syndrome of cerebellar ataxia, areflexia, pes cavus, optic atrophy, and sensorineural hearing loss; (2) no ECG data were available, or (3) they were unwilling or unable to provide consent for the study.

Study Design

A retrospective chart review of patients meeting the inclusion criteria was conducted. Patient history and family history were obtained. Echocardiographic data, when available, and ECG data were collected. Patients were divided into subgroups consisting of (1) typical AHC versus atypical AHC, (2) genotype‐positive versus genotype‐negative, and (3) ATP1A3‐D801N variant positive versus all other ATP1A3 variants. For studies of genetic yield, only unrelated probands were used. The primary outcome measure was sudden death or documented ventricular fibrillation. Additional ECG characteristics examined were the QT interval, heart rate, PR duration, QRS duration, P‐wave axis, QRS‐wave axis, and T‐wave axis values between the cohorts. Early repolarization, ST morphology and elevation/depression, as well as the presence of U waves, T‐wave alternans, and J‐point elevation were also examined. Race and ethnicity were self‐reported. Medications at the time of ECG was noted for each patient, with a focus on medications known to modulate the QT interval.

ECG Evaluation

Patients with typical and atypical AHC received ECG and 24‐hour Holter monitor evaluation as part of routine care. All ECGs consisted of 12 leads and were recorded using a paper speed of 25 mm/s with an amplitude of 10 mm/mV. For patients with multiple ECGs, the first ECG study was used for all analyses. The parameters evaluated from each ECG were heart rate, PR duration, QRS duration, QT interval, P‐wave axis, QRS‐wave axis, and T‐wave axis. Additionally, the presence of early repolarization and presence of U waves and T‐wave alternans was recorded for each patient. The criteria for ST elevation or depression were the presence of elevation or depression in at least 2 leads that measured at least 0.1 mV. The QTc interval was calculated with Bazett’s formula (QTc=QT/RR^1/2)¹⁶ using the manually measured QT interval.¹⁷ The automated QT/QTc calculation was not reported or used in the analysis. The threshold set for shortened QTc interval was <370 milliseconds established by the Gollob criteria.¹⁸ To ensure reproducibility of QTc measurements, 2 pediatric electrophysiologists independently remeasured QT/QTc intervals blinded to clinical status and genotype. Both electrophysiologists manually remeasured the QT/QTc intervals of each ECG used in the study. Twenty ECGs were randomly selected to compare reproducibility of the measurements. The method for QT determination was based on Lepeschkin and Surawicz.¹⁹ Lead II was the preferential lead for remeasurement of QT/QTc, but if the tracing in lead II could not be used for measurement, lead V5 was used. The QT interval was defined as the beginning of the Q wave to the point where a tangent from the steepest part of the T‐wave downslope crossed the isoelectric baseline. If present, U waves were excluded. This method is demonstrated in Figure S1.

Cohort of SUDEP and Sudden Cardiac Arrest Referrals

To explore the possibility that ATP1A3 variants are an independent cause of SUDEP, or sudden cardiac arrest (SCA), or death, we established a cohort using deidentified variant information from the clinical genetic testing company Invitae (San Francisco, CA). A referral database of individuals referred for evaluation of cardiac arrhythmic disease were queried based on the following inclusion criteria: (1) referred to Invitae for clinical genetic testing and (2) referral diagnosis of syncope, seizures, abnormal ECG, abnormal QT interval, SCA, ventricular tachycardia, ventricular fibrillation, or channelopathy. The exclusion criteria were (1) variants determined to be likely benign and benign at the time of genetic testing, (2) individuals with a pathogenic/likely pathogenic variant in another gene associated with cardiac channelopathies or SCA, (3) cardiomyopathy as the indication for genetic testing, (4) atrial ectopy (premature atrial complex) or ventricular ectopy (premature ventricular complex) as the indication for genetic testing, (5) QT prolongation as the indication for genetic testing, and (6) no clear indication for testing available. All resulting variants that fulfilled inclusion criteria and did not meet exclusion criteria were subjected to the 2015 American College of Medical Genetics and Genomics guidelines²⁰ used for classifying the pathogenicity of genetic variants.

Statistical Analysis

Descriptive statistics were used to report the characteristics for patients. Values were expressed as mean±standard deviation unless otherwise noted. An independent t test and Mann‐Whitney test were used to compare parametric and nonparametric demographic data, respectively, between subgroups. An independent t test was used to compare mean heart rate, PR duration, QRS duration, QT interval, QTc interval, P‐wave axis, QRS‐wave axis, and T‐wave axis within each subgroup. A multivariate regression analysis was completed using sex, AHC presentation, and ATP1A3 genotype status as predictors of QTc measurement as the response. Interrater reliability was used to measure agreement between the manually remeasured QT intervals, with 90% agreement being deemed as satisfactory. Agreement was defined as a difference in measurement of ≤10 milliseconds, and disagreement was defined as a difference in measurement of >10 milliseconds. To assess changes in QTc measurements as a function of age, we compared QTc measurements <10 years of age and ≥10 years of age for all individuals and compared several groups including ATP1A3‐D801N–positive individuals, ATP1A3 genotype–positive individuals (excluding ATP1A3‐D801N), ATP1A3 genotype–negative, and atypical AHC.

Results

Cohort Characteristics

A total of 62 individuals met inclusion criteria. Fifty percent were female subjects, the mean age at diagnosis was 2.2±1.4 years, and the mean age at the most recent follow‐up was 11.7±9.4 years. One patient with AHC was evaluated at Baylor College of Medicine, and the remaining patients were evaluated at Duke University School of Medicine. Thirty‐four individuals were White (55%); 7 were Black (11%); 7 were Asian, multiracial, or Native Hawaiian/Pacific Islander (11%); and 14 patients did not report their race (23%). From a general cardiac perspective, a small minority, 2 of 62 (3%), had structural abnormalities by echocardiogram (1 atrial septal defect and 1 ventricular septal defect). No individuals in the study cohort had atrial arrhythmia, sinus pauses, or atrioventricular block. Furthermore, no individuals in the study had implantation of cardiac pacemakers or implantable cardioverter‐defibrillators. No individual underwent electrophysiology study or cardiac ablation, sympathetic denervation, placement of a ventricular assistive device, received extracorporeal membrane oxygenation, or cardiac surgery/transplant. Fifty‐six individuals (90%) met the criteria for typical AHC, whereas 6 patients (10%) met the criteria for atypical AHC criteria. Thirty‐seven patients (77.1%) were genotype‐positive for variants in ATP1A3. Of these 37 patients, 9 (24%) carried the D801N variant. There were no statistically significant differences in demographics between any of these subgroups. There was no significantly higher burden of QT modulatory medications or discernable differences among anticonvulsant therapy regimens between these subgroups. The cohort was composed of unrelated probands except for 3 families that included 2 separate pairs of siblings and 1 family consisting of a mother with familial hemiplegic migraine and her 2 children with AHC. These findings are summarized in Table 1 and Figure 1. Representative 12‐lead ECGs from patients in the ATP1A3‐D801N, ATP1A3‐E815K, and ATP1A3 genotype–negative cohorts are found in Figures S2 through S4. Although the ATP1A3‐D801N variant was the most common, several other variants were identified, including G89D (4/37), E815K (5/37), and G947R (1/37), among others (Table S1).

• Download figure
• Download PowerPoint

ATP1A3‐D801N Genotype Is Associated With a Shorter QTc Interval

Given the known incidence of SUDEP in patients with AHC, we investigated cardiac repolarization in our patient cohort. The interrater reliability between the 2 electrophysiologists for manually remeasuring the QT intervals was confirmed to be 90%. Comparison of QTc measurements among patients with typical AHC versus atypical AHC revealed that those with typical AHC had a shorter mean QTc; however, this difference did not meet statistical significance (387.4±4.6 milliseconds; 95% CI, 378.2–396.5 versus 400.8±15.3 milliseconds; 95% CI, 361.6–440.1, respectively; P=0.36). ATP1A3 genotype–positive individuals had a significantly shorter QTc measurement when compared with genotype‐negative individuals (375.4±5.6 milliseconds; 95% CI, 363.9–386.8 versus 408.4±4.8 milliseconds; 95% CI, 398.4–418.4; P<0.001). This reduction was driven by significantly shorter mean QTc measurements among subjects hosting the ATP1A3‐D801N variant when compared with subjects with all other ATP1A3 variants (344.9±7.0 milliseconds; 95% CI, 328.7–361.1 versus 385.1±6.0 milliseconds; 95% CI, 372.8–397.5; P<0.001). Furthermore, D801N variant–positive individuals had a shorter QT compared with ATP1A3 genotype–negative individuals (408.4±4.8 milliseconds; P<0.001).

The QTc of male subjects versus female subjects in the cohorts was analyzed. Within the ATP1A3‐D801N cohort, there was a statistically significant difference in QTc, with male subjects demonstrating a higher mean QTc of 357.6±5.4 (95% CI, 342.7–372.5) compared with female subjects , with a mean QTc of 329.0±9.9 (95% CI, 297.6–360.5; P=0.025). Conversely, there were no sex‐based difference in the ATP1A3 genotype–positive cohort or the ATP1A3 genotype–negative cohort.

We next compared the ATP1A3‐D801N QTc with the QTc of individuals hosting other ATP1A3 variants. The ATP1A3‐E815K variant is associated with more severe neurologic involvement,²¹ yet in our cohort, this group had a significantly longer QTc compared with D801N QTc (398.8±7.9 milliseconds; 95% CI, 376.8–420.8; P<0.001). There was no significant difference in the QTc between patients with the E815K variant and all other genotype‐positive patients who did not have the D801N variant (398.8±7.9 milliseconds versus 391.9±34.6 milliseconds; 95% CI, 379.8–404; P=0.25). Furthermore, the ATP1A3‐G89D variant QTc was similar to E815K and was longer than D801N (409±5.9 milliseconds; 95% CI, 383.8–434.2; P=0.47). None of these differences were because of demographic differences such as age, sex, and race between each of the comparison groups. These results are summarized in Figure 2. This association was similar when comparing QTc intervals between subgroups (Figure S5). We next completed a multivariate regression analysis to determine the relationship between sex, AHC presentation, and ATP1A3 genotype (predictors) to QTc measurement (response). ATP1A3 genotype status was significantly associated with QTc measurement (P<0.001). Patients with a positive ATP1A3 genotype have a 31.9‐millisecond decrease in QTc as compared with those with ATP1A3 genotype–negative status when accounting for sex (slope: 0.7 milliseconds, P=0.931) and AHC presentation (slope: 4.3, P=0.76). Overall, these findings suggest the ATP1A3‐D801N variant is associated with an abnormally short QT interval (QTc ~345 milliseconds) when compared with other ATP1A3 variants.

• Download figure
• Download PowerPoint

ATP1A3 Genotype Is Not Associated With Other Significant ECG Parameter Changes

To determine whether repolarization was the only electrocardiographic parameter impacted by the ATP1A3 genotype, we compared other ECG measurements among subgroups. The QRS duration did not significantly differ between any of the groups. Individuals with typical AHC had a slightly left‐shifted QRS axis compared with those with atypical AHC (63.1±3.6 versus 80.7±7.3; P=0.03). There were no other differences in ECG parameters. Additionally, when comparing the ECG parameters of patients who were genotype‐positive with those who were genotype‐negative, there were no significant differences in any of the examined ECG parameters. Finally, when the ATP1A3 genotype–positive individuals were compared with ATP1A3 genotype–negative individuals, there were no significant differences in any ECG parameters except a modest change in P‐wave axis (54.6±2.3 versus 43.3±4.3; P=0.01).

Other repolarization characteristics were assessed and were rarely abnormal without a clear association with genotype. Marked J‐point elevation was not seen. Mild ST elevation, which was consistent with early repolarization, was occasionally observed, most often with 1‐mV elevation in the anterolateral leads. Specifically, 2 out of the 9 patients (22%) with the ATP1A3‐D801N variant were found to have early repolarization in the anterolateral leads. When analyzing the inferior leads, 1 out of 9 patients (11%) with the ATP1A3‐D801N variant were found to have early repolarization. None of these patients had early repolarization in the septal leads. No septal lead J‐point elevation was observed in general. Abnormal T‐wave morphology was rare and found in 3 patients (4.8%) who demonstrated isoelectric T waves in inferior and anterolateral leads. U waves were not a common feature and were observed in 2 patients (3.2%). A summary of the ECG parameters can be found in Table 2. Overall, these findings suggest that repolarization timing (QT/QTc) is the predominant ECG difference when comparing AHC and ATP1A3 status.

ATP1A3‐D801N–Associated QTc Shortening Persists Into Adulthood

To determine whether QTc increases with age in patients with AHC, in particular those hosting the D801N variant, we compared QTc versus age at time of ECG study. The mean QTc for ATP1A3‐D801N individuals was 344.9 milliseconds (95% CI, 328.7–361.1) and did not increase significantly based on age. The QTc measurements remained <370 milliseconds throughout childhood and adulthood (age range, 0.87–43 years) with no significant difference between mean QTc <10 and ≥10 years of age. In contrast, 7 out of 28 patients with ATP1A3 genotype–positive not hosting the D801N variant demonstrated a shortened QTc <10 years of age, which increased with age (mean, 385.1 milliseconds; 95% CI, 372.8–397.5). Conversely, the vast majority of patients who were genotype‐negative and had atypical AHC did not have short QTc measurements during their lifespan (mean, 408.4 milliseconds; 95% CI, 398.4–418.4; mean, 413.6 milliseconds; 95% CI, 392.9–434.2, respectively). These results are summarized in Figure 3. These findings suggest that patients with the ATP1A3‐D801N variant have a lifelong shortened QTc. Furthermore, some patients who carry ATP1A3 variants other than D801N may be predisposed to a shortened QTc in early childhood, specifically.

• Download figure
• Download PowerPoint

ATP1A3‐D801N Predisposes to Ventricular Fibrillation With Sedation

Given the association between AHC and SUDEP, and our observation of a markedly reduced QTc in patients with the D801N variant, we next sought to determine whether QT shortening was associated with documented ventricular arrhythmic events. Two patients in our cohort with the ATP1A3‐D801N variant and typical AHC experienced documented recalcitrant ventricular fibrillation, which resulted in fatal cardiac arrest for one of these patients.

The first patient was a 13.8‐year‐old boy who was in his usual state of health when awakening from general anesthesia, using inhaled sevoflurane, for brain magnetic resonance imaging, as is routine at our institution. He developed significant bradycardia that was treated with glycopyrrolate, as well as atropine, which increased his heart rate from 30 beats per minute (bpm) 130 bpm . Subsequently, he became restless and was given 2 incremental doses of dexmedetomidine. Seizure‐like movements and irregular breathing were noted 3 to 4 minutes later. Because of concern for a seizure, midazolam was given, at which time the patient became pale and appeared cyanotic. Concordantly, he was noted to have no carotid or femoral pulse; therefore, cardiopulmonary resuscitation was initiated. Epinephrine was given, spontaneous circulation briefly returned, and he was endotracheally intubated. Cardiopulmonary resuscitation was initiated again because he was noted to have ventricular fibrillation on the monitor. Given that a secure airway had been established and he was adequately ventilated, it seemed plausible to rule out hypoxia as the cause for any bradycardia leading to subsequent arrhythmia. He was defibrillated 4 times before sinus rhythm was noted and maintained. During the cardiopulmonary resuscitation, he was given epinephrine 3 times, lidocaine, and sodium bicarbonate. Before the arrest, his ECG demonstrated a QTc measurement of 364 milliseconds. He had no prior episodes of unexplained syncope or documented arrhythmia. Following his arrest, the ECG demonstrated a QTc of 416 milliseconds which lengthened to 467 milliseconds in follow‐up 10 days after the event before subsequently normalizing.

The second patient was a 4.5‐year‐old girl who was admitted to the hospital because of status epilepticus. She was intubated and sedated using dexmedetomidine, fentanyl, hydromorphone, midazolam, and chloral hydrate, and was in a nonepileptic state. Before the arrest, she experienced 2 episodes of bradycardia marked by a heart rate of approximately 40 bpm. Subsequently, she developed ventricular fibrillation refractory to multiple defibrillations and administration of epinephrine, atropine, lidocaine, amiodarone, lidocaine, calcium, magnesium, and sodium bicarbonate. She ultimately died because of refractory ventricular fibrillation. Of note, her ECG at 8 months of age demonstrated a QTc value of 308 milliseconds. Immediately after arrest and before her death, her ECG demonstrated a QTc value of 514 milliseconds.

Bradycardia occurring just before arrest was a feature common in both of these patients. Both patients who arrested experienced QTc prolongation following the arrest. This finding is nonspecific to this patient population and is a well‐documented occurrence following cardiac arrest.²², ²³, ²⁴ No other subjects in our cohort experienced a documented ventricular arrhythmia, sudden death, or mortality. An ECG of the second patient before her arrest and cardiac telemetry during the arrest demonstrating ventricular tachycardia/ventricular fibrillation can be found in Figure S6. ECGs of the first patient at a time period before his arrest and following his arrest can be found in Figure S7. Interestingly, there was no prior history of syncope in the 2 individuals who experienced ventricular fibrillation, and only 1 individual out of 62 individuals in the entire cohort (2%) had a history of syncope or nonictal loss of consciousness. The individual with a history of syncope carried the ATP1A3‐E815K variant and had typical AHC. Overall, these findings suggest that patients with the ATP1A3‐D801N variant have an increased susceptibility to potentially fatal arrhythmias. Furthermore, because both patients who developed ventricular fibrillation lacked a history of syncope, the only indicator of a potential arrhythmia in the setting of sedation/general anesthesia may be bradycardia. This raises the possibility that risk of fatal arrhythmia is highest when sedation and bradycardia are present in individuals with shortened QTc at baseline (Figure 4).

• Download figure
• Download PowerPoint

ATP1A3 Variants Are Associated With Sudden Death and Sudden Cardiac Arrest

To examine if ATP1A3 variants are independently associated with sudden death or SCA, we used a cohort consisting of 6437 individuals referred for cardiovascular disease or SUDEP genetic testing. Among these, 8 individuals (0.12%) carried rare variants found in ATP1A3 that were otherwise negative for pathologic or likely pathologic variants in known sudden death or SCA genes (Table 3). All variants were classified as variants of uncertain significance at the time of genetic testing. Furthermore, when applying the 2015 American College of Medical Genetics and Genomics criteria to these variants, they were confirmed to be variants of uncertain significance. Within this group of 8 individuals, there were 3 broad categories of indications for genetic testing: an abnormal ECG/QT interval; possible channelopathy; or SCA, ventricular fibrillation, and ventricular tachycardia. A total of 5 out of 8 (62.5%) of these individuals had a referral indication of SCA or ventricular arrhythmia. Two of these 5 individuals had a reported positive family history of SCA. With this cohort of genotype‐negative individuals who experienced sudden unexplained death, these findings suggest that ATP1A3 variants may be associated with SCA and lethal ventricular arrhythmias with a less‐pronounced neurological phenotype.

Discussion

Assessing the underlying clinical phenotype of patients with AHC is critical, because this population is at increased risk for SUDEP; however, the underlying cause of the susceptibility to sudden death is not fully understood. This poses challenges in attempting to identify both individuals who are at risk of sudden death and exogenous risk factors that may place these individuals at even higher risk. Although our study contained one of the largest cohorts of patients with AHC and concurrent ATP1A3 variants with ECG data available, AHC is a rare condition. The exact prevalence has been reported to be approximately 1:100 000 children,²⁵ therefore, we used a retrospective design in our study. We acknowledge that the necessary retrospective nature of this study means that the findings reported are associative.

In the present study, we found that in patients with AHC who are ATP1A3 genotype–positive, and in particular those with the D801N variant, the timing of cardiac repolarization is affected as evidenced by a shorter QT interval compared with others in the cohort. We also showed that in patients with the D801N variant, the QTc was pathologically short, which suggests that these patients may have a predisposition to bradycardia causing potentially fatal ventricular arrhythmias. Our results raise the possibility that the cause of SUDEP among patients with AHC is because of abnormal repolarization and arrhythmia. A minority of patients with the ATP1A3‐D801N variant displayed early repolarization, especially in the anterolateral leads, and the clinical relevance of this remains unknown because it is a common feature in pediatric patients in the setting of normal repolarization times.²⁶, ²⁷, ²⁸, ²⁹ Furthermore, our analysis of QTc measurements as a function of age revealed that patients with the ATP1A3‐D801N variant have shortened QTc measurements that persist through childhood and adulthood. Although patients with other ATP1A3 variants displayed shortened QTc measurements in childhood, this propensity normalized with increasing age. The definition for short QTc adopted in this study comes from the proposed diagnostic criteria report by Gollob et al, which suggests that a QTc measurement <370 milliseconds can be pathologic.¹⁸

These data were further supported by patient survival, which suggested that individuals who experience ventricular fibrillation may have a series of risk factors present just before the arrest, including (1) a susceptible genotype, (2) a shortened QTc measurement, and (3) a sedation‐induced bradycardia. Collectively, these findings suggest that patients with the ATP1A3‐D801N variant may be predisposed to SUDEP because of shortened QTc measurements present throughout childhood and adulthood, particularly during times of anesthesia/sedation. Further investigation is needed to validate this finding and to determine the mechanism of disease, which can provide insight into actions and medications that should be included or omitted during necessary procedures for these patients.

One study utilizing an AHC registry documented 3 patients with AHC (D801N, P808L, and G947R variants) succumbing to sudden death because of cardiac dysfunction, and noted that their autopsy findings consisted of gross cardiology alterations.² One of the 3 patients in that study who experienced sudden death carried the ATP1A3‐D801N variant. This supports the understanding that patients with the ATP1A3‐D801N variant are more likely to experience sudden death, possibly resulting from a shortened QT interval, and potentially also gross cardiac repolarization abnormalities that may increase the likelihood of a fatal cardiac arrhythmia. Additionally, one study analyzed 52 ECGs of an age‐matched control group of patients with epilepsy to patients with AHC. They found that patients with AHC had significantly more repolarization abnormalities as well as a shorter mean QT interval,³⁰, ³¹ specifically in patients with the ATP1A3‐D801N variant.³¹ This finding is supported by the results of our study showing that patients with the ATP1A3‐D801N variant with typical AHC presentation are more likely to experience ECG abnormalities and have a shorter QTc. An additional alternative hypothesis not explored in this study is that individuals with ATP1A3 variants may be predisposed to apnea rather than cardiac arrhythmias.³² Sleep apnea has been documented as a risk factor for coronary events and cardiovascular death³³, ³⁴; however, a link, if any, to cardiac disease remains unclear. It is critical to understand that SCA is multifactorial and can be attributable to many causes as outlined in Figure 4.

Abnormalities in cardiac electrophysiology have been scarcely reported in patients with AHC and are rare in patients with rapid‐onset dystonia parkinsonism or syndrome of cerebellar ataxia, areflexia, pes cavus, optic atrophy, and sensorineural hearing loss. Previous work in principally neuronal models has revealed that differing protein variants in the alpha‐3 subunit of the ATP1A3 gene that result in the syndrome of cerebellar ataxia, areflexia, pes cavus, optic atrophy, and sensorineural hearing loss; rapid‐onset dystonia parkinsonism; and AHC cause specific disruptions that cause reduced affinity at the cytoplasmic surface, widely distributed protein dysfunction, or dysfunction at the cytoplasmic surface in the transmembrane domain, respectively.³⁰, ³¹, ³⁵, ³⁶, ³⁷, ³⁸, ³⁹ Although ATP1A3‐D801N is associated with a mild–moderate form of AHC from a neurological perspective,²¹, ⁴⁰ it has been implicated in more severe cardiac phenotypes.², ³¹ Studies focusing on the D801N variant show through homology modeling that the D801N variant creates a positive dipole which consequently affects the flow of potassium through the pump via electrostatic repulsion⁴¹ and reduced sodium‐potassium ATPase activity and pump current in mammalian cells.⁴², ⁴³, ⁴⁴ However, there remains a paucity of studies using cardiac myocytes and the mechanism of short‐QT pathogenesis in the setting of ATP1A3‐D801N, and the role of ATP1A3 in cardiac repolarization more broadly, remain open questions.

Limitations

The present study has several identifiable limitations. The study is a single‐center experience with a retrospective design that should be validated independently in other centers and prospectively, if possible. Furthermore, the link between ATP‐D801N and short QT is associative and should be evaluated in experimental models to clarify the mechanism. Although there was no difference in the presence of QT modulatory medications between subgroups, polypharmacy is common in this cohort with multiple medical disabilities. Thus, we are not able to exclude drug interactions or the impact of potential drug effects on the cardiac phenotypes noted for different genetic substrates. Furthermore, the use of a genotype‐negative sudden unexplained death cohort consisting of patients who experienced SCA and possess ATP1A3 variants is limited by lack of detailed clinical information confirming short QT in ECGs. It is also limited by the constraint that ATP1A3 as a gene is still under investigation for its role in cardiac disease. Therefore, applying the American College of Medical Genetics and Genomics criteria results in these variants satisfying the criteria for variant of unknown significance, because the ACMG criteria are based on the a priori association of a certain gene with a disease. Nonetheless, this genotype‐negative sudden unexplained death cohort provides important information that contributes to the hypothesis that ATP1A3 variants may be independently associated with SCA. Despite these limitations, our study provides important insights into the underlying pathophysiology associated with SUDEP in patients with AHC.

Conclusions

In this study, we report that patients who are ATP1A3 genotype–positive, particularly patients with the ATP1A3‐D801N variant, consistently experience short‐QTc measurements regardless of age. Also reported is that these persistent repolarization abnormalities may be associated with an increased predisposition to bradycardia, which can cause life‐threatening cardiac arrhythmias. It is crucial to monitor cardiac parameters in this patient population to reduce the likelihood of a fatal cardiac event.

Sources of Funding

M.M.M is supported by the Stead Scholarship Program of the Duke University Department of Medicine and the Basic Science Research Track Scholarship, sponsored by the Burroughs Wellcome Fund Physician‐Scientist Institutional Award to Duke University School of Medicine, Request ID# 1018899. J.E.E. is supported by the National Institutes of Health Clinical and Translational Science Award (UL1TR00255). A.P.L. is supported by NIH K08‐HL136839 and the Duke University School of Medicine.

Disclosures

M.A.M. has a provisional patent for treatment of AHC. J.G., R.N., R.T., T.E.C. received salary from and are stockholders of Invitae. The remaining authors have no disclosures to report.

Footnotes