ECG-only explainable deep learning algorithm predicts the risk for malignant ventricular arrhythmia in phospholamban cardiomyopathy.

BACKGROUND: Phospholamban (PLN) p.(Arg14del) variant carriers are at risk for development of malignant ventricular arrhythmia (MVA). Accurate risk stratification allows timely implantation of intracardiac defibrillators and is currently performed with a multimodality prediction model. OBJECTIVE: This study aimed to investigate whether an explainable deep learning-based approach allows risk prediction with only electrocardiogram (ECG) data. METHODS: A total of 679 PLN p.(Arg14del) carriers without MVA at baseline were identified. A deep learning-based variational auto-encoder, trained on 1.1 million ECGs, was used to convert the 12-lead baseline ECG into its FactorECG, a compressed version of the ECG that summarizes it into 32 explainable factors. Prediction models were developed by Cox regression. RESULTS: The deep learning-based ECG-only approach was able to predict MVA with a C statistic of 0.79 (95% CI, 0.76-0.83), comparable to the current prediction model (C statistic, 0.83 [95% CI, 0.79-0.88]; P = .054) and outperforming a model based on conventional ECG parameters (low-voltage ECG and negative T waves; C statistic, 0.65 [95% CI, 0.58-0.73]; P < .001). Clinical simulations showed that a 2-step approach, with ECG-only screening followed by a full workup, resulted in 60% less additional diagnostics while outperforming the multimodal prediction model in all patients. A visualization tool was created to provide interactive visualizations (https://pln.ecgx.ai). CONCLUSION: Our deep learning-based algorithm based on ECG data only accurately predicts the occurrence of MVA in PLN p.(Arg14del) carriers, enabling more efficient stratification of patients who need additional diagnostic testing and follow-up.

Arrhythmic risk stratification in arrhythmogenic right ventricular cardiomyopathy.

Arrhythmogenic right ventricular cardiomyopathy (ARVC) is a heritable cardiomyopathy characterized by a predominantly arrhythmic presentation. It represents the leading cause of sudden cardiac death (SCD) among athletes and poses a significant morbidity threat in the general population. As a causative treatment for ARVC is still not available, the placement of an implantable cardioverter defibrillator represents the current cornerstone for SCD prevention in this setting. Thanks to international ARVC-dedicated efforts, significant steps have been achieved in recent years towards an individualized, patient-centred risk stratification approach. A novel risk calculator algorithm estimating the 5-year risk of arrhythmias of patients with ARVC has been introduced in clinical practice and subsequently validated. The purpose of this article is to summarize the body of evidence that has allowed the development of this tool and to discuss the best way to implement its use in the care of an individual patient.

The arrhythmogenic cardiomyopathy phenotype associated with PKP2 c.1211dup variant.

BACKGROUND: The arrhythmogenic cardiomyopathy (ACM) phenotype, with life-threatening ventricular arrhythmias and heart failure, varies according to genetic aetiology. We aimed to characterise the phenotype associated with the variant c.1211dup (p.Val406Serfs*4) in the plakophilin­2 gene (PKP2) and compare it with previously reported Dutch PKP2 founder variants. METHODS: Clinical data were collected retrospectively from medical records of 106 PKP2 c.1211dup heterozygous carriers. Using data from the Netherlands ACM Registry, c.1211dup was compared with 3 other truncating PKP2 variants (c.235C > T (p.Arg79*), c.397C > T (p.Gln133*) and c.2489+1G > A (p.?)). RESULTS: Of the 106 carriers, 47 (44%) were diagnosed with ACM, at a mean age of 41 years. By the end of follow-up, 29 (27%) had experienced sustained ventricular arrhythmias and 12 (11%) had developed heart failure, with male carriers showing significantly higher risks than females on these endpoints (p < 0.05). Based on available cardiac magnetic resonance imaging and echocardiographic data, 46% of the carriers showed either right ventricular dilatation and/or dysfunction, whereas a substantial minority (37%) had some form of left ventricular involvement. Both geographical distribution of carriers and haplotype analysis suggested PKP2 c.1211dup to be a founder variant originating from the South-Western coast of the Netherlands. Finally, a Cox proportional hazards model suggested significant differences in ventricular arrhythmia-free survival between 4 PKP2 founder variants, including c.1211dup. CONCLUSIONS: The PKP2 c.1211dup variant is a Dutch founder variant associated with a typical right-dominant ACM phenotype, but also left ventricular involvement, and a possibly more severe phenotype than other Dutch PKP2 founder variants.

The added value of abnormal regional myocardial function for risk prediction in arrhythmogenic right ventricular cardiomyopathy.

AIMS: A risk calculator for individualized prediction of first-time sustained ventricular arrhythmia (VA) in arrhythmogenic right ventricular cardiomyopathy (ARVC) patients has recently been developed and validated (www.ARVCrisk.com). This study aimed to investigate whether regional functional abnormalities, measured by echocardiographic deformation imaging, can provide additional prognostic value. METHODS AND RESULTS: From two referral centres, 150 consecutive patients with a definite ARVC diagnosis, no prior sustained VA, and an echocardiogram suitable for deformation analysis were included (aged 41 ± 17 years, 50% female). During a median follow-up of 6.3 (interquartile range 3.1-9.8) years, 37 (25%) experienced a first-time sustained VA. All tested left and right ventricular (LV and RV) deformation parameters were univariate predictors for first-time VA. While LV function did not add predictive value in multivariate analysis, two RV deformation parameters did; RV free wall longitudinal strain and regional RV deformation patterns remained independent predictors after adjusting for the calculator-predicted risk [hazard ratio 1.07 (95% CI 1.02-1.11); P = 0.004 and 4.45 (95% CI 1.07-18.57); P = 0.040, respectively] and improved its discriminative value (from C-statistic 0.78 to 0.82 in both; Akaike information criterion change > 2). Importantly, all patients who experienced VA within 5 years from the echocardiographic assessment had abnormal regional RV deformation patterns at baseline. CONCLUSIONS: This study showed that regional functional abnormalities measured by echocardiographic deformation imaging can further refine personalized arrhythmic risk prediction when added to the ARVC risk calculator. The excellent negative predictive value of normal RV deformation could support clinicians considering the timing of implantable cardioverter defibrillator implantation in patients with intermediate arrhythmic risk.

Druggable proteins influencing cardiac structure and function: Implications for heart failure therapies and cancer cardiotoxicity.

Dysfunction of either the right or left ventricle can lead to heart failure (HF) and subsequent morbidity and mortality. We performed a genome-wide association study (GWAS) of 16 cardiac magnetic resonance (CMR) imaging measurements of biventricular function and structure. Cis-Mendelian randomization (MR) was used to identify plasma proteins associating with CMR traits as well as with any of the following cardiac outcomes: HF, non-ischemic cardiomyopathy, dilated cardiomyopathy (DCM), atrial fibrillation, or coronary heart disease. In total, 33 plasma proteins were prioritized, including repurposing candidates for DCM and/or HF: IL18R (providing indirect evidence for IL18), I17RA, GPC5, LAMC2, PA2GA, CD33, and SLAF7. In addition, 13 of the 25 druggable proteins (52%; 95% confidence interval, 0.31 to 0.72) could be mapped to compounds with known oncological indications or side effects. These findings provide leads to facilitate drug development for cardiac disease and suggest that cardiotoxicities of several cancer treatments might represent mechanism-based adverse effects.

Individualized Family Screening for Arrhythmogenic Right Ventricular Cardiomyopathy.

BACKGROUND: Clinical guidelines recommend regular screening for arrhythmogenic right ventricular cardiomyopathy (ARVC) to monitor at-risk relatives, resulting in a significant burden on clinical resources. Prioritizing relatives on their probability of developing definite ARVC may provide more efficient patient care. OBJECTIVES: The aim of this study was to determine the predictors and probability of ARVC development over time among at-risk relatives. METHODS: A total of 136 relatives (46% men, median age 25.5 years [IQR: 15.8-44.4 years]) from the Netherlands Arrhythmogenic Cardiomyopathy Registry without definite ARVC by 2010 task force criteria were included. Phenotype was ascertained using electrocardiography, Holter monitoring, and cardiac imaging. Subjects were divided into groups with "possible ARVC" (only genetic or familial predisposition) and "borderline ARVC" (1 minor task force criterion plus genetic or familial predisposition). Cox regression was performed to determine predictors and multistate modeling to assess the probability of ARVC development. Results were replicated in an unrelated Italian cohort (57% men, median age 37.0 years [IQR: 25.4-50.4 years]). RESULTS: At baseline, 93 subjects (68%) had possible ARVC, and 43 (32%) had borderline ARVC. Follow-up was available for 123 relatives (90%). After 8.1 years (IQR: 4.2-11.4 years), 41 (33%) had developed definite ARVC. Independent of baseline phenotype, symptomatic subjects (P = 0.014) and those 20 to 30 years of age (P = 0.002) had a higher hazard of developing definite ARVC. Furthermore, patients with borderline ARVC had a higher probability of developing definite ARVC compared with those with possible ARVC (1-year probability 13% vs 0.6%, 3-year probability 35% vs 5%; P < 0.01). External replication showed comparable results (P > 0.05). CONCLUSIONS: Symptomatic relatives, those 20 to 30 years of age, and those with borderline ARVC have a higher probability of developing definite ARVC. These patients may benefit from more frequent follow-up, while others may be monitored less often.

Circulating Biomarkers of Fibrosis Formation in Patients with Arrhythmogenic Cardiomyopathy.

Arrhythmogenic cardiomyopathy (ACM) is a progressive inheritable disease which is characterized by a gradual fibro-(fatty) replacement of the myocardium. Visualization of diffuse and patchy fibrosis patterns is challenging using clinically applied cardiac imaging modalities (e.g., late gadolinium enhancement, LGE). During collagen synthesis and breakdown, carboxy-peptides are released into the bloodstream, specifically procollagen type-I carboxy-terminal propeptides (PICP) and collagen type-I carboxy-terminal telopeptides (ICTP). We collected the serum and EDTA blood samples and clinical data of 45 ACM patients (age 50.11 ± 15.53 years, 44% female), divided into 35 diagnosed ACM patients with a 2010 ARVC Task Force Criteria score (TFC) ≥ 4, and 10 preclinical variant carriers with a TFC < 4. PICP levels were measured using an enzyme-linked immune sorbent assay and ICTP levels with a radio immunoassay. Increased PICP/ICTP ratios suggest a higher collagen deposition. We found significantly higher PICP and PICP/ICTP levels in diagnosed patients compared to preclinical variant carriers (p < 0.036 and p < 0.027). A moderate negative correlation existed between right ventricular ejection fractions (RVEF) and the PICP/ICTP ratio (r = -0.46, p = 0.06). In addition, significant correlations with left ventricular function (LVEF r = -0.53, p = 0.03 and end-systolic volume r = 0.63, p = 0.02) were found. These findings indicate impaired contractile performance due to pro-fibrotic remodeling. Follow-up studies including a larger number of patients should be performed to substantiate our findings and the validity of those levels as potential promising biomarkers in ACM.

Electrocardiographic findings in patients with arrhythmogenic cardiomyopathy and right bundle branch block ventricular tachycardia.

AIMS: Little is known about patients with right bundle branch block (RBBB)-ventricular tachycardia (VT) and arrhythmogenic cardiomyopathy (ACM). Our aims were: (i) to describe electrocardiogram (ECG) characteristics of sinus rhythm (SR) and VT; (ii) to correlate SR with RBBB-VT ECGs; and (iii) to compare VT ECGs with electro-anatomic mapping (EAM) data. METHODS AND RESULTS: From the European Survey on ACM, 70 patients with spontaneous RBBB-VT were included. Putative left ventricular (LV) sites of origin (SOOs) were estimated with a VT-axis-derived methodology and confirmed by EAM data when available. Overall, 49 (70%) patients met definite Task Force Criteria. Low QRS voltage predominated in lateral leads (n = 37, 55%), but QRS fragmentation was more frequent in inferior leads (n = 15, 23%). T-wave inversion (TWI) was equally frequent in inferior (n = 28, 42%) and lateral (n = 27, 40%) leads. TWI in inferior leads was associated with reduced LV ejection fraction (LVEF; 46 ± 10 vs. 53 ± 8, P = 0.02). Regarding SOOs, the inferior wall harboured 31 (46%) SOOs, followed by the lateral wall (n = 17, 25%), the anterior wall (n = 15, 22%), and the septum (n = 4, 6%). EAM data were available for 16 patients and showed good concordance with the putative SOOs. In all patients with superior-axis RBBB-VT who underwent endo-epicardial VT activation mapping, VT originated from the LV. CONCLUSIONS: In patients with ACM and RBBB-VT, RBBB-VTs originated mainly from the inferior and lateral LV walls. SR depolarization and repolarization abnormalities were frequent and associated with underlying variants.

Towards automatic classification of cardiovascular magnetic resonance Task Force Criteria for diagnosis of arrhythmogenic right ventricular cardiomyopathy.

BACKGROUND: Arrhythmogenic right ventricular cardiomyopathy (ARVC) is diagnosed according to the Task Force Criteria (TFC) in which cardiovascular magnetic resonance (CMR) imaging plays an important role. Our study aims to apply an automatic deep learning-based segmentation for right and left ventricular CMR assessment and evaluate this approach for classification of the CMR TFC. METHODS: We included 227 subjects suspected of ARVC who underwent CMR. Subjects were classified into (1) ARVC patients fulfilling TFC; (2) at-risk family members; and (3) controls. To perform automatic segmentation, a Bayesian Dilated Residual Neural Network was trained and tested. Performance of automatic versus manual segmentation was assessed using Dice-coefficient and Hausdorff distance. Since automatic segmentation is most challenging in basal slices, manual correction of the automatic segmentation in the most basal slice was simulated (automatic-basal). CMR TFC calculated using manual and automatic-basal segmentation were compared using Cohen's Kappa (κ). RESULTS: Automatic segmentation was trained on CMRs of 70 subjects (39.6 ± 18.1 years, 47% female) and tested on 157 subjects (36.9 ± 17.6 years, 59% female). Dice-coefficient and Hausdorff distance showed good agreement between manual and automatic segmentations (≥ 0.89 and ≤ 10.6 mm, respectively) which further improved after simulated correction of the most basal slice (≥ 0.92 and ≤ 9.2 mm, p < 0.001). Pearson correlation of volumetric and functional CMR measurements was good to excellent (automatic (r = 0.78-0.99, p < 0.001) and automatic-basal (r = 0.88-0.99, p < 0.001) measurements). CMR TFC classification using automatic-basal segmentations was comparable to manual segmentations (κ 0.98 ± 0.02) with comparable diagnostic performance. CONCLUSIONS: Combining automatic segmentation of CMRs with correction of the most basal slice results in accurate CMR TFC classification of subjects suspected of ARVC.

Programmed Ventricular Stimulation as an Additional Primary Prevention Risk Stratification Tool in Arrhythmogenic Right Ventricular Cardiomyopathy: A Multinational Study.

BACKGROUND: A novel risk calculator based on clinical characteristics and noninvasive tests that predicts the onset of clinical sustained ventricular arrhythmias (VA) in patients with arrhythmogenic right ventricular cardiomyopathy (ARVC) has been proposed and validated by recent studies. It remains unknown whether programmed ventricular stimulation (PVS) provides additional prognostic value. METHODS: All patients with a definite ARVC diagnosis, no history of sustained VAs at diagnosis, and PVS performed at baseline were extracted from 6 international ARVC registries. The calculator-predicted risk for sustained VA (sustained or implantable cardioverter defibrillator treated ventricular tachycardia [VT] or fibrillation, [aborted] sudden cardiac arrest) was assessed in all patients. Independent and combined performance of the risk calculator and PVS on sustained VA were assessed during a 5-year follow-up period. RESULTS: Two hundred eighty-eight patients (41.0±14.5 years, 55.9% male, right ventricular ejection fraction 42.5±11.1%) were enrolled. At PVS, 137 (47.6%) patients had inducible ventricular tachycardia. During a median of 5.31 [2.89-10.17] years of follow-up, 83 (60.6%) patients with a positive PVS and 37 (24.5%) with a negative PVS experienced sustained VA (P<0.001). Inducible ventricular tachycardia predicted clinical sustained VA during the 5-year follow-up and remained an independent predictor after accounting for the calculator-predicted risk (HR, 2.52 [1.58-4.02]; P<0.001). Compared with ARVC risk calculator predictions in isolation (C-statistic 0.72), addition of PVS inducibility showed improved prediction of VA events (C-statistic 0.75; log-likelihood ratio for nested models, P<0.001). PVS inducibility had a 76% [67-84] sensitivity and 68% [61-74] specificity, corresponding to log-likelihood ratios of 2.3 and 0.36 for inducible (likelihood ratio+) and noninducible (likelihood ratio-) patients, respectively. In patients with a ARVC risk calculator-predicted risk of clinical VA events <25% during 5 years (ie, low/intermediate subgroup), PVS had a 92.6% negative predictive value. CONCLUSIONS: PVS significantly improved risk stratification above and beyond the calculator-predicted risk of VA in a primary prevention cohort of patients with ARVC, mainly for patients considered to be at low and intermediate risk by the clinical risk calculator.

Prevalence and Disease Expression of Pathogenic and Likely Pathogenic Variants Associated With Inherited Cardiomyopathies in the General Population.

BACKGROUND: Pathogenic and likely pathogenic variants associated with arrhythmogenic right ventricular cardiomyopathy (ARVC), dilated cardiomyopathy (DCM), and hypertrophic cardiomyopathy (HCM) are recommended to be reported as secondary findings in genome sequencing studies. This provides opportunities for early diagnosis, but also fuels uncertainty in variant carriers (G+), since disease penetrance is incomplete. We assessed the prevalence and disease expression of G+ in the general population. METHODS: We identified pathogenic and likely pathogenic variants associated with ARVC, DCM and/or HCM in 200 643 UK Biobank individuals, who underwent whole exome sequencing. We calculated the prevalence of G+ and analyzed the frequency of cardiomyopathy/heart failure diagnosis. In undiagnosed individuals, we analyzed early signs of disease expression using available electrocardiography and cardiac magnetic resonance imaging data. RESULTS: We found a prevalence of 1:578, 1:251, and 1:149 for pathogenic and likely pathogenic variants associated with ARVC, DCM and HCM respectively. Compared with controls, cardiovascular mortality was higher in DCM G+ (odds ratio 1.67 [95% CI 1.04; 2.59], P=0.030), but similar in ARVC and HCM G+ (P≥0.100). Cardiomyopathy or heart failure diagnosis were more frequent in DCM G+ (odds ratio 3.66 [95% CI 2.24; 5.81], P=4.9×10-7) and HCM G+ (odds ratio 3.03 [95% CI 1.98; 4.56], P=5.8×10-7), but comparable in ARVC G+ (P=0.172). In contrast, ARVC G+ had more ventricular arrhythmias (P=3.3×10-4). In undiagnosed individuals, left ventricular ejection fraction was reduced in DCM G+ (P=0.009). CONCLUSIONS: In the general population, pathogenic and likely pathogenic variants associated with ARVC, DCM, or HCM are not uncommon. Although G+ have increased mortality and morbidity, disease penetrance in these carriers from the general population remains low (1.2-3.1%). Follow-up decisions in case of incidental findings should not be based solely on a variant, but on multiple factors, including family history and disease expression.

Longitudinal Prediction of Ventricular Arrhythmic Risk in Patients With Arrhythmogenic Right Ventricular Cardiomyopathy.

BACKGROUND: The arrhythmogenic right ventricular cardiomyopathy (ARVC) risk calculator stratifies risk for incident sustained ventricular arrhythmias (VA) at the time of ARVC diagnosis. However, included risk factors change over time, and how well the ARVC risk calculator performs at follow-up is unknown. METHODS: This was a retrospective analysis of patients with definite ARVC and without prior sustained VA. Risk factors for VA including age, nonsustained ventricular tachycardia, premature ventricular complex burden, T-wave inversions on electrocardiogram, cardiac syncope, right ventricular function, therapeutic medication use, and exercise intensity were assessed at the time of 2010 Task Force Criteria based ARVC diagnosis and upon repeat evaluations. Changes in these risk factors were analyzed over 5-year follow-up. The 5-year risk of VA was predicted longitudinally using (1) the baseline ARVC risk calculator prediction, (2) the ARVC risk prediction calculated using updated risk factors, and (3) time-varying Cox regression. Discrimination and calibration were assessed in comparison to observed VA event rates. RESULTS: Four hundred eight patients with ARVC experiencing 132 primary VA events were included. Matched comparison of risk factors at baseline versus at 5 years of follow-up revealed decreased burdens of premature ventricular complexes (-1200/day) and nonsustained ventricular tachycardia (-14%). Presence of significant right ventricular dysfunction and number of T-wave inversions on electrocardiogram were unchanged. Observed risk for VA decreased by 13% by 5 years follow-up. The baseline ARVC risk calculator's ability to predict 5-year VA risk worsened during follow-up (C statistics, 0.83 at diagnosis versus 0.68 at 5 years). Both the updated ARVC risk calculator (C statistics of 0.77) and time-varying Cox regression model (C statistics, 0.77) had strong discrimination. The updated ARVC risk calculator overestimated 5-year VA risk by an average of +6%. CONCLUSIONS: Risk factors for VA in ARVC are dynamic, and overall risk for incident sustained VA decreases during follow-up. Up-to-date risk factor assessment improves VA risk stratification.

Value of genetic testing in the diagnosis and risk stratification of arrhythmogenic right ventricular cardiomyopathy.

BACKGROUND: Arrhythmogenic right ventricular cardiomyopathy (ARVC) is characterized by risk of malignant ventricular arrhythmia (VA). ARVC is diagnosed using an array of clinical tests in the consensus-based Task Force Criteria (TFC), one of which is genetic testing. OBJECTIVE: The purpose of this study was to investigate the value of genetic testing in diagnosing ARVC and its relation to the occurrence of first malignant VA. METHODS: A multicenter cohort of patients with ARVC was scored using the revised 2010 TFC with and without genetic criterion, analyzing any resulting loss or delay of diagnosis. Malignant VA was defined as sustained VA (≥30-second duration at ≥100 beats/min or requiring intervention). RESULTS: We included 402 subjects (221 [55%] male; 216 [54%] proband; 40 [27-51] years old at presentation) who were diagnosed with definite ARVC. A total of 232 subjects (58%) fulfilled genetic testing criteria. Removing the genetic criterion caused loss of diagnosis in 18 patients (4%) (11 of 216 probands [5%] and 7 of 186 relatives [4%]) and delay of diagnosis by ≥30 days in 22 patients (5%) (21 of 216 probands [10%] and 1 of 186 relative [0.5%]). A first malignant VA occurred in no patients who lost diagnosis and in 3 patients (3 of 216 probands [1%] and no relatives) during their diagnosis delay, none fatal. Time-to-event analysis showed no significant difference in time from diagnosis to malignant VA between pathogenic variant carriers and noncarriers. CONCLUSION: Disregarding the genetic criterion of the TFC caused loss or delay of diagnosis in 10% of patients with ARVC (40 of 402). Malignant VA occurred in 1% of cases with lost or delayed diagnosis (3 of 402), none fatal.

Life-threatening ventricular arrhythmia prediction in patients with dilated cardiomyopathy using explainable electrocardiogram-based deep neural networks.

AIMS: While electrocardiogram (ECG) characteristics have been associated with life-threatening ventricular arrhythmias (LTVA) in dilated cardiomyopathy (DCM), they typically rely on human-derived parameters. Deep neural networks (DNNs) can discover complex ECG patterns, but the interpretation is hampered by their 'black-box' characteristics. We aimed to detect DCM patients at risk of LTVA using an inherently explainable DNN. METHODS AND RESULTS: In this two-phase study, we first developed a variational autoencoder DNN on more than 1 million 12-lead median beat ECGs, compressing the ECG into 21 different factors (F): FactorECG. Next, we used two cohorts with a combined total of 695 DCM patients and entered these factors in a Cox regression for the composite LTVA outcome, which was defined as sudden cardiac arrest, spontaneous sustained ventricular tachycardia, or implantable cardioverter-defibrillator treated ventricular arrhythmia. Most patients were male (n = 442, 64%) with a median age of 54 years [interquartile range (IQR) 44-62], and median left ventricular ejection fraction of 30% (IQR 23-39). A total of 115 patients (16.5%) reached the study outcome. Factors F8 (prolonged PR-interval and P-wave duration, P < 0.005), F15 (reduced P-wave height, P = 0.04), F25 (increased right bundle branch delay, P = 0.02), F27 (P-wave axis P < 0.005), and F32 (reduced QRS-T voltages P = 0.03) were significantly associated with LTVA. CONCLUSION: Inherently explainable DNNs can detect patients at risk of LTVA which is mainly driven by P-wave abnormalities.

Automatic Identification of Patients With Unexplained Left Ventricular Hypertrophy in Electronic Health Record Data to Improve Targeted Treatment and Family Screening.

Background: Unexplained Left Ventricular Hypertrophy (ULVH) may be caused by genetic and non-genetic etiologies (e.g., sarcomere variants, cardiac amyloid, or Anderson-Fabry's disease). Identification of ULVH patients allows for early targeted treatment and family screening. Aim: To automatically identify patients with ULVH in electronic health record (EHR) data using two computer methods: text-mining and machine learning (ML). Methods: Adults with echocardiographic measurement of interventricular septum thickness (IVSt) were included. A text-mining algorithm was developed to identify patients with ULVH. An ML algorithm including a variety of clinical, ECG and echocardiographic data was trained and tested in an 80/20% split. Clinical diagnosis of ULVH was considered the gold standard. Misclassifications were reviewed by an experienced cardiologist. Sensitivity, specificity, positive, and negative likelihood ratios (LHR+ and LHR-) of both text-mining and ML were reported. Results: In total, 26,954 subjects (median age 61 years, 55% male) were included. ULVH was diagnosed in 204/26,954 (0.8%) patients, of which 56 had amyloidosis and two Anderson-Fabry Disease. Text-mining flagged 8,192 patients with possible ULVH, of whom 159 were true positives (sensitivity, specificity, LHR+, and LHR- of 0.78, 0.67, 2.36, and 0.33). Machine learning resulted in a sensitivity, specificity, LHR+, and LHR- of 0.32, 0.99, 32, and 0.68, respectively. Pivotal variables included IVSt, systolic blood pressure, and age. Conclusions: Automatic identification of patients with ULVH is possible with both Text-mining and ML. Text-mining may be a comprehensive scaffold but can be less specific than machine learning. Deployment of either method depends on existing infrastructures and clinical applications.

A new prediction model for ventricular arrhythmias in arrhythmogenic right ventricular cardiomyopathy.

AIMS: Arrhythmogenic right ventricular dysplasia/cardiomyopathy (ARVC) is characterized by ventricular arrhythmias (VAs) and sudden cardiac death (SCD). We aimed to develop a model for individualized prediction of incident VA/SCD in ARVC patients. METHODS AND RESULTS: Five hundred and twenty-eight patients with a definite diagnosis and no history of sustained VAs/SCD at baseline, aged 38.2 ± 15.5 years, 44.7% male, were enrolled from five registries in North America and Europe. Over 4.83 (interquartile range 2.44-9.33) years of follow-up, 146 (27.7%) experienced sustained VA, defined as SCD, aborted SCD, sustained ventricular tachycardia, or appropriate implantable cardioverter-defibrillator (ICD) therapy. A prediction model estimating annual VA risk was developed using Cox regression with internal validation. Eight potential predictors were pre-specified: age, sex, cardiac syncope in the prior 6 months, non-sustained ventricular tachycardia, number of premature ventricular complexes in 24 h, number of leads with T-wave inversion, and right and left ventricular ejection fractions (LVEFs). All except LVEF were retained in the final model. The model accurately distinguished patients with and without events, with an optimism-corrected C-index of 0.77 [95% confidence interval (CI) 0.73-0.81] and minimal over-optimism [calibration slope of 0.93 (95% CI 0.92-0.95)]. By decision curve analysis, the clinical benefit of the model was superior to a current consensus-based ICD placement algorithm with a 20.3% reduction of ICD placements with the same proportion of protected patients (P < 0.001). CONCLUSION: Using the largest cohort of patients with ARVC and no prior VA, a prediction model using readily available clinical parameters was devised to estimate VA risk and guide decisions regarding primary prevention ICDs (www.arvcrisk.com).