Expanded adaptation of an artificial intelligence model for predicting chemotherapy-induced cardiotoxicity using baseline electrocardiograms

Introduction

While effective as chemotherapeutics, anthracyclines can cause cancer therapy-related cardiac dysfunction (CTRCD), which adversely affects the prognosis of patients with malignancies1–5. Despite guideline recommendations6–9, repeated echocardiograms are rarely performed10 with delayed diagnosis of CTRCD leading to unrecoverable cardiac dysfunction11. Recently, artificial intelligence (AI) was shown to be capable of detecting reduced left ventricular ejection fraction (LVEF) solely from electrocardiogram (ECG)12. Furthermore, this model was predictive of a future decrease in LVEF. Therefore, we hypothesized that an AI model detecting reduced LVEF (AI-EF model) could predict CTRCD from ECGs.

Purpose

To assess whether the AI-EF model could detect patients at a high risk of CTRCD by analyzing ECGs taken immediately prior to the initiation of cardiotoxic chemotherapy.

Methods

Among patients who received chemotherapy with a regimen including anthracyclines in two institutions between June 1st, 2015 and October 1st, 2020, those who underwent both an ECG and echocardiogram ≤90 days prior to initial treatment were selected. The ECGs were analyzed by the AI-EF model and patients were divided into two groups according to the scores from the model. CTRCD was defined as LVEF <53% and ≥10% decrease in LVEF from the baseline at any time after the start of chemotherapy13. The cumulative incidence of CTRCD was compared for the two groups using Kaplan-Meier curves, log-rank test, a univariate Cox proportional hazard model, and a multivariable Cox proportional hazard model adjusting for known risk factors for CTRCD. Finally, a prediction model for CTRCD using readily available clinical variables with the AI-EF score was compared with the model using the same variables without the AI-EF score.

Results

1,158 patients were included in this study. 99 of them developed CTRCD during follow-up. The AI-EF model displayed excellent risk stratification of developing CTRCD: while 7.1% in the low AI-EF score group developed CTRCD, 12.9% of the patients in the high AI-EF score group developed CTRCD (hazard ratio (HR), 2.14; 95% confidence interval (CI), 1.43–3.19; log-rank p<0.001; Figure 1). This finding was robust across subgroups such as cancer types, the initial dose of anthracycline and baseline LVEF, and consistent after adjusting for multiple risk factors (adjusted HR, 2.10; 95% CI, 1.37–3.22; p<0.001; Figure 2). Furthermore, the addition of the AI-EF score significantly improved the accuracy of predicting CTRCD compared to clinical features alone (time-dependent area under the received operating curve (AUROC) for 2 years, 77.1; 95% CI, 71.8–82.3 for the model with AI-EF score and AUROC 73.9; 95% CI, 69.0–80.1 for the model without AI-EF score; p=0.02).

Conclusion

The AI-EF model, by utilizing baseline ECG, could stratify patients according to the risk of CTRCD and robustly augmented CTRCD prediction.

Funding Acknowledgement

Type of funding sources: Private grant(s) and/or Sponsorship. Main funding source(s): American Heart AssociationVerily

Generating and evaluating a ranked candidate gene list for potential vertebrate heart field regulators

The vertebrate heart develops from two distinct lineages of cardiomyocytes that arise from the first and second heart fields (FHF and SHF, respectively). The FHF forms the primitive heart tube, while adding cells from the SHF allows elongation at both poles of the tube. Initially seen as an exclusive characteristic of higher vertebrates, recent work has demonstrated the presence of a distinct FHF and SHF in lower vertebrates, including zebrafish. We found that key transcription factors that regulate septation and chamber formation in higher vertebrates, including Tbx5 and Pitx2, influence relative FHF and SHF contributions to the zebrafish heart tube. To identify molecular modulators of heart field migration, we used microarray-based expression profiling following inhibition of tbx5a and pitx2ab in embryonic zebrafish (Mosimann & Panakova, et al, 2015; GSE70750). Here, we describe in more detail the procedure used to process, prioritize, and analyze the expression data for functional enrichment.

Electrophysiologic characteristics of accessory atrioventricular connections in an inherited form of Wolff-Parkinson-White syndrome

INTRODUCTION

A familial form of Wolff-Parkinson-White syndrome (WPW) occurs in association with hypertrophic cardiomyopathy and intraventricular conduction abnormalities. This syndrome, demonstrating autosomal dominant inheritance and segregating with a high degree of penetrance but variable expressivity, has been genetically linked to chromosome 7q3. The purpose of this study is to detail the electrophysiologic characteristics of accessory atrioventricular connections (AC) in four members of a kindred with this syndrome.

METHODS AND RESULTS

We clinically evaluated 32 members of a single kindred and identified 20 individuals with ventricular preexcitation, abnormal intraventricular conduction including complete AV block and/or ventricular hypertrophy. Genetic linkage analysis mapped the disease gene in this kindred to the chromosome 7q3 locus (maximum logarithm of the odds score = 6.88, theta = 0); recombination events in affected individuals reduced the genetic interval from 7 centimorgans (cM) to 5 cM. Electrophysiologic study of four individuals with preexcitation, identified seven AC (1 right sided, 3 septal, and 3 left sided). All four individuals had inducible orthodromic tachycardia; while three had multiple AC. Bidirectional conduction was demonstrated in 6 of 7 AC. Successful ablation was accomplished in 5 of 7 AC.

CONCLUSION

The electrophysiologic characteristics and location of AC in family members having this complex cardiac phenotype are similar to those seen in individuals with isolated WPW. Identification of WPW in more than one family member should prompt clinical evaluation of relatives for additional findings of ventricular hypertrophy or conduction abnormalities.

Genetic heterogeneity of familial atrial myxoma syndromes (Carney complex)

Familial atrial myxomas may present as a component of the Carney complex, an autosomal dominant multiple neoplasia and lentiginosis syndrome. Although a gene defect for the Carney complex in some families maps to chromosome 2p, this syndrome is genetically heterogeneous, and other genetic loci also produce familial atrial myxomas.

New understandings in the genetics of congenital heart disease

In spite of the amazing success during the past half century in diagnosis and treatment of congenital heart disease, very little is known of its cause. However, a genetic cause has been clearly established for many forms of cardiovascular disease, and new understandings in the molecular genetics of congenital heart disease provide further insight. Progress has been quite impressive for some cardiovascular abnormalities, whereas in other areas the findings are more preliminary. For example, the molecular genetic cause of supravalvular aortic stenosis and the heart disease associated with Marfan syndrome has been clearly established. Impressive progress has been made in conotruncal defects, Holt-Oram syndrome, Alagille syndrome, and total anomalous pulmonary venous connection. In other areas, such as patent ductus arteriosus and atrioventricular septal defect, the findings are more preliminary. Taken as a whole, the prospect of understanding the genetic basis of congenital heart disease has never been better. Understanding the cause of congenital heart disease will provide new insights into both normal cardiac development and the pathophysiologic basis of congenital heart defects.

Identification, characterization, and chromosomal localization of the human homolog (hES) of ES/130

The chicken extracellular matrix glycoprotein ES/ 130 is necessary for epithelial--mesenchymal transformation in the developing heart and is also expressed in noncardiac chicken tissues such as limb and notochord. We have identified hES, the human homolog of chicken ES/130. Fluorescence in situ hybridization analysis (FISH) localizes hES to human chromosome 20p11.2-p12. FISH analyses of individuals with 20p12 deletions and affected by Alagille syndrome exclude hES as a candidate gene for this disorder. Reverse transcriptase-polymerase chain reaction studies reveal that hES is expressed in both fetal and adult human tissues and that hES expression in the left ventricle is increased in the failing adult heart. Further studies will evaluate how hES mutations may relate to congenital human cardiac and skeletal anomalies as well as cardiac remodeling in the adult.

Missense mutation in the pore region of HERG causes familial long QT syndrome

BACKGROUND

Long QT syndrome (LQT) is an inherited cardiac disorder that results in syncope, seizures, and sudden death. In a family with LQT, we identified a novel mutation in human ether-a-go-go-related gene (HERG), a voltage-gated potassium channel.

METHODS AND RESULTS

We used DNA sequence analysis, restriction enzyme digestion analysis, and allele-specific oligonucleotide hybridization to identify the HERG mutation. A single nucleotide substitution of thymidine to guanine (T1961G) changed the coding sense of HERG from isoleucine to arginine (Ile593Arg) in the channel pore region. The mutation was present in all affected family members; the mutation was not present in unaffected family members or in 100 normal, unrelated individuals.

CONCLUSIONS

We conclude that the Ile593Arg missense mutation in HERG is the cause of LQT in this family because it segregates with disease, its presence was confirmed in three ways, and it is not found in normal individuals. The Ile593Arg mutation may result in a change in potassium selectivity and permeability leading to a loss of HERG function, thereby resulting in LQT.

Familial Hypertrophic cardiomyopathy with Wolff-Parkinson-White syndrome maps to a locus on chromosome 7q3

We have mapped a disease locus for Wolff-Parkinson-White syndrome (WPW) and familial hypertrophic cardiomyopathy (FHC) segregating in a large kindred to chromosome 7 band q3. Although WPW syndrome and FHC have been observed in members of the same family in prior studies, the relationship between these two diseases has remained enigmatic. A large family with 25 surviving individuals who are affected by one or both of these conditions was studied. The disease locus is closely linked to loci D7S688, D7S505, and D7S483 (maximum two point LOD score at D7S505 was 7.80 at theta = 0). While four different FHC loci have been described this is the first locus that can be mutated to cause both WPW and/or FHC.

Genetic heterogeneity of heart-hand syndromes

BACKGROUND

Heart-hand syndromes compose a class of combined congenital cardiac and limb deformities. The proto-typical heart-hand disorder is Holt-Oram syndrome, which is characterized by cardiac septation defects and radial ray limb deformity. We have recently mapped the Holt-Oram syndrome gene defect to the long arm of human chromosome 12 in two families. The role of this disease locus in the pathogenesis of related conditions such as heart-hand syndrome type III (cardiac conduction disease accompanied by skeletal malformations) or familial atrial septal defects is unknown.

METHODS AND RESULTS

Clinical evaluations and genetic linkage analyses were performed in five additional kindreds with Holt-Oram syndrome and also in one kindred with heart-hand syndrome type III and one kindred with familial atrial septal defect and conduction disease. Holt-Oram syndrome in all five kindreds mapped to chromosome 12q2. These studies and previous data provide odds of greater than 10(25):1 that the Holt-Oram syndrome disease gene is at chromosome 12q2. In contrast, neither the phenotypically similar disorder heart-hand syndrome type III nor the locus responsible for a familial atrial septal defect with atrioventricular block maps to chromosome 12q2.

CONCLUSIONS

We demonstrate that heart-hand syndromes are genetically heterogeneous. Conditions that clinically appear to be partial phenocopies of Holt-Oram syndrome arise from distinct disease genes.

An evaluation of ribonuclease protection assays for the detection of beta-cardiac myosin heavy chain gene mutations

BACKGROUND

Ribonuclease (RNase) protection has been used to identify beta-cardiac myosin heavy chain (MHC) gene mutations that cause familial hypertrophic cardiomyopathy (FHC). Since more than 10 different mutations within this gene have been demonstrated to cause FHC in unrelated individuals, the genetic diagnosis of this condition will involve screening the beta-MHC gene. The accuracy with which RNase protection identifies such mutations is critical to defining the utility of this methodology in detecting mutations that cause FHC.

METHODS AND RESULTS

Twelve unrelated individuals with FHC were selected for further study because their beta-MHC genes had been screened for mutations by use of RNase protection, and no mutation was found. We performed linkage analysis of the families of these 12 probands using polymorphic short tandem repeats within the beta-MHC gene to determine whether FHC was genetically linked to the MHC locus on chromosome 14. FHC was not genetically linked to the MHC locus in 11 families whose beta-cardiac MHC gene did not contain mutations detectable by RNase protection.

CONCLUSIONS

We conclude that RNase protection is a sensitive method for screening for mutations within the beta-cardiac MHC gene. Further, mutations in the noncoding regions of the beta-MHC gene and mutations in the alpha-cardiac MHC gene are not a common cause of FHC. Negative RNase protection assays of affected individuals suggest that their FHC is due to mutations at other loci.

Need for invasive cardiological assessment and intervention: a ten year review

The uptake of invasive cardiological investigation and cardiopulmonary bypass procedures by the North West Surrey Health District was audited over the years 1979-88. Growth was almost continuous throughout the ten year period. The need within the district each year for coronary angiography seemed to be between 111 and 171 and for surgical revascularisation of the myocardium between 63 and 96 procedures; the first figure is the mean of the second quinquennial period (1984-88) and the second figure the total for 1988. After correction for the standardised mortality ratio and catchment area, the national requirement should lie between 690 and 1070 coronary angiograms and 390 and 600 coronary artery bypass graft operations per million population each year. There is a further national requirement for 70 valvar heart operations and 30 miscellaneous procedures per million population each year. Owing to delays in the provision of services, 20 patients died of a cardiovascular cause while they were on the waiting list for investigation or surgery. In the United Kingdom the annual target to be achieved by 1990 was 300 coronary artery bypass procedures per million population.