Prospects for precision genetic medicine in congenital heart disease

Abstract

Precision medicine, defined as tailoring medical care individually based upon relevant factors, is primarily implemented currently through the use of genetic variation. Over the past thirty years, the possibility of determining specific genetic variants underlying congenital heart disease has increased dramatically. This has created the potential for using precision genetic approaches to improve care and outcomes for patients and families with congenital heart disease. In this review, recent advances in understanding the roles of genetic variants in various outcomes, in developing novel therapeutic approaches, and in refining clinical trials for congenital heart disease are discussed.

Introduction

The promise of precision medicine is that health outcomes can be maximized through tailoring of the approach to individual patients based on the relevant pathogenic factors [1]. While conceptualized as a multilayered approach that includes environment factors and social determinants of health, much of the focus to date is on the use of genetic and genomic data. For congenital heart disease (CHD), there has been an acceleration in the identification of the relevant causal genetic variation, driven by rapid advances in genomic technologies as well as the assembly of large CHD cohorts to support genetic discovery [2–7]. Clinical genetic testing is increasingly being performed for patients with CHD, and rational approaches are evolving for various clinical scenarios [8•,9]. Armed with more genetic information about their patients, care providers are now grappling with how to implement precision genomic medicine for patients and families with CHD. In this review, the opportunities and obstacles to that implementation will be discussed. An overview of the approach vis-à-vis CHD is shown in Figure 1.

Figure 1.

Overview of the precision-medicine paradigm for congenital heart disease.

Precision genetics for outcomes in congenital heart disease

Previously, exome sequencing studies from the National Heart, Lung, and Blood Institute (NHLBI)-funded Pediatric Cardiac Genomics Consortium (PCGC) have established that damaging de novo variants underlying CHD are associated with extracardiac anomalies (ECAs) and neurodevelopmental delays and disabilities (NDD) [2–4]. In a new study, Boskovski and colleagues examined the role of clinically significant de novo variants (single nucleotide, small insertions/deletions, and copy number variants (CNVs)) on surgical outcomes [10••]. They found that these de novo variants were associated with worse transplant-free survival, even after adjusting for known risk factors, including age, Society of Thoracic Surgeons– European Association for Cardio-Thoracic Surgery (STAT) mortality category, and center (adjusted hazard ratio (HR) 3.51, 95% CI 1.96–6.06). As patients with these genetic variants are more likely to have ECAs, the authors examined the interactions between genetic variation, ECAs, and survival. They observed that the de novo variants conferred no additional risk for transplant-free survival for patients with ECAs, but did for those without ECAs (HR 11.21, 95% CI 4.12–29.13). Similarly, the clinically significant genetic variants were associated with a longer time to extubation postoperatively (adjusted HR 0.74, 95% CI 0.60–0.91). Thus, this study showed that clinically relevant de novo genetic variants affect perioperative outcomes and longer-term survival, mediated both directly and through ECAs, which also negatively affect them.

With improved survival of individuals born with CHD, less favorable outcomes, particular for neurodevelopment, have emerged as a critical issue. Increasingly, it has become clear that the genetic determinants of CHD can directly mediate NDD. In a 20-year retrospective cohort study from Wales, Dowden and colleagues examined a country-wide registry for CNVs that are associated with CHD [11••]. Unsurprisingly, 22q11del was the most prevalent followed closely by 15q11–q13del and then 16p11.2del and 1q21.1dup. The overall male:female ratio was 1.4 with heterogeneity among the pathogenic CNVs. Among the children with CHD and a pathogenic CNV, 37% had NDD, which is increased over expectation based on the CHD literature (Figure 2a). The authors then examined the patterns of CHD and NDD for the four pathogenic CNVs with the highest correlation of CHD and NDD (Figure 2b). They found that the distributions of subclinical NDD phenotypes varied among those four pathogenic CNVs, although global developmental delays were the commonest in all four. They also observed that the distribution of CHD types varied among those four CNV cohorts (not shown, see Figure 4b in Ref [11••]).

Figure 2.

(a) Venn diagram illustrating the overlap in the number of subjects harboring a pathogenic congenital heart disease-associated copy number variant who have CHD and/or NDD across the cohort, including a separate indicator of those patients with neither CHD nor NDD. (b) Distribution of the clinical phenotypes of NDDS across four copy number variants showing the highest comorbidity of NDD and CHDs.

Reprinted from [11••].

Another recent study applying a precision genetic approach to CHD outcomes, from Morton and colleagues, examined the role of genetic variation in cancer risk among CHD survivors [12••]. This work was premised on the observation that adults born with CHD have 1.4–2-fold increase in cancer prevalence relative to the general population. This risk has largely been attributed to this group’s increased exposure to radiation during medical care (e.g., cardiac catheterizations). Using exome sequencing data from the PCGC’s study CHD GENES and controls, they examined the rates of rare heterozygous loss-of-function changes in 723 cancer-risk genes, of which 38 are also known to cause CHD. They observed a highly significantly increased risk of harboring a cancer-risk variant among the CHD cohort (odds ratio 1.3, 95% confidence interval: 1.2–1.5). This increased rate among CHD was comparable in younger and older CHD subjects and was robust to ancestry. Analysis of the subset of 38 cancer-risk genes that also cause CHD revealed an even high risk (odds ratio 7.2, 95% CI, 4.2–12.2). The results for these and other gene subsets are shown in Table 1. While this study did not examine rates of cancer per se (the CHD cohort was principally pediatric aged), it provides interesting proof of principle for using a precision genetic approach to assess cancer risk among CHD survivors.

Table 1.

Patients with congenital heart disease and rare loss-of-function variants in subsets of the Catalogue of Somatic Mutations in Cancer–Cancer Gene Consensus cancer-risk genes.

Genes	No.	Patients with CHD	Control participants	Odds ratio (95% CI)	Binomial P valuea
Total participants	14,251	4443	9808	NA	NA
All cancer risk	723	642	1099	1.34 (1.21–1.49)	2.3×10−11
OMIM CHDb	38	68	18	7.19 (4.23–12.22)	< 2.2×10−16
Regulatory	216	143	166	1.93 (1.54–2.42)	1.4×10−12
Regulatory and OMIM CHDb	17	40	9	9.89 (4.80–20.40)	< 2.2×10−16
LoF cancer mechanism	227	240	376	1.43 (1.21–1.69)	1.6×10−7
Recessive LoF	135	158	274	1.28 (1.05–1.57)	1.7×10−7
Dominant LoF	46	53	50	2.36 (1.60–3.47)	3.5×10−8

Abbreviations: NA, not applicable; LoF, loss of function; OMIM, Online Mendelian Inheritance in Man.

^aBonferroni significance P-value threshold: 1.7×10⁻³.

^bThe OMIM CHD genes with dominant patterns of transmission. Adapted from [10••].

A recent elegant study of the genetics of hypoplastic left heart syndrome (HLHS) from Krane and colleagues suggests that a precision genetics approach might be helpful in predicting which fetuses with incipient HLHS would or would not benefit from in utero intervention with balloon aortic valvuloplasty [13••]. First, these authors performed exome sequencing on 87 patients with HLHS and their parents. They took de novo missense variants predicted to be damaging and undertook gene-set enrichment in gene categories related to cell-fate commitment and differentiation as well as cell-cycle biology. Next, Krane and colleagues compared nuclear RNA sequencing of right ventricular cardiomyocytes from healthy hearts at three different developmental stages to those from HLHS hearts, discarded tissues procured during operations in infants. They observed increased tetraploid nuclei in HLHS, consistent with premature cell-cycle withdrawal. Analysis of differently expressed genes pointed to cell-cycle biology and overlapped with the gene modules implicated with de novo mutations. Finally, Krane and colleagues compared induced pluripotent stem cells (iPSCs) derived from three patients with damaging de novo missense variants in genes or gene families implicated from the sequenced cohort and compared those to iPSC derived from healthy controls. The iPSCs were differentiated to cardiomyocyte progenitors and then cardiomyocytes. Analysis of bulk RNA sequencing once again implicated perturbations in cell-cycle biology. Taken together, these complicated studies implicated fundamental derangements in cardiac precursor cell and cardiomyocyte specification and lineage commitment, arising from specific genetic defects (scheme depicted in Figure 3). However, it is clear that this applies to a subset of patients with HLHS. The authors suggested that the intrinsic defects in cardiomyocyte replication and increased apoptosis may doom intrauterine aortic valvuloplasty for some fetuses with incipient HLHS. This implies that a precision-medicine approach might be applied to inform case selection for such risky interventions. Another path forward, also noted by the authors, is the development of mechanistic therapies to overcome the perturbations in cell-cycle biology. Of note, the broad approach to deeply phenotype forms of CHD modeled in this study could readily be applied to other forms of CHD, potentially providing insights into their pathogenesis and some complex outcomes such as heart failure.

Figure 3.

Scheme depicting the identified steps during cardiac development at which HLHS- related abnormalities interfere with normal development and contribute to the complex congenital heart disease phenotype. CM indicates cardiomyocyte; CP, cardiac progenitor; FHF, first heart field; PSC, pluripotent stem cell; SHF, second heart field; UPR, unfolded protein response.

Reprinted from [13••].

Mechanistic treatment of congenital heart disease

Since the very beginning of therapeutic interventions for CHD such as the Blalock–Taussig–Thomas shunt for certain forms of CHD with cyanosis, the entire enterprise has been premised upon understanding the hemodynamic derangements. Why a patient is born with CHD makes very little or no difference in the therapeutic approach. Because discoveries of causal genetic variation underlying CHD expose the pathogenesis of the CHD, this hypothetically provides opportunities to develop mechanistic therapies that might improve outcomes or obviate certain invasive interventions.

To date, there is no proven mechanistic therapy for CHD, but there is a tantalizing example of this approach. Andelfinger and colleagues described treatment of two critically ill young infants with Noonan syndrome due to RIT1 gain-of-function alleles who had severe hypertrophic cardiomyopathy [14]. As RIT1 and the other proteins altered in Noonan syndrome contribute to signal transduction through the RAS/mitogen-activated protein kinase (MAPK) pathway, reduction of signaling using an inhibitor of the MAPK kinases, MEK1 and MEK2, seemed rational. Indeed, preclinical data with a mouse model provided proof of principal. Andelfinger and colleagues treated the two babies with the MEK inhibitor trametinib with notable reversal of the hypertrophic cardiomyopathy. Relevant for this review is the fact that both patients also had pulmonary valve stenosis with valves that were characterized as dysplastic. In both, the valve gradient was eliminated after several months of treatment and the appearance of the valve leaflets normalized (G. Andelfinger, personal communication). While much work is needed to assess whether the side-effect profile is acceptable and to demonstrate treatment efficacy for Noonan syndrome-related pulmonary valve stenosis, this limited experience provides proof of principal. Of note, this would be precision medicine as there would be no a priori reason to believe that MEK inhibition would work for all patients with pulmonic stenosis, rather, it would be genotype-specific.

A related example, albeit lacking any therapeutic evidence so far, is the progressive valve stenosis observed among patients with the commonest form of CHD, bicuspid aortic valve (BAV). As discovered by Vidu Garg and Deepak Srivastava, a small percentage of BAV is caused by pathogenic variants in NOTCH1 and is associated with premature valve calcification [15]. Subsequent work showed that NOTCH1 signaling normally represses osteoblast-like calcification pathways mediated by BMP2 [16], leading to the notion that restoration of NOTCH signaling in patients with BAV attributable to altered NOTCH1 biology might prevent or slow valve calcification. Given that BAV per se is often asymptomatic, such therapy might suffice for many affected individuals.

These two examples both related to valve disease. Could mechanistic therapy be useful in CHD more broadly? Particularly postnatally, certain morphologic abnormalities are almost certainly not amenable (e.g., abnormalities of ventricular looping, conotruncal septation, and positioning), but, for instance, recurrence of subaortic stenosis or promoting septal-defect closure might be possible to treat.

Precision genetics for congenital heart-disease clinical trials

Designers of clinical trials for CHD already use some genetic information when developing inclusion and exclusion criteria. Particularly for trials for which neurodevelopmental outcomes or growth will serve as primary or secondary endpoints, patients with certain genetic traits such as Down syndrome, 22q11 deletion disorders, and Noonan syndrome are excluded in order to avoid confounding. Alternatively, study designs could be used in which individuals genetically at risk for outcomes relevant to their endpoints can be stratified with the treatment arms.

CHD genetic discoveries of the last few years have revealed that causal genetic variation, both single-nucleotide changes and larger genomic events (copy number variation), are more frequently present as nonsyndromic CHD and are not associated with identifiable facial dysmorphia. This is all the more the case for infants, for whom NDD associated with certain classes of causal genetic variation are not yet apparent. Thus, genetic testing of infants with CHD is necessary to determine who is at risk for NDD or poor growth.

The NHLBI-supported Pediatric Heart Network recently launched randomized controlled trial (RCT) called TEAM 4 Growth (Training in Exercise Activities and Motion for Growth, NCT04702373), which innovatively will incorporate a precision genetics approach. This RCT will evaluate growth in infants born with hypoplastic left heart syndrome or other related forms of CHD with single right ventricles after the Norwood procedure, comparing standard of care to an intervention with passive range-of-motion exercise program. Secondary endpoints will examine the effects of this intervention on neurodevelopment and on bone mineral density. To avoid the potential confounding of genetic factors as described above, all subjects will have re-sequencing of a panel of CHD genes to find pathogenic single-nucleotide variants and small insertions/deletions, as well as array genotyping to identify pathogenic copy number variants, being done through a collaboration with the NHLBI-funded Pediatric Cardiac Genomics Consortium. Enrollment for TEAM 4 Growth will exceed the desired final numbers to account for the need to remove genotype+ subjects after the genetic analyses. Taken as a whole, TEAM 4 Growth will test the utility of using precision genetics for CHD-related RCTs.

Conclusions

Over the past thirty years, our understanding of the nature of genetic variation contributing to the pathogenesis of CHD has expanded dramatically. The earliest instances were generally those easiest to recognize: aneuploidies (e.g., Down and Ullrich–Turner syndrome), what-were-then-called microdeletions (e.g., DiGeorge/velocardiofacial and Williams syndromes), and multiple organ-malformation traits. More recently, we have come to realize that most causal variation can be found only when genetic testing is performed. Indeed, genetic testing is increasing showing efficacy in the setting of prenatal CHD, when complete phenotyping is even more challenging [17–19•]. As genomic technologies have continued to improve and become less costly, the salient issue is changing: how can we most precisely use genetic information about CHD to improve care? Aside from informing reproductive counseling for affected individuals and immediate relatives, it is becoming clear that a precision-medicine approach can inform our expectation for the outcomes of individual patients. Excitingly, genetic findings are suggesting novel therapeutic approaches for certain aspects of CHD. Finally, incorporation of genetic information into RCTs directed at improving outcomes for CHD may eliminate an important source of confounding, thereby allowing us to more accurately assess therapeutic interventions directed toward outcomes such as neurodevelopment and growth.