Lack of Evidence for the Role of the p.(Ser96Ala) Polymorphism in Histidine-Rich Calcium Binding Protein as a Secondary Hit in Cardiomyopathies

Inherited forms of arrhythmogenic and dilated cardiomyopathy (ACM and DCM) are characterized by variable disease expression and age-related penetrance. Calcium (Ca2+) is crucially important for proper cardiac function, and dysregulation of Ca2+ homeostasis seems to underly cardiomyopathy etiology. A polymorphism, c.286T>G p.(Ser96Ala), in the gene encoding the histidine-rich Ca2+ binding (HRC) protein, relevant for sarcoplasmic reticulum Ca2+ cycling, has previously been associated with a marked increased risk of life-threatening arrhythmias among idiopathic DCM patients. Following this finding, we investigated whether p.(Ser96Ala) affects major cardiac disease manifestations in carriers of the phospholamban (PLN) c.40_42delAGA; p.(Arg14del) pathogenic variant (cohort 1); patients diagnosed with, or predisposed to, ACM (cohort 2); and DCM patients (cohort 3). We found that the allele frequency of the p.(Ser96Ala) polymorphism was similar across the general European–American population (control cohort, 40.3–42.2%) and the different cardiomyopathy cohorts (cohorts 1–3, 40.9–43.9%). Furthermore, the p.(Ser96Ala) polymorphism was not associated with life-threatening arrhythmias or heart failure-related events across various patient cohorts. We therefore conclude that there is a lack of evidence supporting the important role of the HRC p.(Ser96Ala) polymorphism as a modifier in cardiomyopathy, refuting previous findings. Further research is required to identify bona fide genomic predictors for the stratification of cardiomyopathy patients and their risk for life-threatening outcomes.

Disease onset and severity are highly variable.Even within families carrying the same genetic variant, variable disease expression and age-related incomplete penetrance are frequently observed [1].Therefore, additional genetic and non-genetic factors (secondary hits) are believed to contribute to disease development and progression.For example, intensive exercise might increase disease penetrance and enhance the risk for life-threatening VA in ACM patients [11].Similarly, the presence of multiple (likely) pathogenic variants in ACM-associated genes may increase disease severity [12].Nevertheless, the prediction of severe outcomes among ACM and DCM patients remains exceedingly difficult, necessitating the identification of novel and accurate markers for risk stratification.
Recent and significant focus has been on the disturbance of calcium (Ca 2+ ) homeostasis within cardiomyocytes with regard to cardiomyopathy etiology.Ca 2+ plays a crucial role in the excitability and contraction of the heart.In failing cardiomyocytes, disturbed Ca 2+ handling contributes to HF and life-threatening VA [13].Therefore, dysfunction of factors that regulate Ca 2+ homeostasis, such as the histidine-rich calcium-binding protein (HRC), may contribute to the development and progression of inherited cardiomyopathies [14].HRC can interact with triadin when [Ca 2+ ] levels increase, thereby regulating RyR2 and modulating SR Ca 2+ release.When [Ca 2+ ] is low in the SR, HRC binds to the sarcoplasmic/endoplasmic reticulum Ca 2+ -ATPase (SERCA2a), which suppresses SERCA2a function [14,15].The serine residue at position 96 is a phosphorylation site of potential importance to triadin affinity (Figure 1).In one candidate gene study, a polymorphism in the HRC gene, c.286T>G p.(Ser96Ala)), was found to confer a four-fold higher risk of life-threatening VA and SCD in a small cohort of 123 patients with idiopathic DCM [16].
To follow up on this observation, the aim of our study was to examine (1) the presence of the HRC p.(Ser96Ala) polymorphism in the general population (control cohort); (2) the effect of the HRC p.(Ser96Ala) polymorphism on major cardiac events in three patient cohorts, namely carriers of the PLN p.(Arg14del) pathogenic variant (cohort 1); patients diagnosed with, or predisposed to, ACM (cohort 2); and DCM patients (United Kingdom (UK) Biobank, cohort 3).We hypothesized, based on the earlier observation, that the HRC p.(Ser96Ala) polymorphism may contribute to the risk of major disease manifestation, which could potentially improve risk stratification.] rises, HRC is phosphorylated by Fam20C at Ser96 position, which induces interaction with triadin to modulate RyR2 function.When the serine residue is altered for an alanine residue (Ala96Ala variant), phosphorylation is not performed, and therefore affinity for SERCA2a and triadin remains unaffected.HRC; histidine-rich calcium-binding protein, SR; sarcoplasmic reticulum, Ca 2+ ; calcium, SERCA; sarcoplasmic/endoplasmic reticulum Ca 2+ -ATPase, PLN; phospholamban, RyR2; ryanodine receptor 2, Ser; serine, Ala; alanine, Fam20C; family with sequence similarity 20C.

Patient Characteristics
In cohort 1, 1005 carriers of the PLN p.(Arg14del) variant passed quality control (QC) and were included.Due to incomplete or unavailable health record data, 157 patients were excluded, leaving 848 patients for data analysis.In cohort 2, among 1033 ACM patients of European (-American) ancestry, 882 patients with complete health record data were included for further analysis.Finally, 1031 DCM patients of European ancestry were found in cohort 3, of which 985 individuals passed QC and were used for downstream analyses (Figure 2).
Finally, in cohort 3 (DCM, Table 3), the mean age at enrollment was 61 years for all three polymorphism groups.Compared to cohorts 1 and 2, predominantly males were included in the analysis: 71% in WT, 68% for heterozygous, and 79% for homozygous carriers.The mean follow-up time of all three HRC groups was approximately 10 years.No statistical differences were found between the groups.

HRC p.(Ser96Ala) Polymorphism in General and Cardiomyopathy Populations
The minor allele frequency (MAF) of the HRC p.(Ser96Ala) polymorphism in the control cohorts ranged from 40.3 to 42.2%, as analyzed in different databases.
In the three patient cohorts, the MAF of the HRC p.(Ser96Ala) polymorphism was 40.9% among PLN p.(Arg14del) carriers (cohort 1), 43.9% for the overall ACM cohort (cohort 2), and 41.3% for the UK Biobank DCM cohort (cohort 3).In more detail, the Dutch ACM cohort (n = 491) showed a MAF of 43.6%, while a MAF of 43.0% was found among US ACM patients (n = 337) and 50.9% among the Swiss ACM cohort (n = 54).Data regarding the different observed frequencies are summarized in Table 4.

Heart Failure and HRC Polymorphism
Among PLN p.(Arg14del) carriers from cohort 1, HF was observed in 42/257 (16%) WT HRC carriers, 53/345 (15%) that were heterozygous, and 15/113 (13%) p.(Ser96Ala) homozygotes (Table 1).No significant association was found between the HRC polymorphism and the HF outcome using logistic regression (OR (95% CI) 0.858 (0.615-1.188), p = 0.360; Table 5).The prevalence and severity of progressive HF in ACM might be low given a currently debated definition of HF among these cases, particularly in the right dominant subforms of the disease [17].In addition, overlapping criteria for congestive HF and DCM are frequently used in the research literature (i.e., LVEF < 45%).For these two reasons, no analyses for the risk of HF outcomes were performed for cohorts 2 and 3.

Discussion
In this study, we examined the role of the previously identified HRC p.(Ser96Ala) polymorphism in relation to major cardiac events in clinical cohorts diagnosed with, or predisposed to, different forms of cardiomyopathy.Most carriers develop symptoms between their third or fifth decade of life, suggesting an incomplete penetrance.In addition, a wide range of symptoms among patients carrying the same genetic variant can be found [18].Therefore, improvement in proper risk stratification is needed in these patients.We established that the HRC p.(Ser96Ala) polymorphism is frequent (40-42%) in the general population, with highly similar frequencies among the different studied patient cohorts.Importantly, we found that the p.(Ser96Ala) polymorphism did not significantly contribute to the risk of major cardiac events among PLN p.(Arg14del) carriers, ACM patients, or DCM patients, contrasting with previous reports.
Recently, studies have started to explore the role of common genetic variants in modifying the risk of cardiomyopathies.Genome-wide association studies for DCM have identified several common genetic variants associated with disease risk [19].Notably, common genetic variants have been shown to modify the penetrance and expressivity of HCM in patients with rare variants [20].Given the marked incomplete penetrance and variable expressivity of rare variants-and lack of known pathogenic variants in many index patients with ACM and DCM-common variants may similarly contribute to expressivity in these cardiomyopathies.This is especially relevant as the prediction of major adverse events, including HF and VA, remains exceedingly challenging.
In the present study, we assessed the risk of HF and VA among a broad range of patient cohorts, namely, PLN p.Arg14del carriers, ACM index patients and family members, and DCM patients.This approach was chosen given that the phenotypic outcomes of these entities share overlapping features, among which there is a significant risk of life-threatening VA.Classically, ACM was described as arrhythmogenic right ventricular cardiomyopathy (ARVC) with sole involvement of the right ventricle; however, biventricular and left dominant forms are now increasingly recognized [5].Conversely, a decline in right ventricular function is a predictor of worse outcomes in DCM patients [21].Carriers of PLN p.(Arg14del) can be diagnosed with either of these cardiomyopathies, reflecting the clinical and genetic overlap between these disease entities [10].Finally, in both ACM and DCM, disturbance of Ca 2+ handling is known to drive the development of life-threatening arrhythmias and HF [22,23].
Given this apparent functional mechanism, the p.(Ser96Ala) polymorphism could potentially influence Ca 2+ regulation and thereby contribute to VA or deterioration of Ca 2+ handling.Indeed, a marked effect was described in an initial study of 123 idiopathic DCM patients-where p.(Ser96Ala) homozygotes had an over four-fold risk of major arrhythmic events [16].In contrast, we were unable to confirm a role for the HRC p.(Ser96Ala) polymorphism in stratifying the risk of major events among three larger cohorts of patients with (predisposition to) cardiomyopathy.Furthermore, directionally inconsistent results were found in our study compared to the study of Arvanitis et al., as the p.(Ser96Ala) polymorphism seemed to be protective for major cardiac events [16].We note that several of our cohorts included individuals with preclinical disease or genetic predisposition only, which contrasts with the established DCM patients analyzed in the work of Arvanitis et al.; however, we performed several sensitivity analyses, including index patients only, which generally showed consistent effects.In addition, the one significant association between the index and life-threatening VA, even if not significant after correction for multiple tests, was directionally inconsistent compared to the aforementioned study.Furthermore, it is possible that our study remains underpowered to detect a small effect of p.(Ser96Ala) on major events.Nevertheless, the 95% CI of our estimates-in all cohorts and analyses-excludes a large effect, especially one as large as described by Arvanitis et al.

Study Design and Patient Selection
In this study, individuals from four different patient and control cohorts were included: Cohort 1 included individuals of Dutch ancestry enrolled in the Dutch ACM registry in whom the pathogenic p.(Arg14del) variant in PLN was identified (accessed on 7 March 2023).Within this database, we selected PLN p.(Arg14del) carriers, and this will be referred to as the PLN registry.This registry includes data from individuals obtained from three Dutch university medical centers (Groningen, Amsterdam, and Utrecht) [24].The study cohort consisted of both index patients (i.e., first affected family member tested positive for the disease-associated genetic variant, mostly because of suspected inherited cardiac disease) and relatives after genetic cascade testing.
Cohort 2 included individuals enrolled in the ACM registries from the Netherlands [24], Switzerland, and The United States (USA, data accessed on 8 March 2023).Index and relatives with a definitive ARVC diagnosis based on the 2010 modified Task Force Criteria (TFC) and relatives with a pathogenic variant in an ARVC-related gene (who do not meet TFC) or gene elusive were included.
Cohort 3 included DCM patients identified within the UK Biobank cohort.Patients with a clinical DCM diagnosis were identified using International Classification of Diseases (ICD-10 code I42.0) [25].
Control cohort included individuals that belong to the "general population".Population data consisted of three publicly available databases: (1) Genome Aggregation Database (gnomAD v2.1.1,https://gnomad.broadinstitute.org/,accessed on 23 March 2023), (2) 998 individuals from The Genome of the Netherlands (GoNL, accessed on 25 May 2023) [26], and (3) participants of the UK Biobank (accessed on 26 April 2023).The UK Biobank is a large population-based prospective study that included 500,000 UK participants between 40 and 69 years [27].Additionally, 31,400 individuals who were referred to the Department of Genetics of the UMC Utrecht, the Netherlands (inclusion between 2017 and 2023, data accessed on 5 May 2023) were evaluated irrespective of diagnosis.In this group, a whole exome sequencing (WES)-based genetic test was available as part of regular clinical care.
For all cohorts, only individuals with European (-American) ancestry were selected for further analysis.Written informed consent was provided by all participating patients.Furthermore, all data were fully anonymized before data could be accessed.This study was conducted according to the Declaration of Helsinki.Study protocol was approved by UMC Utrecht (Biobank number .Use of UK Biobank data was performed under application number 17,488 and was approved by the local Massachusetts General Hospital Institutional Review Board.

Genetic Data Extraction
Cohort 1 and 2: DNA samples of PLN p.(Arg14del) carriers, ACM patients, and preclinical gene variant carriers enrolled in the ACM registry were genotyped on the Illumina Global Screening Array (-24 v3.0 BeadChip, San Diego, CA, USA) at the Human Genomics Facility (HuGe-F), Erasmus Medical Center, Rotterdam, the Netherlands and at the Genetic Resources Core Facility (GRCF), at John Hopkins University, Baltimore, the USA.The p.(Ser96Ala) polymorphism (T/G) (rs3745297) in HRC was extracted on imputed data after standard sample QC (INFO score > 0.99; removal of samples was performed for those who revoked consent, had a mismatch between genetically predicted and selfreported sex, were outliers for heterozygosity or missingness, had putative sex chromosome aneuploidy, or were ancestral outliers in a Principal Component (PC) Analysis) using Plink v1.9 and v2.0.The HRC polymorphism was categorized as either WT (TT), heterozygous (TG), or homozygous (GG).
Cohort 3: UK Biobank participants underwent dense genotyping using the UK Biobank Axiom Array or the Affymetrix UK BiLEVE Axiom Array (Thermo Fisher Scientific, Waltham, MA, USA), as described by Bycroft et al. [28].We used the version 3 imputed data, where the p.(Ser96Ala) polymorphism attained near-perfect imputation accuracy (INFO > 0.99).Standard sample QC was performed as described before.
Control cohort: the HRC p.(Ser96Ala) polymorphism was extracted from gnomAD (WES data used) [29]; GoNL (whole-genome sequencing (WGS) data used) [26]; the UK Biobank (version 3 imputed data; INFO score > 0.99; sample QC described above) [28]; and in patients in whom WES was performed as part of regular clinical care, irrespective of diagnosis, in UMC Utrecht, Utrecht, The Netherlands between 2017 and 2023.

Clinical Outcomes
The first primary outcome was defined as the first occurrence of a life-threatening VA event, including sustained VT or ventricular fibrillation (VF), appropriate ICD therapy, and (aborted) SCD.The second primary outcome was defined as the first occurrence of an HF-related event.This included hospitalization for HF-related complaints, implantation of a left ventricular assistance device (LVAD), heart transplantation, or HF-related death.A secondary outcome was a composite of both life-threatening VA and/or HF-related events.For cohorts 1-2, clinical information from the first cardiac evaluation and follow-up visits was retrospectively extracted from the ACM and PLN registries, as described before [24].Age at enrollment in the registry was defined by presenting with clinical presentation due to cardiomyopathy-related symptoms, SCD, or family screening [24].
For the UK Biobank DCM cohort (cohort 3), participants attended an entry assessment at centers across the UK to provide baseline characteristics.Follow-up data were collected via hospital event data and death registry linkage; main outcomes (VT, SCD, and ICD implantation) were defined using ICD codes, as previously described [25].The latest update of health record linkage data was performed in October 2020.Outcomes and covariates of interest were collected and presented as descriptive statistics for each registry cohort; categorical variables were analyzed using Chi-Square tests, and continuous variables were analyzed using Kruskal-Wallis tests.In each dataset, logistic regression was performed with life-threatening VA, HF, or both (composite outcome) modeled as outcome; the HRC p.(Ser96Ala) was modeled as a predictor, adjusting for covariates of sex, age at enrollment (in the registry cohorts), and the first 12 PCs of ancestry.Logistic regression analyses were repeated within subgroups, i.e., restricting to index cases or relatives.Additionally, we performed analyses using only incident events modeled in a Cox proportional hazard regression (time-to-event).Cox models were adjusted for the same covariates as in the logistic regression analysis and right-censored at last follow-up at the cardiology outpatient clinic or death if the primary outcome had not occurred.Start of incident time was modeled in two ways, namely, (1) from birth (which effectively is true since the genetic variant is already present at birth) and ( 2) from enrollment in the ACM or PLN registry.In addition to the aforementioned outcomes, mortality was also assessed in time-to-event analyses.Differences were considered statistically significant when p < 0.05.Statistical analyses were performed with R version 4.0.3, using the "survival" and "survminer" packages [30,31].

UK Biobank DCM Cohort
In the UK Biobank DCM cohort, logistic regression was used, pooling incident and prevalent events to maximize case numbers and power.Outcomes included VA, SCD, and ICD implantation.Models were adjusted for age, sex, genotyping array, and the first 12 PCs of ancestry.Because logistic regression may not correctly model temporality and competing risks, we also performed analyses using only incident events modeled in a Cox proportional hazard regression, using function coxph() from R package "survival" [30].Cox models were adjusting for the same covariates and right-censored at death to account for competing risks of death.Start of incident time was modeled in two ways, namely, (1) from enrollment in UK Biobank and (2) from DCM diagnosis.In addition to the aforementioned outcomes, mortality was also assessed in time-to-event analyses.Differences were considered statistically significant when p < 0.05.

Conclusions
In this study, we found that the p.(Ser96Ala) polymorphism is common in the general population and cardiomyopathy-affected patients.Furthermore, there is a lack of evidence for the role of the p.(Ser96Ala) HRC polymorphism in modifying the risk of major cardiac events among cardiomyopathy patients.Our data indicate that any possible effect is, at most, small, limiting its use as a sole predictor of severity among ACM and DCM patients.Further research is required to identify bona fide predictors for stratification of cardiomyopathy patients and their risk for life-threatening outcomes.

Limitations
Despite the unique (and rare) composition of our included cohorts, this study could be subjected to several limitations.A larger part of the PLN p.(Arg14del) pathogenic variant carriers and ACM patients are classified as still being in the preclinical phase.Therefore, clinical events might be low compared to a cohort that includes only diagnosed patients, such as the DCM patients from the UK Biobank.In addition, lifestyle factors such as diet, environment, and exercise might influence and accelerate disease manifestation and events.However, these data were not available when we performed our analyses.In addition, some clinical, MRI, or echocardiographic parameters have not been reliably collected (e.g., medication use) or were not available.However, we have the impression that the unavailability of these data does not hamper the major conclusions and genetic observations of this study.When estimating a population MAF, most of the studies make use of a controlled setting in which cases and controls are matched.However, whether this group represents the population frequencies might not be taken into account [32].Therefore, extrapolation of the observed allele frequencies to the general population can be inadequate.Finally, we note that our study was largely focused on individuals of European ancestry, potentially limiting the generalizability to other ancestry groups.As illustrated in a single study on Japanese paroxysmal atrial fibrillation patients, a MAF of 26% was reported, which was similar to population data of East Asian ancestry found in gnomAD [33].

Table 4 .
Minor allele frequencies of the HRC polymorphism in general and registry cohorts.

Table 5 .
Logistic regression analyses in PLN p.(Arg14del) carriers were modeled with HRC p.(Ser96Ala) as predictor.Analyses were adjusted for sex, age at enrollment, and principal components.

Table 6 .
Logistic regression analysis in ACM patients was modeled with HRC p.(Ser96Ala) as predictor.Analyses were adjusted for sex, age at enrollment, and principal components.

Table 7 .
Logistic regression analyses in DCM patients were modeled with HRC p.(Ser96Ala) as predictor.Analyses were adjusted for sex, age at enrollment, genotype array, and principal components.