Clinical Spectrum Of CACNA1C Variants, Revisited

CACNA1C is a gene encoding the CaV1.2 calcium channel and several cardiac conditions are potentially associated with pathogenic variants in this gene. The aim of this study is to explore genotype-phenotype correlations related to CACNA1C ever described variants and vast phenotypic spectrum. We analyzed 102 patients with CACNA1C variants (CACNA1Cv) (9 our cohort and 93 from literature). We studied the association between CACNA1Cv and clinical parameters: arrhythmias, structural heart defects, cardiomyopathy and survival. We followed the American College Medical Genetics (ACMG) scoring system to grade variants’ pathogenicity and their domains.


Abstract
Background CACNA1C is a gene encoding the CaV1.2 calcium channel and several cardiac conditions are potentially associated with pathogenic variants in this gene. The aim of this study is to explore genotype-phenotype correlations related to CACNA1C ever described variants and vast phenotypic spectrum.

Methods
We analyzed 102 patients with CACNA1C variants (CACNA1Cv) (9 our cohort and 93 from literature). We studied the association between CACNA1Cv and clinical parameters: arrhythmias, structural heart defects, cardiomyopathy and survival. We followed the American College Medical Genetics (ACMG) scoring system to grade variants' pathogenicity and their domains.

Results CACNA1Cv
with high ACMG scores were associated with higher mortality than variants with lower scores (p < 0.001). CACNA1Cv in Cytoplasmic and Transmembrane domains were associated with higher mortality than other domains (p = 0.005). Multivariate analysis for higher ACMG scores, indicates cardiomyopathy and a lesser extent domain, as independent risk factor for mortality (p = 0.031 and p = 0.04). Cytoplasmic domain variants were frequently associated with long-QT syndrome; C-terminal variants were often linked to Brugada syndrome. Parental mosaicism was relatively high (4-5%) and must not be overlooked in parents' phenotypic analysis and in calculation of disease recurrence risk

Conclusion
To the best of our knowledge, this is the rst study trying to create genotype-phenotype correlation and better risk strati cation in CACNA1Cv in relation to survival and long-term results. Background CACNA1C is a gene encoding the CaV1.2 calcium channel, which is critical for the plateau phase of the cardiac action potential, cellular excitability, and excitation-contraction coupling 1,2,3,4,5 . Several conditions are potentially associated with pathogenic variants in this gene including electrical, myocardial and structural defects. Moreover, extracardiac involvement is described, as well as a multisystemic disorder (Timothy Syndrome) characterized by different degrees of cardiac involvement (arrhythmic and structural), ngers syndactyly, developmental delay and recurrent infections.
Abnormalities related to CACNA1C variants vary greatly and may include any of the followings: i. electrical defects such as long QT syndrome (LQTS), Brugada syndrome (BrS), short QT syndrome (SQTS) and sudden cardiac death (SCD); ii. myocardial abnormalities, including hypertrophic cardiomyopathy (HCM); iii. structural heart defects such as tetralogy of Fallot and ventricular septal defects 6,7,8,9 . For some of these disease entities a causal association has recently been questioned 10 .
It is di cult to estimate the phenotypic spectrum of CACNA1C variants due to its wide variability. Historically, two major variants were described (p.Gly406Arg and p.Gly402Ser); however, with the progressive use of next generation sequencing (NGS) more variants are described and genetic heterogeneity is increasing. Moreover, genotype-phenotype correlation is still undone. Previous studies have only discussed fragmented aspects of the phenotypic spectrum.

Methods
The aim of this study is to explore this relationship, based on our cohort and review of current evidence on known CACNA1C variants.

Bambino Gesù Children Hospital cohort (OPBG):
We reviewed medical records of patients referred to our tertiary care centre for molecular analysis of CACNA1C. All data, including the molecular and cardiac diagnosis, were extracted from our database, which includes records on electrocardiograms (ECG), Holter ECG monitoring, echocardiography, stress test results and angiography procedures. The study protocol conforms to ethical guidelines of the 1975 Declaration of Helsinki, as re ected in a priori approval by the Institution's Human Research Committee.
We performed, on peripheral blood DNA extracted from patients, molecular analysis of CACNA1C (mRNA: NM_000719) coding regions and anking splicing regions through high-throughput targeted resequencing using an Arrhythmia custom panel and successively analyzed with NextSeq550 platform (Illumina, San Diego, CA). This panel includes causative genes for the following syndromes: LQTS, BrS, Catecholaminergic Polymorphic Ventricular Tachycardia (CPVT) and SQTS. Variant analysis was performed with Illumina Variant Studio Software and Integrative Genome Viewer (IGV). Sanger sequencing con rmed all genetic variants detected in the index cases on re-extracted DNA, using standard protocols.

Literature Review
We searched PubMed for any published studies describing CACNA1C variant(s). Search strategy included "CACNA1C" and any of the following terms: "Timothy", "Brugada", "Long QT", "LQTS", "Cardiomyopathy" and "Rhythm". Inclusion criteria were: any article reporting data on CACNA1C and clinical presentation of the patient(s) involved; exclusion criteria were: duplicated cohorts and/or articles, molecular studies and articles written in language other than English. Full papers were carefully reviewed and considered according to the previous criteria. All articles were independently analyzed by three authors (BC, MG, FC).
In case of dispute, an agreement was negotiated among the three authors. References of selected papers were cross-checked with the same inclusion and exclusion criteria. Papers included at the end of the screening process were fully reviewed, and study characteristics were extracted and tabulated in an MS Excel for Windows (Microsoft Corporation, Redmond, Washington)-available upon request.

Variants classi cation
For each variant, we considered the ACMG (American College of Medical Genetics) classi cation to establish the pathogenicity of the molecular variants 10 . This classi cation allowed us to classify variants using speci c terminology identi ed in Mendelian disorders ('pathogenic' (ACMG score 5), 'likely pathogenic' (ACMG score 4), 'uncertain signi cance -VUS' (ACMG score 3), 'likely benign' (ACMG score 2), and 'benign' (ACMG score 1)). We excluded from the analysis patients with variants identi ed by the ACMG score as "benign" or "likely benign", distributed uniformly across all regions. We used this classi cation and stipulated it in further statistical analysis and in supplementary table 1A and 1B.

Clinical parameters
We analyzed different parameters for each patient (literature and OPBG cohort): family history, prenatal ndings, birth weight, gender, median age at diagnosis, and survival.
Electrocardiographic features have been investigated speci cally for: abnormalities of QT corrected interval (QTc), PR, QRS, ST, J wave and T wave, Brugada pattern, presence and type of atrioventricular block (AVB), sinus node dysfunction, atrial tachyarrhythmias, ventricular tachycardia/ brillation and torsade de pointes.
Standard Echocardiogram data were collected with a focus on ndings suggestive for hypertrophic cardiomyopathy (HCM) and ventricular septal defects.
We have also considered the presence of cyanosis, history of syncope or aborted SCD, hypotension, congenital heart defects and type of management (surgical, conservative or hemodynamic studies) including post-operative complications. The presence of dysmorphic features and syndactyly were evaluated. The analysis included any abnormal growth pattern, developmental delay, history of seizures, hypoglycemia, sepsis/recurrent infections and nally genetic investigation, with peculiar attention to CACNA1C variants.

Statistical Analysis
All statistical analysis was performed using SPSS Statistics 21 package (IBM Corporation, Armonk, NY, USA). Continuous variables are presented as mean values (standard deviation (SD) and range) or median values (interquartile range, IQR), as appropriate. Categorical variables are expressed as absolute numbers or percentages. Fisher's exact test was used to compare groups < 5. Survival data were analyzed and graphically reported by the Kaplan-Meier method. Log rank test was used for comparison between groups. Patients who did not experience an event (cardiac arrest, aborted cardiac arrest) were censored at the time of the last follow-up. Univariate analysis was performed by logistic regression analysis for each variable (Arrhythmic event, LQT, BrS, ICD/PMK, HCM, CHD and mortality). Variables with P values less than 0.2 in the univariate analysis were considered eligible for entering multivariable analysis. Multiple regression analysis was also performed. Results are expressed as odds ratios (OR) with 95% con dence intervals (CI). P-value was considered signi cant when ≤ 0.05.

OPBG Cohort
We identi ed 19 patients from our database (10 females and 9 males). Cardiac diagnosis varied from: LQTS (n = 13), BrS (n = 2), CPVT (n = 3), and asymptomatic (n = 1) teenager who is a rst-degree relative of an individual with SCD. The age at rst observation was < 18-years-old in 76% of patients. Eight (42%) patients showed a positive family history of LQTS or SCD. Supplementary material Table 1A includes list of CACNA1C variants and domain localization in our OPBG Cohort.
Although rare, the following variants (p.Val1707Ile, p.Ala1717Gly, p.Thr1870Met, p.Arg1973Gln) were previously reported with con icting interpretations of pathogenicity and potential benign effect: ClinVar ID 190622, 93411, 191567 and 93419 respectively 12,13,14 . For these reasons we decided to exclude them from statistical analysis in order to avoid bias in terms of correlation to morbidity and mortality. Thus, we excluded 10 patients with any of these variants from our analysis. This reduced our cohort to 9 patients.

Literature Review analysis
We identi ed 235 articles from the literature (Fig. 1). 185 articles were excluded according to the aforementioned criteria (1 article not in English language; 118 reporting basic science studies; 66 duplicated cohorts). Fourteen patients of these remaining 50 articles were further excluded because CACNA1C variant was mentioned but without any speci c molecular details: no nucleotide nor amino acid references (Table 1B in  5. Intramembrane, pore-forming; 6. C-terminal, cytoplasmic.
In supplementary materials table 2 we annotated amminoacidic division for each proteic domain.
For each domain group, we reported the main ndings including gender, family history, median age at diagnosis, QTc, survival, SCD, ICD, tachyarrhythmia, Brugada pattern, HCM, and CHD.
We summarized relevant clinical data related to CACNA1C variants in a bar graph included in Fig. 2.
Moreover a detailed analysis including full description of domain to domain analysis is included in supplementary le 1. The le includes relative references for this speci c section.

Outcomes
We performed univariate analysis for each outcome (Arrhythmic event, LQT, BrS, ICD/PMK, HCM, CHD and mortality), with N-terminal domain as reference ( We performed Kaplan Meier analysis for survival related to CACNA1C localization and depending on ACMG pathogenicity scoring ( Fig. 3a and 3b).
In order to explore potential domains at higher risk for mortality at follow-up, we performed another analysis (Fig. 3c) considering patients with variants in cytoplasmic and transmembrane domains versus others.
To explore the relationship between variants ACMG score, localization and HCM, we performed a Coxregression analysis. Domains were categorized in two main groups: Cytoplasmic and Transmembrane vs other domains. Results HCM and high score ACMG resulted as a predictor of mortality independently from variants domain localization.

Discussion
CACNA1C pore-forming, alpha-1C subunit of the voltage-gated calcium channel gives rise to L-type calcium currents mediates in ux of calcium ions into the cytoplasm, and thereby triggers calcium release from the sarcoplasm 1,2 . CaV1.2-the predominant L-type calcium channel in the cardio-vascular system and in the brain has an intermediate voltage-dependence of activation 3 . It is composed of four homologous but non-identical domains (repeats I, II, III, IV), each consisting of six membrane-spanning helices (S1-S6). Helices S1 through S4 of each repeat form voltage-sensing domains (VSD); helices S5, S6 and the connecting P-loop of all four repeats together form the channel pore 16 .
CACNA1C variants are relatively rare and maybe related to high clinical heterogeneity, not only in arrhythmic conditions but also in structural progressive myocardial disease and CHD. Previous studies are mainly divided into two groups: those discussing basic science studies and animal models (which were not the topic of this study) and those describing heterogeneous cohorts of patients. To the best of our knowledge, this is the rst comprehensive study which includes previously described variants and a cohort of 9 patients from our centre. Variants were classi ed according to their localization, in order to identify genotype-phenotype correlation and potential risk strati cation in relation to survival and longterm results in each category.
Different survival rates are identi ed according to ACMG variants pathogenicity score and variants localization, as explained by Kaplan-Meier curves. In fact, Fig. 3a considers the different variant localisation domains (N-terminal, Extracellular, Transmembrane, Cytoplasmic and C-terminal) with a pvalue of 0.028.
While Fig. 3b includes Kaplan Meier analysis for survival expressed in years depending on the degree of variant pathogenicity according to ACMG score (Log rank = 0.000). Pathogenic and likely pathogenic variants are shown in a single plot (ACMG score = 4-5), the second plot represents Variant of Uncertain Signi cance (VUS) with ACMG score = 3, and the last plot represents Not Applicable (NA) variants for the lack of data (ACMG score < 3). There is signi cance difference between the three plots with a log rank of 0.000. This means that validity of the ACMG score is supported by the present analysis, with the score of pathogenicity (ACMG score 4-5) that result associated to high mortality.
This study might help in determining the prognosis especially in terms of active management for early therapeutic intervention, close follow-up and potential bene t from early implantation of ICD/PMK.
The last Kaplan Meier curve (Fig. 3c)  that contains ve hydrophobic transmembrane segments (S1, S2, S3, S5, and S6) and one positively charged transmembrane segment (S4). S4 segments represent the voltage-sensor and are characterized by a series of positively charged amino acids at every third position. LQT electrophysiological phenotype is characterized by loss of current density and gain-of-function shift in activation leading to increased steady-state current 4 .
Brugada syndrome is signi cantly related to C-terminal domain (p = 0-002). Calcium binding region is localized in C-terminal domain. Variants might cause loss of a low-a nity interaction with CALM1 or loss of channel inactivation by Ca (2+) and calmodulin 17 .
Multivariate analysis has shown that HCM and ACMG variants score pathogenicity (4-5) are independent predictors of mortality. The p-value is 0.031 for HCM and 0.04 for variants pathogenicity (supplementary Table 3). There is no signi cance considering variants localisation domain. This suggestes that the differences among different regions in Figs. 3 and 5 is possibly due to a higher incidence of HCM in CACNA1C variants located in the transmembrane and cytosolic domains, as showed in Fig. 2. In order to exclude other known genetic factors, we analysed with exome sequencing our patient with the classic CACNA1C variant without identifying any pathogenic variants of genes related to HCM. This data strengthens the potential role of this gene on structural myocardial changes. A similar nding has been reported previously in literature 8 where authors have demonstrated that exome sequencing identi ed only one variant that was related to a complex phenotype in a family where different members showed LQTS, HCM and CHDs.
In order to give a comprehensive vision of genotype phenotype correlation, we summarized in Fig. 4 the topographic representation of different variants domains and associated major clinical features and outcome.
Another aspect that need to be considered is the mosaicism. Mosaicism refers to the co-existence, in an individual, of cells with different genotypes, although derived from a single zygote, so that some cells may present with the gene mutation (and the resulting loss/gain of function), and others do not 18 . The incidence of mosaicism in the human is underestimated, especially in the low-grade mosaicism. While somatic mosaicism has been implicated in over 30, monogenic disorders, mosaicism is rarely reported in LQTS 19 . In this review we identi ed ve reports 3,20,21,22,23 with parental mosaicism. It makes a relatively high percentage for such a small cohort, reaching 4-5% of the total analysed series. These observations have important consequences for genetic counselling, as previously identi ed de novo mutations may represent parental mosaicism. A shared partial phenotype should not be dismissed as a benign or insigni cant nding but should be evaluated further to rule out the possibility of parental mosaicism concealing a potentially fatal heritable disease.
Another previously and relatively widely reported aspect related to CACNA1C variants is the association with multisystemic disorder, well recognized as Timothy syndrome. It is characterized by cardiac involvement (LQTS, HCM, CHD), hand/foot variable syndactyly, facial dysmorphic features (depressed nasal bridge, low-set ears, thin vermilion border of the upper lip, and round face), and neurodevelopmental features including global developmental delays and autism spectrum disorders. Another less investigated aspect is immunode ciency and recurrent infections [24][25][26][27] .

Conclusions
In conclusion, our study did not identify any sex predominance in any domain. The domain most frequent associated with LQTS is the cytoplasmic domain (table 1). The domain potentially linked to Brugada syndrome is N-terminal. Mortality is statistically signi cant considering ACMG score p = 0.000 (Fig. 3a) and localisation domain p = 0.028 (Fig. 3b), even more signi cant when we considered Cytoplasmic and Transmembrane vs other groups p = 0.005 (Fig. 3c). HCM and pathogenicity are considered independent risk factor for mortality from localisation of domains (p = 0.031 and 0.04). To the best of our knowledge this is the rst comprehensive study trying to create phenotype genotype correlation and better risk strati cation for patients carrying variants with CACNA1C, even if we have to consider the limitations of the study.

Study Limitation
Due to the rarity of the condition, the main limitation of the study is the number of patients. Such a small number may prevent to nd any conclusive ndings regarding localization domain as an independent predictor of death. Another limitation regards the attempt of using published data to draw meaningful conclusions. There were some di culties in retrieving data for patients from literature review and some genetic tests (fourteen patients) were unfortunately unavailable. We used in our review analysis information that we have found. Moreover, for younger ones, instead, there is a brief follow-up, which may lead to underestimation of long-term morbidity and mortality. The retrospective study is also a limitation, but this is necessary since this is a rare cardiac condition.

Declarations
Ethics approval The study protocol conforms to ethical guidelines of the 1975 Declaration of Helsinki, as re ected in a priori approval by the Institution's Human Research Committee.

Consent for publication
Not applicable Availability of data and materials The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request.

Competing interests
The authors declare that they have no competing interests Funding