Improving sudden cardiac death risk stratification in hypertrophic cardiomyopathy using established clinical variables and genetic information

Hypertrophic cardiomyopathy (HCM) is the most prevalent genetic myocardial disease [1, 2, 22]. It can be morphologically defined by non-secondary, left ventricular hypertrophy. Histopathological characteristics of HCM are hypertrophied myocytes with cellular disarray and marked interstitial fibrosis [10]. The most common cause of HCM is the presence of disease-associated variants in genes coding for sarcomeric proteins [1, 20]. The wide availability of cardiac imaging capabilities and the increasing awareness of HCM has enabled clinicians to successfully and reliably identify more patients suffering from HCM at earlier stages of the disease [12, 15, 21, 23] and with different disease manifestations, e.g. the dynamic obstruction of blood-flow in the left ventricular outflow tract (LVOT) [4, 10, 21].

HCM remains one of the main causes of sudden cardiac death (SCD) in young patients even though the majority of the patients present mild symptoms at diagnosis. Previous studies and registers describing the natural history of HCM have reported an increased risk of SCD in HCM patients in comparison to the general population. Several risk factors have been identified over the course of the last 50 years that are associated with an increased risk of SCD. Although SCD is a tragic and devastating complication, the absolute number of events that is reported in HCM patient registries is relatively low and ranges at around 1.1% per year [14, 17, 24]. An adequate SCD risk assessment for all patients with HCM is essential and includes comprehensive clinical and family history, Holter-monitoring, ECG, cardiac imaging and exercise testing [4, 16, 26]. Careful consideration is warranted when identifying patients that qualify for a primary prophylactic ICD implantation. On one hand, the emphasis certainly relies on reaching a higher sensitivity to avoid missing patients with a higher risk of SCD. On the other hand, increasing the sensitivity using the currently available clinical tools will put more patients at risk of complications of device therapy without having a need.

An underlying disease-causing mutation can be identified in up to 60 percent of the patients with HCM [1, 4, 20]. In accordance with ESC and AHA guidelines, genetic testing should be discussed and evaluated for each patient suffering from HCM [1, 4, 6, 13, 27]. The majority of gene variants can be found in cardiac myosin-binding protein C (MYBPC3) gene and the myosin heavy chain (MYH7) gene [2, 20] with sarcomere mutations being associated with worse outcome [3, 5, 7, 8, 11, 29]. Still, the use of this genetic information in clinical routine is still sparse in many countries.

The ESC HCM guidelines recommend the incorporation of the 5-year SCD Risk Score Calculator (ESC RSC) in the decision making (which in effect relatively favors specificity). The 5-year Risk Score Calculator model has been developed through a multicenter study of 3675 patients [26]. Using clinical parameters only, the model estimates the individual 5-year SCD risk and divides the patients into three categories: patients with low (< 4% risk in 5 years), intermediate and high risk (> 6% per 5 years) [4, 26]. Current guidelines recommend primary prophylactic ICD-implantation considering the individual SCD risk for each patient with the threshold being 5-year SCD risk ≥ 6% for the ESC model (Class IIb recommendation) or the presence of ≥ 1 risk factor for the 2020 AHA/ACC approach (Class IIa/b recommendation) [4, 27]. Genetic results are not integrated into the SCD risk stratification, although the incidence of malignant arrhythmia is reported to be higher in patients with variants in genes encoding sarcomeric proteins [17, 18, 29, 30]. Patients carrying a disease-causing variant (genotype-positive) are reported to have a worse outcome than patients without a disease-causing mutation (genotype-negative) [17, 18, 29, 30]. The relatively high costs of genetic sequencing and low accessibility in many countries has prohibited integration in the risk models.

The aim of this study is twofold. Firstly, we aim to validate the ESC RSC and the AHA HCM risk stratification guidelines in a German HCM patient population in a longitudinal study. Secondly, to investigate the benefit of integrating genetic findings into the ESC 5-year risk model.

HCM patients have an increased risk for SCD in comparison to the general population. While HCM remains one of the most common causes of SCD in the young, the absolute risk for HCM individuals is relatively low [14, 17, 19, 24]. By advancing diagnostic standards, more patients with HCM are diagnosed, especially at early stages of the disease. This subsequently increases the difficulty in identifying patients with an appropriate primary prophylactic ICD indication.

The ESC RSC was validated in different external patient populations, mainly from the UK, Spain, Greece and Italy. To our knowledge, this is the first time that the score is validated in a larger German patient population. Besides the European model, we also evaluated the AHA HCM SCD risk stratification approach and compared both methodologies in the same cohort. Both approaches performed well in identifying patients with risk for SCD, achieving an AUC of 0.74 (ESC RSC) and 0.70 (AHA SCD risk stratification). The definition in the cutoffs for ICD indication is chosen differently in both models: The ESC working-group decided to implement a cut-off favoring specificity, which we can find in our cohort with 83.3%, albeit at the cost of a low sensitivity at 28.6% (cut-off of 6%). The AHA SCD risk stratification approach expectedly showed high sensitivity with 92.9%, but at the cost of a low specificity (27.9%) at the AHA cut-off ≥ 1 risk factor. Similar results were shown in previous studies [9, 24, 31]. We evaluated the integration of genetic findings into the risk models to further improve patient stratification. To avoid complexity and increase the ease of use in the daily routine, we relied on the ESC RSC and integrated the new variable: genotype. A higher SCD event rate in genotype-positive patients has been previously reported in literature [18, 29, 30]. While sarcomeric variants in general were shown to result in higher SCD risk, results of contemporary studies do not allow to do a gene or even variant specific prediction. Moreover, the event rate in HCM patients is relatively low, which makes the expansion of the risk score for specific variants and genes currently infeasible. Due to this reason and for simplicity in daily clinical routines, we have chosen to classify the patient population in 2 groups, genotype-positive (LP or P) and genotype-negative patients (Class I-III according to ACMG). The weighting of the genetic factor was calculated analog to the original formula using cox proportional hazards model with data from previously published literature [18, 30]. The modified Genotype-model based on ESC-RSC (with the integration of genetic findings into the formula) seams to improve prediction especially in the patient population identified as low/intermediate risk.

Extreme values and outliers could affect the diagnostic performance of the scores. The clinical decision making regarding the indication of prophylactic ICD implantation can be assumed to be binary, either the cut-off value is reached, and an ICD implantation is warranted, or the cut-off is not reached and an ICD implantation would not be recommended. Using the existing cut-offs from guidelines, we analyzed the performance of the approaches again in a binary approach. Although the original ESC RSC showed a low sensitivity and misclassified 10 patients as low-risk, it correctly identified 96% of the patients that did not have a higher risk of malignant arrhythmia (achieving a very high specificity). The Genotype-model achieved the highest precision, followed by ESC RSC and AHA SCD risk stratification. This is reiterated in the calculation of NNT per risk stratification approach, where the lowest NNT was observed in the Genotype-model. However, it should be noted that the cut-off values for recommending primary prophylactic ICD implantation are determined by the cardiac societies and follow physician consideration, patient expectations, expert opinions and socioeconomic considerations.

A total of 14 events were observed over a follow-up time of 5.77 years. The ESC RSC identified 4 patients correctly, 3 patients were classified as intermediate risk. However, 7 patients were incorrectly classified as low-risk. The modified Genotype-model misclassified 2 cases as low risk. Even with the highly sensitive approach of the AHA recommendations, 1 event would not have been identified in our patient population, but quiet some “unnecessary” ICD implantations would have been performed, judged within the observation period of 5 years. These facts underline the importance of shared decision making.

In summary, current risk stratification approaches are validated in this German population and can aid in the SCD risk stratification in HCM patients. The Genotype-model is not presented as a substitute to the original guidelines, but provides the ability to be integrated in the contemporary care of HCM patients. The derivation of the new model has its limitations: The addition of the genetic score was based on data from an external HCM patient population and might result in changes in model calibration and adaptations of the cut-offs. Furthermore, this is a monocenter study and a model-optimization, fitting, and validation in an external, larger, independent HCM patient population needs to be performed in future studies. Patients treated with novel myosin inhibitors, such as Mavacamten, where not included in this study, as well as children with HCM. Additionally, there remains a small subpopulation of HCM patients that do not present with the typical risk-factors for electrical vulnerability and appear to be misclassified in the current guidelines. Further research is warranted to understand this patient population, especially to identify further SCD risk factors.