HCM Risk-SCD

Clinical background

Hypertrophic cardiomyopathy (HCM) is one of the commonest inherited heart diseases and a leading cause of sudden cardiac death in young adults. The decision on a primary preventive implantable cardioverter defibrillator (ICD) is central but difficult: the annual incidence of sudden cardiac death averages 0.5 to 1 per cent, which means that most patients are never affected, while ICD implantation carries real risks in the form of inappropriate shocks, infection and procedural complications. Earlier risk stratification rested on a binary model with a limited number of risk factors, with the result that too many patients ended up in a grey zone where the decision became arbitrary. HCM Risk-SCD was developed to replace that model with a continuous, individualised probability estimate, and was adopted as the primary risk stratification tool in the 2014 ESC guidelines. The 2023 ESC guidelines on cardiomyopathies retain HCM Risk-SCD as the recommended tool (class I, level of evidence B) [1,4].

Calculating HCM Risk-SCD

The model calculates the individual 5-year probability of sudden cardiac death, defined as a composite of sudden death or an appropriate ICD shock. The prognostic index is calculated as:

$$\text{PI} = 0{.}15939858 \times \text{MWT} - 0{.}00294271 \times \text{MWT}^2 + 0{.}0259082 \times \text{LAD} + 0{.}00446131 \times \text{LVOT}$${\max} + 0{.}4583082 \times \text{FH}{\text{SCD}} + 0{.}82639195 \times \text{NSVT} + 0{.}71650361 \times \text{syncope} - 0{.}01799934 \times \text{age}

where MWT is the maximal left ventricular wall thickness (mm), LAD is the left atrial diameter (mm), LVOTmax is the maximal LVOT gradient at rest or on Valsalva (mmHg), FHSCD is a family history of sudden cardiac death (0 or 1), NSVT is non-sustained ventricular tachycardia on ambulatory monitoring (0 or 1), syncope is unexplained syncope (0 or 1) and age is the age at assessment (years). The 5-year probability is obtained as:

$$P_{\text{SCD, 5 years}} = 1 - 0{.}998^{\exp(\text{PI})}$$

The derivation cohort consisted of 3,675 consecutive patients with HCM from six European centres, recruited retrospectively. Over a median follow-up of 5.7 years (24,313 patient-years) there were 198 events (5 per cent), defined as sudden death or an appropriate ICD shock. Of eight predefined predictors, one (an abnormal blood pressure response to exercise) was rejected because it lacked association at the 15 per cent level; the remaining seven form the variables of the model. Internal validation was performed by bootstrapping. The calibration slope was 0.91 (95 per cent CI 0.74 to 1.08), the C-index 0.70 (95 per cent CI 0.68 to 0.72) and the D-statistic 1.07 (95 per cent CI 0.81 to 1.32) [1].

Interpretation in practice

The ESC guidelines define three risk bands from the calculated 5-year risk. Each band corresponds to a class of recommendation for an ICD, but the decision must always be made in dialogue with the patient after an overall clinical assessment.

An important addition is that the 2023 ESC guidelines formally recognise further risk modifiers that may influence the ICD decision independently of the score, above all extensive late gadolinium enhancement (LGE comprising 15 per cent or more of the left ventricular mass) and a reduced left ventricular ejection fraction (below 50 per cent) [4].

Validation and performance

The first external validation was published by Vriesendorp et al. in 2015 and comprised 706 patients from two tertiary referral centres in the Netherlands and Belgium. Over a follow-up of 7.7 ± 5.3 years there were 42 events (5.9 per cent). The c-statistic was 0.69 (95 per cent CI 0.57 to 0.82), significantly better than older risk factor-based models (C-index 0.55 and 0.60) [2].

In a Brazilian cohort of 187 patients (mean age 41.5 years, follow-up 8.3 years, 16 events) the discrimination of HCM Risk-SCD was appreciably poorer. The AUC was 0.58 at both the 4 and the 6 per cent threshold, and Kaplan–Meier analysis showed no significant difference in event-free survival between the risk groups. At the 4 per cent threshold, sensitivity was 65 per cent and specificity 54 per cent; at the 6 per cent threshold, sensitivity fell to 44 per cent. The American ACC/AHA model performed somewhat better in this population (AUC 0.63, sensitivity 81 per cent) [3].

A Chinese validation from Fuwai Hospital comprised 534 patients recruited between 1992 and 2010, with a mean follow-up of 6.96 years and 31 SCD events. The group with a calculated risk of 4 per cent or higher had a significantly higher incidence of SCD (8.68 per cent versus 3.42 per cent, p = 0.01), and the actual 5-year risk (4.31 per cent) agreed well with the predicted risk (4.65 per cent). The AUC was, however, only 0.660 in the whole population, and the optimal threshold was calculated as 3.23 per cent, that is, lower than the ESC threshold of 4 per cent [4].

A large Japanese multicentre study (REVEAL-HCM) with 3,611 patients showed that the ESC model's risk stratification between intermediate and high risk was problematic: the 5-year incidence of SCD events was 2.9 per cent in class IIa (risk 6 per cent or higher) and 9.3 per cent in class IIb (risk 4 to below 6 per cent), that is, the reverse of what was expected. The ACC/AHA guidelines stratified the corresponding classes better [5].

In summary, HCM Risk-SCD performs with moderate discrimination (a C-index of around 0.70) in European populations, but performance falls in non-Caucasian cohorts and sensitivity at the recommended thresholds is often low, particularly at the 6 per cent threshold.

Limitations

The model is not validated for patients under 16 years, for previous survivors of sudden cardiac arrest (in whom secondary prevention applies whatever the score) or for infiltrative or metabolic phenocopies of HCM such as amyloidosis or Fabry disease. Patients who had undergone septal reduction (septal myectomy or alcohol ablation) were excluded from the derivation cohort, and the reliability of the model after such procedures is unclear.

HCM Risk-SCD does not include late gadolinium enhancement on cardiac MRI, left ventricular apical aneurysm, genetic information or the ejection fraction as variables, even though these factors have shown prognostic value. This is a deliberate trade-off to keep the model applicable with data available at a standard work-up, but it means that the score may underestimate risk in patients in whom these factors are prominent.

The quadratic term for the maximal left ventricular wall thickness means that the risk contribution of wall thickness plateaus and eventually declines at very high values, which is biologically reasonable but may be counter-intuitive in extreme hypertrophy.

That the score was derived retrospectively at tertiary referral centres introduces a selection skew: the cohort represents a more symptomatic and more severely affected patient population than that seen in general cardiology practice, which may exaggerate the calculated risk when the model is applied to broader populations.

References

1. O'Mahony C, Jichi F, Pavlou M, et al. A novel clinical risk prediction model for sudden cardiac death in hypertrophic cardiomyopathy (HCM risk-SCD). Eur Heart J. 2014;35(30):2010-20. PMID: 24126876
2. Vriesendorp PA, Schinkel AF, Liebregts M, et al. Validation of the 2014 European Society of Cardiology guidelines risk prediction model for the primary prevention of sudden cardiac death in hypertrophic cardiomyopathy. Circ Arrhythm Electrophysiol. 2015;8(4):829-35. PMID: 25922410
3. Oliveira Antunes M, Fernandes F, Arteaga-Fernandez E, et al. Validation of ACC/AHA and ESC Sudden Cardiac Death Risk Guidelines in Diverse Hypertrophic Cardiomyopathy Cohort: Stratification HCM Study. Glob Heart. 2024;19(1):94. PMID: 39713197
4. Hang F, Fan C. Validation of the HCM Risk-SCD Model in a Chinese Hypertrophic Cardiomyopathy Cohort. J Clin Med. 2025;14(20):7355. PMID: 41156225
5. Amano M, Kitaoka H, Yoshikawa Y, et al. Validation of Guideline Recommendation on Sudden Cardiac Death Prevention in Hypertrophic Cardiomyopathy. JACC Heart Fail. 2025;13(6):1014-1026. PMID: 40088231