ABSTRACT & COMMENTARY
Predicting Sudden Death Risk in Hypertrophic Cardiomyopathy
By Michael H. Crawford, MD, Editor
O’Mahony C, et al. A novel clinical risk prediction model for sudden cardiac death in hypertrophic cardiomyopathy (HCM Risk-SCD). Eur Heart J 2014;35:2010-2020.
The role of sudden death in hypertrophic cardiomyopathy (HCM) can be ameliorated by an implantable cardioverter-defibrillator (ICD), but the risks of an ICD must be balanced with the risk of sudden death. Traditional clinical risk factors tend to overestimate sudden death risk, hence these investigations at six European centers used data from a retrospective longitudinal cohort study to develop a prognostic model in 3675 patients with clinical HCM (85%) or familial disease. The subjects were ≥ 16 years of age with no history of ventricular arrhythmias, who met standard echocardiographic criteria for HCM and had no other explanation for left ventricular hypertrophy. Sudden death was defined as death within an hour of new symptoms, a nocturnal death, or an appropriate ICD shock in the 15% with ICDs. A prediction equation based on multiple clinical variables was derived from a development population (5 of the 6 centers) and tested in a validation cohort (one of the centers). The new equation was compared to traditional clinical risk factors: severe hypertrophy (wall thickness ≥ 30 mm), nonsustained ventricular tachycardia (NSVT), family history of sudden death, and unexplained syncope. Blood pressure response to treadmill exercise was not included because it has not been independently associated with sudden death.
During a median follow-up of 5.7 years, 5% of the patients experienced sudden death. Six clinical factors that predicted sudden death at the 15% significance level were included in the model: maximal wall thickness, left atrial diameter, left ventricular outflow tract (LVOT) gradient, family history of sudden death, NSVT, and unexplained syncope. The C-Index for the equation was 0.70. If an ICD is placed in 16 patients with a ≥ 4% predicted risk of sudden death, one patient will potentially be saved over 5 years. The authors concluded that this validated risk prediction model for HCM patients provides accurate information about sudden death risk using readily available clinical data.
COMMENTARY
The current four or five clinical variables analyzed in a binary fashion are a crude predictor of risk and have been shown to have a low predictive value for sudden death. Their use leads to an overestimation of risk and the placement of ICDs in many low-risk individuals. This new model benefits from a large patient population that is well characterized and includes the full spectrum of HCM patients from severe symptomatic disease to family members identified by echocardiography. It is highly accurate with a C-statistic of 0.70. Using the cutoff of 4% or higher predicted rate of sudden death over 5 years, 71% of the sudden death victims were identified. Also, the formula could be placed online and it could be updated as new information is obtained.
Although this is an improvement over our current risk prediction approach, it does have limitations. There is no consensus what constitutes high risk; 4% was chosen here because it most accurately predicted who would have sudden death. Perhaps for clinical purposes a different cutoff would be better. Also, the factors that the investigators included in their initial analysis were only those that had been proven independently predictive in other multivariate studies. Hence, exercise blood pressure response was not included. There may be other variables that would improve the formula; for example, genetic studies and MRI late gadolinium enhancement. In addition, age may be important clinically. In two individuals with identical risk scores, you might treat a 20-year-old differently from a 70-year-old. Although a large diverse population was studied, certain subgroups were poorly represented such as elite athletes, pediatric patients, Asians, and past myectomy or alcohol septal ablation patients. This risk predictor should be used with caution in such patients. Interestingly, once LV wall thickness was > 35 mm, the risk of sudden death fell, so it might not be accurate in this group also. Finally, the effects of various drug therapies were not considered. Many of these issues could be addressed with more data, which could be incorporated into the model. No prediction formula will ever be 100% accurate, but this one seems like a significant step toward more accurate risk prediction in HCM. We await the online version.
