Concerns About the HCM Risk-Kids Study

To the Editor Norrish et al¹ recently published a risk prediction model for sudden cardiac death equivalents (SCDE) in childhood hypertrophic cardiomyopathy with a C statistic of 0.69 (95% CI, 0.66-0.72) that, if applied, recommends implantable cardioverter-defibrillator implantation in 249 of 527 patients (47.2%) (out of patients with a risk calculated above 6%, 223 did not reach an end point within 5 years’ follow-up and 26 did have an event). However, only 6.5% of total group (34 of 527) experienced SCDE. What concerns us about this publication is the inadequate evidence for the surprising conclusion that a left-ventricular outflow tract gradient (LVOTG) appeared protective in the final algorithm, which was arrived at by using statistical procedures to impute data for missing predictors (48.5% of patients had missing data). The conclusion seems dubious when one scrutinizes the raw data. Table 3¹ shows that the group with SCDE had a median gradient that was marginally but nonsignificantly higher than those without SCDE but that 31.7% of patients with SCDE had an LVOTG of 30 mm Hg or greater, whereas only 20.7% of patients without SCDE had an LVOTG of 30 mm Hg or greater; thus, an excess of patients with SCDE had a hemodynamically significant gradient. That the presence of an LVOTG should be protective in childhood when numerous studies in adult hypertrophic cardiomyopathy (HCM) shows it being a risk factor that is included in the European Society of Cardiology HCM risk algorithm² seems inherently unlikely. A possible explanation for this paradox could be a modifying effect by β-blocker therapy. From the data in Table 2,¹ 40.2% received β-blockers. There are now 5 separate retrospective cohort studies from the United Kingdom, Sweden, Taiwan, and the United States showing β-blocker therapy to be associated with a reduction in mortality in both adult and pediatric patients with HCM, and several studies show the effect to be dose related.³ A 2017 study of risk factors for SCDE in childhood HCM⁴ found, in contrast to the study by Norrish et al,¹ that an LVOTG at presentation was a risk factor for sudden death with a relative risk of 3.7. It remained a significant independent risk factor in the final multivariate model (with only 8.3% missing data), together with electrocardiographical (ECG) limb-lead voltage sum or ECG risk score, severity of hypertrophy, and a protective effect of early β-blocker dose.⁴ It is a reasonable hypothesis that patients with an outflow obstruction would be more likely to receive β-blocker therapy and might receive larger doses. Thus, a confounding effect of β-blocker dose needs to be examined when assessing risk factors for SCDE. Furthermore, the inclusion in the risk algorithm of published ECG measures that have been found to be correlated with SCDE in childhood HCM^4,5 will likely improve further versions of risk-assessing algorithms.