Prediction of cardiac death in patients with hypertrophic cardiomyopathy using plasma adipokine levels

BACKGROUNDS AND AIMS

Hypertrophic cardiomyopathy (HCM) causes cardiac death through both sudden cardiac death (SCD) and death due to heart failure (HF). Although adipokines lead to adverse cardiac remodeling in HCM, the prognostic value of plasma adipokines in HCM remains unknown. We aimed to predict cardiac death in patients with HCM using plasma adipokines.

METHODS AND RESULTS

We performed a multicenter prospective cohort study of patients with HCM. The outcome was cardiac death including heart transplant, death due to HF, and SCD. With data from 1 institution (training set), a prediction model was developed using random forest classification algorithm based on 10 plasma adipokines. The performance of the prediction model adjusted for 8 clinical parameters was examined in samples from another institution (test set). Time-to-event analysis was performed in the test set to compare the rate of outcome events between the low-risk and high-risk groups determined by the prediction model. In total, 389 (267 in the training set; 122 in the test set) patients with HCM were included. During the median follow-up of 2.7 years, 21 patients experienced the outcome event. The area under the covariates-adjusted receiver-operating characteristics curve was 0.89 (95 % confidence interval [CI] 0.71-0.99) in the test set. revealed the high-risk group had a significantly higher risk of cardiac death (hazard ratio 17.8, 95 % CI 2.1-148.3, P = 0.008).

CONCLUSION

The present multicenter prospective study demonstrated that a panel of plasma adipokines predicts cardiac death in patients with HCM.

Covariate-specific ROC curve analysis can accommodate differences between covariate subgroups in the evaluation of diagnostic accuracy

OBJECTIVES

We present an illustrative application of methods that account for covariates in receiver operating characteristic (ROC) curve analysis, using individual patient data on D-dimer testing for excluding pulmonary embolism.

STUDY DESIGN AND SETTING

Bayesian nonparametric covariate-specific ROC curves were constructed to examine the performance/positivity thresholds in covariate subgroups. Standard ROC curves were constructed. Three scenarios were outlined based on comparison between subgroups and standard ROC curve conclusion: (1) identical distribution/identical performance, (2) different distribution/identical performance, and (3) different distribution/different performance. Scenarios were illustrated using clinical covariates. Covariate-adjusted ROC curves were also constructed.

RESULTS

Age groups had prominent differences in D-dimer concentration, paired with differences in performance (Scenario 3). Different positivity thresholds were required to achieve the same level of sensitivity. D-dimer had identical performance, but different distributions for YEARS algorithm items (Scenario 2), and similar distributions for sex (Scenario 1). For the later covariates, comparable positivity thresholds achieved the same sensitivity. All covariate-adjusted models had AUCs comparable to the standard approach.

CONCLUSION

Subgroup differences in performance and distribution of results can indicate that the conventional ROC curve is not a fair representation of test performance. Estimating conditional ROC curves can improve the ability to select thresholds with greater applicability.