Frequency of recovery and relapse in patients with nonischemic dilated cardiomyopathy on guideline-directed medical therapy.

Several key clinical questions, such as which patients with dilated cardiomyopathy (DC) will recover, how many will relapse, when will they relapse, and predictors of relapse, have sparse data. The present study examines the frequency and predictors of recovery and relapse in patients with DC. One hundred eighty-eight patients of a nonischemic DC cohort having baseline left ventricular ejection fraction (LVEF) ≤ 40% were divided into 3 groups: improved group with sustained recovery of LVEF to >40% with a net increase in LVEF of ≥ 10% from baseline, not-improved group without change or decrease in LVEF compared with that in baseline including patients with an increase in LVEF <10%, and relapsed group with decrease in LVEF ≥ 10% after initial improvement. Follow-up duration was 50 ± 31 months. One hundred ten patients (59%) did not improve. Of the 78 patients (41%) who improved, 50 (64%) had sustained improvement. Remaining 28 (36%) of the 78 improved patients relapsed on further follow-up of 36 ± 25 months. Baseline LVEF was similar in the 3 groups. Mean LVEF increased from 29 ± 8% to 50 ± 7% (p <0.001) in the improved group, changed from 27 ± 9% to 25 ± 9% (p = 0.95) in the not-improved group, and, after increasing from 30 ± 7% to 52 ± 6%, it decreased to 34 ± 9% (p <0.001) in the relapsed group. Multivariate analysis showed that the only variable associated with recovery of LVEF was shorter QRS duration (odds ratio 0.31, 95% confidence interval 0.15 to 0.67, p = 0.003). Recurrence of left ventricular systolic dysfunction was associated with long QRS duration (odds ratio 3.52, 95% confidence interval 1.27 to 9.76, p = 0.01). In conclusion, with currently recommended medical therapy, 1/4 of patients with nonischemic DC have sustained improvement, and >1/3 of those who improve relapse. QRS duration predicted both recovery and relapse. The survival rate of patients in the improved group was significantly better than that in the other 2 groups (p = 0.03, log-rank).