Primary-prevention ICDs are indicated for most patients with LVEF ≤35%. Some patients improve their LVEF to >35% during the life of their first ICD. In present practice, when the battery is depleted, ICDs most often are replaced regardless of LVEF. In patients with recovered LVEF who have never received appropriate ICD therapy, the utility of ICD generator replacement remains unclear.
We enrolled patients with a primary-prevention ICD originally implanted for LVEF ≤35%, who underwent ICD generator change within our state’s largest multihospital health system. Patients who required appropriate ICD therapy for VT/VF prior to generator change were excluded. Cumulative incidence curves were Fine-Gray adjusted for the competing risk of death.
Among 951 generator changes, 423 patients (69±12 y, 65% men, 29% Black, LVEF 34±15%, 222 [52%] ischemic) met inclusion criteria. Over 3.4±2.2 years after generator change, 78 (18%) received appropriate therapy for VT/VF. Compared to patients with recovered EF>35% (n=161 [38%]), those with LVEF persistently ≤35% (n=262 [62%]) more likely required ICD therapy (p=0.005; 5-year rates: 13% vs. 25%). ROC analysis ( AUC 0.66, p<0.001) revealed the optimal cutoff for VT / VF prediction was LVEF 45%, which was supported by the plot of hazard vs. EF as a continuous variable, modeled by restricted cubic splines. There was much lower VT/VF incidence among those with LVEF ≥45% vs. <45% (p<0.001); 5-year rates: 6% vs. 25%. These findings were similar for patients with either ischemic or nonischemic cardiomyopathy (HR 3.9, p<0.01; and HR 8.5, p=0.035).