Pre-LVAD Blood-Borne Infection is a Harbinger of Poor Prognosis after LVAD Implantation

Purpose Infections are common post-LVAD and are associated with increased mortality. Multiple risk factors have been identified including age and INTERMACS profile. However, the effect of pre-LVAD infections on post-LVAD outcomes is poorly studied. We aimed to determine the impact of pre-LVAD infections on post-LVAD infection risk and survival. Methods We retrospectively reviewed 106 pts implanted with LVAD at a large volume center. Infection data including type (bacterial, fungal, and viral) and source (blood-borne vs others) were collected pre- and post-LVAD. The cohort was divided into 4 groups based on source and type of pre-LVAD infection: Group 1: blood-borne bacterial or fungal, Group 2: other bacterial or fungal, Group 3: viral, Group 4: no infection. Survival free from any post-LVAD infection and survival were assessed using Cox proportional hazards. Multivariable analyses were performed adjusting for age and INTERMACS profile (≤ 2 vs > 2). Results Among 106 pts (age 58±1 yr, 82% male, 62% white), 47 pts (44%) had pre-LVAD infection: 40 pts (38%) bacterial or fungal, 7 (7%) viral; 13 (12%) had blood-borne infection. 58 pts (55%) had a post-LVAD infection. Median time to any infection was 21 days (IQR: 7-39) for the entire cohort and 11 days (IQR: 4-36) for pts with pre-LVAD blood-borne infection. Pts in Group 1 had significantly worse survival free from infection compared to other groups (Group 1 vs 4, HR CI: 3.3 (1.7-6.6)) and significantly worse survival compared to Group 4 (HR CI: 5.5 (1.4-22.1)) (Figure). Notably, survival at 1 yr was 59% for Group 1 compared to 93% for Group 4. After adjusting for age and INTERMACS profile, this relationship remained significant for the composite endpoint, but not survival alone. For pts in Group 1 with INTERMACS profile ≤ 2, 1 yr survival was 57%. Conclusion Occurrence of pre-LVAD blood-borne infection was associated with 72% increase in mortality or post-LVAD infection. This finding may carry important implications for patient selection.