High mortality associated with gram-negative bacterial bloodstream infection in liver transplant recipients undergoing immunosuppression reduction

BACKGROUND Immunosuppression is an important factor in the incidence of infections in transplant recipient. Few studies are available on the management of immunosuppression (IS) treatment in the liver transplant (LT) recipients complicated with infection. The aim of this study is to describe our experience in the management of IS treatment during bacterial bloodstream infection (BSI) in LT recipients and assess the effect of temporary IS withdrawal on 30 d mortality of recipients presenting with severe infection. AIM To assess the effect of temporary IS withdrawal on 30 d mortality of LT recipients presenting with severe infection. METHODS A retrospective study was conducted with patients diagnosed with BSI after LT in the Department of Liver Surgery, Renji Hospital from January 1, 2016 through December 31, 2017. All recipients diagnosed with BSI after LT were included. Univariate and multivariate Cox regression analysis of risk factors for 30 d mortality was conducted in the LT recipients with Gram-negative bacterial (GNB) infection. RESULTS Seventy-four episodes of BSI were identified in 70 LT recipients, including 45 episodes of Gram-positive bacterial (GPB) infections in 42 patients and 29 episodes of GNB infections in 28 patients. Overall, IS reduction (at least 50% dose reduction or cessation of one or more immunosuppressive agent) was made in 28 (41.2%) cases, specifically, in 5 (11.9%) cases with GPB infections and 23 (82.1%) cases with GNB infections. The 180 d all-cause mortality rate was 18.5% (13/70). The mortality rate in GNB group (39.3%, 11/28) was significantly higher than that in GPB group (4.8%, 2/42) (P = 0.001). All the deaths in GNB group were attributed to worsening infection secondary to IS withdrawal, but the deaths in GPB group were all due to graft-versus-host disease. GNB group was associated with significantly higher incidence of intra-abdominal infection, IS reduction, and complete IS withdrawal than GPB group (P < 0.05). Cox regression showed that rejection (adjusted hazard ratio 7.021, P = 0.001) and complete IS withdrawal (adjusted hazard ratio 12.65, P = 0.019) were independent risk factors for 30 d mortality in patients with GNB infections after LT. CONCLUSION IS reduction is more frequently associated with GNB infection than GPB infection in LT recipients. Complete IS withdrawal should be cautious due to increased risk of mortality in LT recipients complicated with BSI. IS