P1525: CLINICAL PROGNOSTIC SCORING MODEL FOR VIRAL ENCEPHALITIS IN PATIENTS UNDERGOING ALLO-HAEMATOPOIETIC STEM CELL TRANSPLANTATION
Yu-Chen He,Chen-Hua Yan,Yuan Kong,Xiao-Lu Zhu,Hai-Xia Fu,Yun He,Feng-Rong Wang,Yuan-Yuan Zhang,Xiao-Dong Mo,Wei Han,Jing-Zhi Wang,Yu Wang,Huan Chen,Yu-Hong Chen,Xiang-Yu Zhao,Ying-Jun Chang,Lan-Ping Xu,Kai-Yan Liu,Xiao-Jun Huang,Xiao-Hui Zhang
DOI: https://doi.org/10.1097/01.hs9.0000972976.88590.63
2023-01-01
HemaSphere
Abstract:Topic: 30. Infections in hematology (incl. supportive care/therapy) Background: As a radical therapy for haematological malignancies, haematopoietic stem cell transplantation (HSCT) has been widely performed in recent years. However, numerous complications still threaten the outcomes of post-transplant patients, including viral encephalitis (VE), which is a common but fatal central nervous system infection in immunocompromised individuals. Currently, there is still a lack of research on the prognosis of patients with VE after HSCT. Aims: The principal objective of this project was to establish a clinical prognostic scoring model to recognize post-transplant VE patients who are at a higher risk for death. Methods: In our study, 115 VE patients who underwent HSCT from 2014 to 2021 were divided into derivation and validation cohorts containing 69 and 46 patients, respectively. The clinical baseline data were analysed first. Subsequently, we screened out the prognostic factors in the derivation cohort using Cox regression analysis and then constructed a clinical prognostic scoring model. After grouping patients in the derivation and validation cohorts according to their different prognoses, we estimated the performance of this model with the Kaplan-Meier method, the area under the receiver operating characteristic curve (AUC), the calibration plot and the decision curve analysis (DCA). Results: We performed univariate Cox regression analysis to select prognostic predictors for VE after HSCT. Variables with p<0.10 were subsequently included in the multivariate analysis. Patients with unconsciousness (p=0.012; RR, 3.416; 95%CI, 1.307-8.930), paresthesia (p=0.011; RR, 3.807; 95%CI, 1.354-10.623), cytomegalovirus (CMV) positivity in cerebrospinal fluid (p=0.014; RR, 5.311; 95%CI, 1.403-20.103) and grade III to IV acute graft-versus-host disease (aGVHD) (p=0.043; RR, 2.657; 95%CI, 1.031-6.848) were identified as predictors for our clinical prognostic scoring model. According to the regression coefficient, a score of 1 was assigned to each predictor. The overall scores of this model ranged from 0 to 4, upon which a 2-category scoring model was established: low (score of 0-1) and high (score of 2-4). In the derivation cohort, the corresponding 16-month OS according to each risk group was 78.7% (low) and 13.9% (high) (p<0.001). The AUC was 0.818 (95%CI, 0.724-0.913), showing good discrimination power. In the validation cohort, the 16-month OS results for the risk groups were 85.3% (low) and 20.0% (high) (p<0.001), and the AUC was 0.793 (95%CI, 0.644-0.942). Furthermore, we performed calibration plots which demonstrated good consistency among the prediction results in the derivation and validation cohorts. DCA exhibited the good accuracy of this model, and patients could benefit from its application. Summary/Conclusion: VE is a common but fatal complication of HSCT. Our study established a clinical prognostic scoring model to select high-risk post-transplantation patients with VE and validated its discrimination, calibration and clinical practicality. Keywords: CNS, Infection, Prognostic factor, Allogeneic hematopoietic stem cell transplant