Risk prediction for liver injury in Epstein-Barr virus infection in pediatric respiratory tract infections

Abstract

Background Epstein-Barr virus (EBV) infection is likely to co-occur in pediatric respiratory tract infections (RTIs). Liver injury is the common complication of EBV infection. The detailed risk factors for liver injury in EBV infection remain elusive. We aimed to investigate the incidence, characteristics and potential risk factors for liver injury in EBV infection for early risk prediction. Methods We retrospectively recruited the pediatric RTIs cases with EBV infection according to a predefined criteria from our hospital between January 2015 and December 2017. We extracted the clinical and laboratory data from the electronical medical records. The impact of age, gender, and various parameters on the liver injury risk was investigated. Univariate logistic regression analysis was performed to analyse the association between clinical/laboratory parameters and liver injury. The related indexes were enrolled in the multivariate logistic regression analysis. Decision curve analysis was used to yield the value of related parameters in predicting liver injury. Receiver operating curve (ROC) analysis was applied to produce the C-index of white blood cell (WBC) count for liver injury. We also tested the non-linear association between WBC count and alanine aminotransferase (ALT). Results A total of 216 pediatric RTIs with EBV infection were enrolled. EBV infection is more likely to occur during the winter season. Cytomegalovirus infection was independently associated with liver injury in EBV infection (OR = 6.972, 95% CI = 1.648–29.490, p = 0.008). WBC count was independently associated with liver injury in EBV infection (OR = 1.169, 95% CI = 1.051–1.301, p = 0.004). The P interaction value between WBC count and cytomegalovirus was 0.149. The decision curve analysis showed that WBC count had larger area under curve compared with platelet (PLT) and birthweight (BW). ROC analysis yielded the c-index of WBC count: 0.75 and cut-point of 8.3. The turning point of WBC count in its association with ALT was 16.8. The p value before and after the turning point was < 0.001 and 0.123, respectively. Conclusions Cytomegalovirus co-infection demonstrated 5.972 more times of liver injury risk in EBV infection. WBC count was an independent biomarker for liver injury before the turning point of 16.8 in EBV infection. More attention should be paid to the risk of EBV infection in the winter. Cytomegalovirus infection and WBC count merit attention in the monitoring of possible liver injury in EBV infection among pediatric RTIs.