The interaction between soluble CD73 and 90-day mortality from patients with non-septic shock and sepsis shock: a secondary analysis from the prospective FINNAKI study

Abstract

Fundamental researches have shown that soluble CD73 (sCD73) can inhibit inflammatory response and limit excessive tissue damage caused by continuous immune cell activation. A Finnish prospective, observational study of acute kidney injury (FINNAKI) showed no association between sCD73 and 90-day mortality in sepsis patients. Clinical data of this study was used for secondary analysis to explore whether the relationship between sCD73 and 90-day mortality was consistent in septic shock and non-septic shock patients. The FINNAKI study was a prospective, observational cohort study conducted in 17 intensive care units (ICUs) in Finland from September 1st, 2011 to February 1st, 2012. Sepsis/septic shock was defined according to Sepsis-1 definition. Demographic characteristics, treatment, comorbidities and 90-day mortality of the patients were analyzed. To evaluate the difference (interaction test) between the relationship of sCD73 and 90-day mortality in septic shock and non-septic shock patients, likelihood ratio test was used to integrate the product term (sCD73×septic shock or non-septic shock) into multivariable Logistic regression. Sensitivity analysis was performed with the definition of Sepsis-3. The interaction between sCD73 and 90-day mortality in patients with septic shock and non-septic shock were verified by generalized additive model (GAM). A total of 588 patients with severe sepsis/septic shock were enrolled. 164 patients died in 90 days, and the 90-day mortality was 27.89%. Based on the Sepsis-1 definition, there were 159 non-septic shock patients and 429 septic shock patients. Compared with the non-septic shock patients, lactate (Lac) level, sequential organ failure assessment (SOFA) score, fluid balance on the first day, and ratio of mechanical ventilation, 12-hour acute kidney injury (AKI), renal replacement therapy (RRT), and postoperative ICU transition in the septic shock patients were significantly increased and the proportion of emergency admission to ICU was significantly decreased. Based on the Sepsis-3 definition, there were 383 non-septic shock patients and 205 septic shock patients; the results of clinical data analysis between the two groups were similar to those based on Sepsis-1. Based on Sepsis-1, there was no significant difference in 90-day mortality between non-septic shock and septic shock patients [23.90% (38/159) vs. 29.37% (126/429), P > 0.05]. However, based on Sepsis-3, the 90-day mortality of patients with septic shock was significantly higher than that of patients with non-septic shock [37.56% (77/205) vs. 22.72% (87/383), P < 0.01]. Multivariate Logistic regression analysis and interaction test showed that after adjusting all confounding factors (except the number of complications) in non-sepsis shock and sepsis shock patients, sCD73 and 90-day mortality were significantly different in both Sepsis-1 and Sepsis-3. The P values for interaction tests were 0.046 and 0.027, respectively. In patients with non-septic shock, sCD73 tended to be positively associated with 90-day mortality [Sepsis-1: odds ratio (OR) = 1.46, 95% confidence interval (95%CI) was 0.99-2.13, P = 0.053; Sepsis-3: OR = 1.34, 95%CI was 1.02-1.74, P = 0.034]. In septic shock patients, sCD73 tended to be negatively associated with 90-day mortality (Sepsis-1: OR = 0.91, 95%CI was 0.69-1.20, P = 0.494; Sepsis-3: OR = 0.80, 95%CI was 0.55-1.17, P = 0.249). The results of GAM model validation were consistent with the results of Logistic regression equation cross validation. The relationship between sCD73 and 90-day mortality is significantly different from patients with non-sepsis shock and sepsis shock. In patients with non-sepsis shock, sCD73 is trend to positively associated with 90-day mortality, and there is a negative trend between sCD73 and 90-day mortality in patients with septic shock.