Plasma TNFR1 predicts major adverse kidney events in hospitalized patients with COVID-19

Background: Patients hospitalized with COVID-19 are at risk for major adverse kidney events (MAKE). Predicting which patients will experience progression of disease and poor outcomes remains elusive. We sought to develop a biomarker-based risk model for predicting MAKE in patients hospitalized with COVID-19. Methods: We applied least absolute shrinkage and regression methodology (LASSO) to a prospectively enrolled cohort of 432 patients admitted with COVID-19 who had blood specimens collected (median 2 days [IQR 2-4 days] from admission) from March 2020-January 2021, at three large academic medical centers. Clinical variables and 26 plasma biomarkers were used as candidate features in the prediction models for the outcome of MAKE, defined as KDIGO stage 3 AKI, dialysis-requiring AKI, or in hospital mortality. Cross-validation was used for optimal shrinkage parameter selection and model AUCs were optimism-corrected using bootstrapping. Results: MAKE occurred in 85 (20%) patients within 60 days of admission. Application of LASSO to the 26 biomarkers selected IL-12, IL-13, IL-6, Tie2, FLT1, NGAL, MCP1, YKL40, Ang1, Ang2, and TNFR1, which yielded an AUC of 0.88 (95% CI 0.85-0.91). Plasma TNFR-1 alone had an AUC of 0.88 (0.84,0.91). When LASSO was applied on the clinical variables and TNFR1, 4 clinical variables (age, black race, obesity, WHO COVID severity score) and TNFR1 were selected with an AUC was 0.88 (95% CI 0.87-0.89). Random Forest models of biomarkers and clinical variables had similar prediction performance. A cutoff of TNFR1 at 3005 pg/ml had a sensitivity of 69%, specificity of 89%, PPV of 60% and NPV of 92% for occurrence of MAKE over 60 days. Conclusions: In this multi-center cohort study, plasma TNFR1 by itself produced a robust prediction for MAKE in patients hospitalized with COVID-19 that did not improve when combined with multiple clinical variables and was equivalent to combined inputs of 10 other plasma biomarkers.