RISK PREDICTION OF COVID-19 SEVERITY USING URINARY L-FABP AND THE SOFA SCORE

The Coronavirus Disease 2019 (COVID-19) pandemic has placed unprecedented strain on healthcare systems. A critical challenge is the early identification of patients who, despite presenting with mild symptoms, are at high risk of clinical deterioration. The Sequential Organ Failure Assessment (SOFA) score is the standard index for assessing multi-organ failure, calculated from vital signs and blood tests. Urinary L-type fatty acid-binding protein (L-FABP), in contrast, is a non-invasive biomarker whose gene is activated by hypoxia-responsive elements like HIF-1. Consequently, its levels rise early in response to renal ischemia or oxidative stress, suggesting it may increase before overt clinical decline. This study aimed to examine the added clinical value of combining admission L-FABP levels with the SOFA score to enhance risk stratification for COVID-19 severity.

We conducted a single-center, retrospective cohort study of patients with reverse transcriptase-polymerase chain reaction (RT-PCR)-confirmed COVID-19 admitted to the National Center for Global Health and Medicine in Japan between January 2020 and April 2022. From 1,025 recruited patients, 842 were eligible for analysis after applying exclusion criteria. Urine samples were collected to measure urine L-FABP levels, and SOFA scores were assessed, at two time points: on admission (day 0) and on day 7. The primary endpoint was disease severity at day 7, categorized as mild, moderate, or severe based on respiratory status. Predictive performance was evaluated using receiver operating characteristic (ROC) curve analysis. The area under the curve (AUC) was calculated to quantify performance, and optimal cut-off points were determined by maximizing Youden’s index.

The cohort included 842 patients. The number of severe cases increased from 7 at admission to 34 by day 7. Significant disease progression was observed; approximately 15% of initially mild patients deteriorated to moderate or severe, and 25 initially moderate patients progressed to severe. For predicting severe outcomes, the combined L-FABP+SOFA model yielded the highest AUC of 0.92, compared to the SOFA score (AUC 0.90) and L-FABP (AUC 0.81) alone. For identifying patients who progressed to moderate or severe disease, the combined model (AUC 0.88) demonstrated significantly superior predictive accuracy over either marker individually (p < 0.001). Notably, urinary L-FABP itself exhibited a very high sensitivity of 94.1% for detecting severe cases at a cut-off of 11.9, underscoring its utility as a first-line screening tool.

The integration of urinary L-FABP and the SOFA score provides superior accuracy for predicting COVID-19 severity, a finding strengthened by our large, real-world dataset. Our findings support a practical two-tiered triage strategy consisting of an initial, non-hospital screening with a urinary L-FABP test, which is painless, safe, and repeatable, where a level of 6 or more suggests hospital referral. Subsequently, in the hospital, a SOFA score ≥2 in these patients indicates a high likelihood of progression and need for hospitalization. This SOFA cut-off can be tailored to a center’s clinical priorities and resources. While these retrospective, single-center findings require external validation, this strategy may improve patient triage and optimize healthcare resource allocation. This abstract was also submitted for the 68th Annual Meeting of the Japanese Society of Nephrology.