A CXCL10-enriched hepatic-risk phenotype reveals a cell-count paradox in severe COVID-19

Background Cell-count–derived inflammatory indices, including the neutrophil-to-lymphocyte ratio (NLR), platelet-to-lymphocyte ratio (PLR), and systemic immune-inflammation index (SII), are widely used to assess COVID-19 severity, but hepatic dysfunction and thrombocytopenia may alter their prognostic meaning. We tested whether FIB-4-defined hepatic risk modifies the interpretation of cell-count inflammation and whether this discordance has a cytokine or chemokine correlate. Methods We applied an a priori 2 × 2 framework to 575 hospitalized adults with COVID-19 undergoing CT pulmonary angiography, classified by cardiometabolic burden and FIB-4-defined hepatic risk. Associations with oxygen requirement, ICU admission, in-hospital death, composite ICU/death and prolonged hospitalization were analyzed. A separate non-overlapping cytokine cohort of 90 patients was used for cytokine/chemokine profiling. Results Hepatic-risk phenotypes had higher adjusted odds of adverse outcomes but paradoxically lower NLR, PLR, SII and derived NLR. Significant index × hepatic-risk interactions showed that NLR, SII and dNLR retained their expected risk gradients in non-hepatic-risk patients but were attenuated or absent in hepatic-risk patients. In the cytokine cohort, the cellular pattern was reproduced, CXCL10/IP-10 and IL-8/CXCL8 were selectively elevated, and hepatic-risk patients were enriched in the NLR-low/CXCL10-high quadrant. Cell-normalized CXCL10 metrics and a CXCL10–NLR discordance score supported disproportionate chemokine enrichment relative to suppressed peripheral cell-count readouts. Conclusion FIB-4-defined hepatic-risk COVID-19 identifies a CXCL10/IP-10-enriched, peripherally cell-count-suppressed inflammatory phenotype. Conventional NLR-, PLR-, SII- and dNLR-based risk interpretation may underestimate clinical risk when hepatic risk is present.