Biomarkers trajectories in critically ill patients with COVID-19 acute respiratory distress syndrome: insights from latent class growth analysis

Background: Acute respiratory distress syndrome (ARDS) is a common condition requiring intensive care, with limited effective treatments due to its clinical and biological heterogeneity. Efforts in critical care medicine have identified ARDS sub-phenotypes, hyperinflammatory and hypoinflammatory, which suggest underlying heterogeneity in patient responses. Studies have detected a low prevalence of the hyperinflammatory phenotype in ARDS patients related to Coronavirus Disease 2019 (COVID-19). To identify targeted therapeutic interventions, this study applied growth mixture models to longitudinal biomarker data to determine specific latent trajectory groups. Methods: This is a secondary analysis of a cohort study on patients with COVID-19 pneumonia admitted to an Italian intensive care unit (ICU) with at least two assessments of inflammatory marker levels within 28 days of admission. Plasma levels of interleukin 6 (IL-6), interleukin-8 (IL-8), soluble tumour necrosis factor receptor 1 (sTNFR-1), intercellular adhesion molecule 1 (ICAM-1), soluble receptor for advanced glycation end products (sRAGE) and angiopoietin-2 (Ang2) were assessed. Results: Fifty-eight patients were analysed, with a total of 201 level assessments performed. None showed hyperinflammatory phenotype within 48 h of ICU admission. Latent class growth analysis identified distinct trajectories for sTNFR-1, ICAM-1, and sRAGE, with sTNFR-1 class 1 including 39.7% of patients showing elevated levels escalating during ICU stay. Compared to sTNFR-1 class 2, class 1 patients were older (63.8 ± 14.7 versus 58.4 ± 12.7, P = 0.0486), had higher baseline inflammatory marker levels (IL-6, IL-8, sRAGE, and sTNFR-1), higher proportions of prone positioning and continuous renal replacement therapy utilization, and a higher mortality rate (56.5% versus 28.6%, p = 0.0333). Conclusions: In our cohort of critically ill patients with severe COVID-19, we identified distinct latent trajectories based on longitudinal data of inflammatory markers. sTNFR-1 levels identified a subgroup with a hypoinflammatory phenotype, demonstrating a mortality rate comparable to that typically observed in ARDS hyperinflammatory phenotype. These findings point out the criticality of delineating distinct patient subgroups within the context of ARDS and COVID-19, enhancing clinical management and optimizing patient prognosis.