COVID-19 Subphenotypes on Admission Are Associated with Mortality and a Differential Response to Therapeutics

Abstract

Rationale: The novel coronavirus disease 2019 (COVID-19) is a clinical syndrome caused by the SARS-CoV-2 virus and has resulted in widespread morbidity and mortality. Current treatment algorithms assess severity of illness and assign therapies based on oxygen requirement. We hypothesize that COVID-19 is a heterogenous syndrome and subphenotypes on admission may vary in both clinical course and response to therapy. Methods: We included all patients aged 18 years or older, admitted with laboratory-confirmed COVID-19 to the Mount Sinai Health System hospitals between March 1, and August 30, 2020. Demographic data, medical comorbidities, and clinical and laboratory data within 24 hours of the first COVID-19 admission were collected. The primary outcome, determined across all COVID-19 admissions, was mortality;secondary outcomes included intubation, intensive care unit (ICU) admission and length of stay. We employed latent class analysis to admission variables without consideration of clinical outcomes to determine the optimal number of COVID-19 subphenotypes and then assigned each patient to a subphenotype class. We then employed linear mixed effect models to determine if admission COVID-19 subphenotypes were associated with mortality, mechanical ventilation, ICU admission and length of stay. We then explored whether subphenotypes demonstrated a differential response to treatment with convalescent plasma and Tocilizumab. Results: A total of 4620 patients were included. These patients had a median age of 67 years (IQR 55-78), 56.9% were male, and 76.7% had 0 or 1 comorbid conditions. Latent class analysis identified that a six-subphenotype model was the best fit, based on cAIC, aBIC, entropy, and likelihood ratio. Each subphenotype was clinically distinct. Review of admission data identified a predominantly female, hyperinflammatory subphenotype;a renal failure subphenotype;a young, less hypoxic subphenotype;an older, more coagulopathic subphenotype;a younger, less coagulopathic subphenotype;and a multiorgan dysfunction subphenotype. As compared to the young, less hypoxic subphenotype, all subphenotypes had increased odds of mortality (Odds Ratio (OR) range from 3.2 to 51.6), intubation (OR range from 3.7 to 14.3) and ICU admission (OR range from 2.1 to 5.0) (Table 1). Exploratory analyses found significantly different mortality across subphenotypes treated by convalescent plasma and Tocilizumab. Conclusions: Six clinically distinct COVID-19 subphenotypes on admission were identified and were associated with varying odds of mortality, intubation and ICU admission. Exploratory analyses suggest that subphenotypes may differentially respond to treatment which may have implications for treatment algorithms.