ABSTRACT
Introduction: Due to variability in the host response, a uniform treatment strategy for severe COVID-19 may be inadequate. We applied unsupervised clustering methods to large cohorts of COVID-19 ICU patients to derive and validate clinical phenotypes, and to explore treatment responses in these phenotypes. Method(s): Phenotypes were derived in 13.279 critically ill COVID-19 patients admitted to 82 Dutch ICUs from September 2020 to February 2022. Twenty-one features were selected from clinical characteristics measured within 24 h after ICU admission. Phenotypes were assigned using consensus k means clustering. External validation was performed in 6225 critically ill COVID-19 patients admitted to 55 Spanish ICUs from February 2020 to December 2021. Individual patient data on corticosteroids therapy enabled us to investigate phenotype-specific responses in this cohort. Result(s): Three distinct clinical phenotypes were derived (Fig. 1A). Patients with phenotype 1 (43%) were younger, had lower APACHE IV scores, higher BMI as well as a lower P/F ratio and 90-day in-hospital mortality (18%, Fig. 1A). Phenotype 2 patients (37%) were older and had slightly higher APACHE IV scores compared with phenotype 1, a lower BMI, and higher mortality compared to phenotype 1 (24%, p = 2.95e-07). Phenotype 3 (20%) included the oldest patients with the most comorbidities and highest APACHE IV scores, severe renal and metabolic impairment, and the worst outcome (47% mortality, p = 6.6e-16 and p = 6.6e-16 versus phenotypes 1 and 2, respectively). Phenotype distribution and outcome were very similar in the validation cohort (Fig. 1B). This cohort also revealed that corticosteroid therapy only benefited phenotype 3 (65% vs. 54% mortality, p = 2.5e-03, Fig. 1C). Conclusion(s): COVID-19 ICU phenotypes based on clinical data are related to outcome and treatment responses. This can inform treatment decisions as well as randomized trials employing precision medicine approaches.
ABSTRACT
Introduction: A subgroup of critically ill COVID-19 patients develops pulmonary fibroproliferation (PF), which is associated with worse outcomes. We explored the kinetics of fibrosis markers and ventilatory parameters prior to and following use of steroids to treat suspected PF. Furthermore, we investigated the effects of early dexamethasone (DEXA) treatment, the current standard-of-care for COVID-19, on the incidence and time to development of PF and clinical outcomes. Methods: We included 191 critically ill COVID-19 patients spanning two treatment cohorts: no DEXA treatment (pre-DEXA cohort, n = 67) and dexamethasone treatment as standard-of-care (DEXA cohort, n = 124). Kinetics of circulating fibrosis markers and ventilatory parameters were analyzed in suspected PF patients prior to and following initiation of steroid therapy as well as in patients in whom PF was not suspected. Furthermore, associations between PF and clinical outcomes were explored. Results: Patients with suspected PF exhibited higher circulating fibrosis markers, lower lung compliance and PaO2/FiO2 ratios, and increased dead space ventilation. Incidence of suspected PF was 28% in the pre-DEXA cohort and 25% in the DEXA cohort (p = 0.61), and time to development of suspected PF was also similar between cohorts (16 [12-21] vs. 19 [14-23] days from ICU admission, p = 0.11). Time on ventilator, LOS in ICU and mortality were significantly higher in suspected PF patients than in no suspected PF patients, with no differences between the cohorts (Fig. 1). Conclusions: Increased circulating fibrosis markers reflect development of PF in critically ill COVID-19 patients, which is associated with prolonged ICU length of stay and high mortality rates. Introduction of dexamethasone as standard-of-care is not associated with altered incidence of PF or improved clinical outcomes in patients with PF. (Figure Presented).