Novel Outcome Biomarkers Identified With Targeted Proteomic Analyses of Plasma From Critically Ill Coronavirus Disease 2019 Patients

Abstract

Objectives: Coronavirus disease 2019 patients admitted to the ICU have high mortality. The host response to coronavirus disease 2019 has only been partially elucidated, and prognostic biomarkers have not been identified. We performed targeted proteomics on critically ill coronavirus disease 2019 patients to better understand their pathophysiologic mediators and to identify potential outcome markers.

Design: Blood was collected at predetermined ICU days for proximity extension assays to determine the plasma concentrations of 1,161 proteins.

Setting: Tertiary care ICU and academic laboratory.

Subjects: All patients admitted to the ICU suspected of being infected with severe acute respiratory syndrome coronavirus 2, using standardized hospital screening methodologies, had blood samples collected until either testing was confirmed negative on ICU day 3 (coronavirus disease 2019 negative) or until ICU day 10 if the patient positive (coronavirus disease 2019 positive).

Interventions: None.

Measurements and main results: Age- and sex-matched healthy control subjects and ICU patients who were either coronavirus disease 2019 positive or coronavirus disease 2019 negative were enrolled. Cohorts were well-balanced with the exception that coronavirus disease 2019 positive patients suffered bilateral pneumonia more frequently than coronavirus disease 2019 negative patients. Mortality rate for coronavirus disease 2019 positive ICU patients was 40%. Feature selection identified the top performing proteins for identifying coronavirus disease 2019 positive ICU patients from both healthy control subjects and coronavirus disease 2019 negative ICU patients (classification accuracies 100%). The coronavirus disease 2019 proteome was dominated by interleukins and chemokines, as well as several membrane receptors linked to lymphocyte-associated microparticles and/or cell debris. Mortality was predicted for coronavirus disease 2019 positive patients based on plasma proteome profiling on both ICU day 1 (accuracy 92%) and ICU day 3 (accuracy 83%). Promising prognostic proteins were then narrowed down to six, each of which provided excellent classification performance for mortality when measured on ICU day 1 CMRF-35-like molecule, interleukin receptor-12 subunit B1, cluster of differentiation 83 [CD83], family with sequence similarity 3, insulin-like growth factor 1 receptor and opticin; area-under-the-curve =1.0; p = 0.007).

Conclusions: Targeted proteomics with feature classification easily distinguished both healthy control subjects and coronavirus disease 2019 tested negative ICU patients from coronavirus disease 2019 tested positive ICU patients. Multiple proteins were identified that accurately predicted coronavirus disease 2019 tested positive patient mortality.

Keywords: biomarkers; coronavirus disease 2019; host response; inflammation; intensive care unit.

Figure 1.

Targeted proteomics accurately differentiates coronavirus disease 2019 positive (COVID19+) ICU patients from healthy control subjects. A, Subjects plotted in 2D following dimensionality reduction of their respective proteomes by stochastic neighbor embedding. Yellow dots represent healthy control subjects, whereas purple dots represent age- and sex-matched COVID19+ ICU patients (ICU day 1 plasma). The dimensionality reduction shows that based on the plasma proteome, the two cohorts are distinct and easily separable (100% classification accuracy). The axes are dimension less. B, Feature classification demonstrating the top 20 plasma proteins that classify coronavirus disease 2019 status with their % association.

Figure 2.

Targeted proteomics accurately differentiates coronavirus disease 2019 positive (COVID19+) ICU patients from coronavirus disease 2019 negative (COVID19–) ICU patients. A, Subjects plotted in 2D following dimensionality reduction of their respective proteomes by stochastic neighbor embedding. Yellow dots represent COVID19+ ICU patients, whereas purple dots represent age- and sex-matched COVID19– ICU patients (ICU day 1 plasma for both populations). The dimensionality reduction shows that based on the plasma proteome, the two cohorts are distinct and easily separable (100% classification accuracy). The axes are dimension less. B, Feature classification demonstrating the top 20 plasma proteins that classify coronavirus disease 2019 status with their % association.

Figure 3.

Targeted proteomics accurately differentiates coronavirus disease 2019 positive (COVID19+) patients that lived or died on ICU days 1 and 3. A, COVID19+ ICU patients plotted in 2D following dimensionality reduction of their respective outcomes (alive or dead) by stochastic neighbor embedding. Green dots represent COVID19+ ICU patients who survived, whereas red dots represent COVID19+ ICU patients who died (ICU day 1 plasma for both populations). The dimensionality reduction shows that based on the plasma proteome, the two cohorts are distinct and easily separable (92% classification accuracy). The axes are dimension less. B, COVID19+ ICU patients plotted in 2D following dimensionality reduction of their respective outcomes (alive or dead) by stochastic neighbor embedding (ICU day 3 plasma for both populations). Green dots represent COVID19+ ICU patients who survived, whereas red dots represent COVID19+ ICU patients who died. The dimensionality reduction shows that based on the plasma proteome, the two cohorts are reasonably distinct and separable (83% classification accuracy). The axes are dimension less. C, Feature classification demonstrating the top 21 plasma proteins obtained on ICU day 1 that classify outcome for COVID19+ ICU patients with their % association. D, Feature classification demonstrating the top 21 plasma proteins obtained on ICU day 3 that classify outcome for COVID19+ ICU patients with their % association.

Figure 4.

Time course for the top six plasma proteins that predicted coronavirus disease 2019 (COVID19) outcome. Green lines represent six coronavirus disease 2019 positive (COVID19+) ICU patients who survived (one patient was discharged by ICU day 7), whereas red lines represent four COVID19+ ICU patients who died. Receiver operating characteristic analyses for all six proteins measured on ICU day 1 had area-under-the-curves of 1.0 (p = 0.007), indicating excellent classification performance. Their respective cutoff values were: CMRF-35-like molecule (CLM-1, 7.8), interleukin receptor-12 subunit B1 (IL12RB1, 3.3), cluster of differentiation 83 (CD83, 3.3), family with sequence similarity 3 (FAM3B, 4.7), insulin-like growth factor 1 receptor (IGF1R, 3.8) and opticin (OPTC, 3.6).