Evaluate the use of neurological biomarkers to predict discharge outcomes in COVID-19 patients.  

Altered levels of brain-derived molecular biomarkers in patients with non-neurological critical illness associates with worse outcomes following these systemic insults. We hypothesized that COVID-19 critical illness would increase expression of brain-derived biomarkers and portend worse outcomes.

38 adults admitted for COVID-19 at a single tertiary care medical center were prospectively enrolled (M_age = 63.63 ±19.51, 53% female, 71% requiring ICU admission) and clinical information collected including discharge disposition to home/rehabilitation (n=18) or expired/skilled nursing facility (SNF; n=20). Plasma GFAP, Tau, NfL, and UCHL1 were measured by digital ELISA.

COVID-19 patients admitted to the ICU exhibited significantly higher levels of NfL (p=0.003,d=1.25) and GFAP (p=0.03,d=0.88).

We used binary logistic regression to determine if biomarkers predicted discharge outcome. Models were examined for best fit using biomarker level, age, ICU status, and history of prior neurological disease. A model including NfL level (Wald’s χ² =6.614, p=0.010, OR=1.043, 95%CI (1.010, 1.076)) predicting disposition was significant (χ² = 22.247, p<0.001, Nagelkerke R² = .591). The model’s prediction success was 84.2% (90.0% for home/rehab and 77.8% for SNF/expired)

A model including GFAP level (Wald’s χ² =3.055, p=0.080, OR=1.003, 95%CI (1.000, 1.007)) and ICU status (Wald’s χ² =4.073, p=0.044, OR=0.096, 95%CI (0.010, 0.935)) on disposition was also significant (χ² = 17.377, p<0.01, Nagelkerke R²= .490). The model successfully predicted disposition status at 78.9% (85% for home/rehab and 72.2% for SNF/expired). Adding age, ICU status, or prior neurological history did not improve outcome prediction.

COVID-19 patients requiring ICU admission exhibit increases in circulating brain-derived proteins. Higher levels of GFAP and NfL is associated with worse discharge outcomes, even after controlling for age, ICU status and prior neurological disease. Future work examining COVID-19 recovery will help determine if these biomarkers are predictive of long-term neurological consequences.