Elevated SARS-CoV-2 in peripheral blood and increased COVID-19 severity in American Indians/Alaska Natives

Abstract

Epidemiological data across the United States show health disparities in COVID-19 infection, hospitalization, and mortality by race/ethnicity. While the association between elevated SARS-CoV-2 viral loads (VLs) (i.e. upper respiratory tract (URT) and peripheral blood (PB)) and increased COVID-19 severity has been reported, data remain largely unavailable for some disproportionately impacted racial/ethnic groups, particularly for American Indian or Alaska Native (AI/AN) populations. As such, we determined the relationship between SARS-CoV-2 VL dynamics and disease severity in a diverse cohort of hospitalized patients. Results presented here are for study participants (n = 94, ages 21-88 years) enrolled in a prospective observational study between May and October 2020 who had SARS-CoV-2 viral clades 20A, C, and G. Based on self-reported race/ethnicity and sample size distribution, the cohort was stratified into two groups: (AI/AN, n = 43) and all other races/ethnicities combined (non-AI/AN, n = 51). SARS-CoV-2 VLs were quantified in the URT and PB on days 0-3, 6, 9, and 14. The strongest predictor of severe COVID-19 in the study population was the mean VL in PB (OR = 3.34; P = 2.00 × 10^-4). The AI/AN group had the following: (1) comparable co-morbidities and admission laboratory values, yet more severe COVID-19 (OR = 4.81; P = 0.014); (2) a 2.1 longer duration of hospital stay (P = 0.023); and (3) higher initial and cumulative PB VLs during severe disease (P = 0.025). Moreover, self-reported race/ethnicity as AI/AN was the strongest predictor of elevated PB VLs (β = 1.08; P = 6.00 × 10^-4) and detection of SARS-CoV-2 in PB (hazard ratio = 3.58; P = 0.004). The findings presented here suggest a strong relationship between PB VL (magnitude and frequency) and severe COVID-19, particularly for the AI/AN group.

Keywords: COVID-19; SARS-CoV-2; hospitalization; intensive care; race/ethnicity; viral load.

Figure 1.

Factors associated with severe disease in hospitalized patients. AI/AN: American Indian/Alaska Native; Rx: treatment. Data presented as OR and 95% CI as determined by logistic regression modeling. The outcome variable was the development of severe disease across the 14-day sampling period. Co-variates in the model included race/ethnicity, sex, pregnancy status, age, days symptomatic prior to enrollment, BMI, treatment with remdesivir and/or steroids, and all co-morbidities that occurred in >5 patients. Statistical significance was set at P ⩽ 0.05. *Indicates statistical significance after the multiple test correction.

Figure 2.

Viral load dynamics in hospitalized patients stratified according to race/ethnicity, and disease severity. Viral loads (log10 copies/1000 cells) are presented as mean value and SD. Values are offset on each of the days to provide visual clarity. Significant differences between groups on each of the days were determined by the Wilcoxon rank-sum test. Cumulative viral loads (all) were determined by taking the mean value of each patient across all days, followed by comparison of the American Indian/Alaska Native (AI/AN) versus non-AI/AN groups with the Wilcoxon rank-sum test. (A) Upper respiratory tract and (B) peripheral blood viral levels for AI/AN (n = 43) and non-AI/AN (n = 51) patients. SARS-CoV-2 was detected in peripheral blood (⩾1 timepoint(s)) in 58.1% of the AI/AN and 33.3% of the non-AI/AN. (C) Upper respiratory tract viral loads in non-severe (AI/AN [n = 20] and non-AI/AN [n = 34]) and (D) severe (AI/AN [n = 23] and non-AI/AN [n = 17]) patients. (E) Peripheral blood viral loads in non-severe (AI/AN [n = 20] and non-AI/AN [n = 34]) and (F) severe (AI/AN [n = 23] and non-AI/AN [n = 17]) patients. SARS-CoV-2 was detected in peripheral blood (⩾1 timepoint(s)) in 25.0% of the AI/AN and 23.5% of the non-AI/AN with non-severe disease, whereas SARS-CoV-2 was detected in 77.0% of the AI/AN and 52.9% of the non-AI/AN with severe disease. *Indicates P ⩽ 0.05.

Figure 3.

Demographic, clinical, and virological predictors of severe disease and detection of SARS-CoV-2 in peripheral blood in hospitalized patients. (A) Data presented as OR and 95% CI as determined by logistic regression modeling. The outcome variable was development of severe disease across the 14-day sampling period. Co-variates in the model included race/ethnicity, sex, pregnancy status, age, days symptomatic prior to enrollment, BMI, treatment with remdesivir and/or steroids, all co-morbidities that occurred in >5 patients, and two SARS-CoV-2 metrics for both the upper respiratory tract and peripheral blood: mean viral load and frequency of detection. (B) Data are presented as hazard ratios (HR) and 95% CI as determined by a multiple-event-per-subject conditional proportional hazard model. An event was defined as detection of SARS-CoV-2 in blood (any level) and time-to-event was defined by the onset of symptoms prior to hospitalization until the day of measurement (right-censored). Co-variates in the model included race/ethnicity, sex, pregnancy status, age, days symptomatic prior to enrollment, BMI, treatment with remdesivir and/or steroids, and all co-morbidities that occurred in >5 patients. Statistical significance was set at P ⩽ 0.05. VL: viral load; CKD: chronic kidney disease; AI/AN: American Indian/Alaska Native; HTN: hypertension; RDV Rx: remdesivir treatment; BMI: body mass index; HLP: hyperlipidemia. *Indicates statistical significance after multiple test correction.