Post by Nadica (She/Her) on Jul 19, 2024 1:24:10 GMT
Blood transcriptomic analyses reveal persistent SARS-CoV-2 RNA and candidate biomarkers in post-COVID-19 condition - Published April 24, 2024
With an estimated 65 million individuals affected by post-COVID-19 condition (also known as long COVID),1 non-invasive biomarkers are direly needed to guide clinical management. To address this pressing need, we used blood transcriptomics in a general practice-based case-control study. Individuals with long COVID were diagnosed according to WHO criteria, and validated clinical scales were used to quantify patient-reported outcomes.2 Whole blood samples were collected from 48 individuals with long COVID and 12 control individuals matched for age, sex, time since acute COVID-19, severity, vaccination status, and comorbidities (appendix 1 p 2). Digital transcriptomic analysis was performed using the nCounter (Nanostring Technologies, Seattle, WA, USA) platform, as described for critical COVID-19.3 Consequently, 212 genes were identified to be differentially expressed between individuals with long COVID and controls (figure A), of which 70 remained significant after adjustment for false discovery rate correction (appendix 1). Several viral RNAs were upregulated: nucleocapsid, ORF7a, ORF3a, Mpro (a nirmatrelvir plus ritonavir [Paxlovid] target), and antisense ORF1ab RNA. Specifically, the upregulation of antisense ORF1ab RNA suggests ongoing viral replication. SARS-CoV-2-related host RNAs (ACE2/TMPRSS2 receptors, DPP4/FURIN proteases) and RNAs prototypical for memory B-cells and platelets4 were also upregulated (figure A). Multivariable logistic regression identified antisense SARS-CoV-2 and FYN RNA concentrations as independent predictors of long COVID (corrected for age and sex; appendix 1 p 2). Receiver operating characteristic curve analysis showed significant discrimination (area under curve [AUC] 0·94, 95% CI 0·86–1·00) between individuals with long COVID (n=48) and controls (n=12), with 93·8% sensitivity and 91·7% specificity (figure B). Single biomarkers antisense SARS-CoV-2 (AUC 0·78, 0·65–0·90) and FYN RNA (AUC 0·89, 0·79–0·99) were significant predictors with lower sensitivity (52·1% and 72·9%, respectively) but similar specificity (91·7% and 100%, respectively; figure B). Upon summarising transcriptomic results into biological pathways, we found significantly decreased immunometabolism in individuals with long COVID, which was negatively correlated with the blood viral load (appendix 1 p 3). A qualitative analysis of individual SARS-CoV-2 transcript positivity revealed significant differences between individuals with long COVID and controls for antisense (65% vs 25%), ORF7a (60% vs 25%), and nucleocapsid (50% vs 8%) RNAs (figure C). Similarly, the SARS-CoV-2 transcript positivity with respect to the total blood viral load was also significantly different (60% vs 8%). By use of multivariable logistic regression, we found that age and sex were not associated with the distinction between a low and high viral RNA load status. Conversely, the number of comorbidities (odds ratio [OR] 1·61, 95% CI 1·14–2·49) and COVID vaccine doses (OR 0·36, 0·14–0·79) emerged as independent predictors of distinguishing between low and high viral RNA load status (appendix 2). We found that viral and immune parameters, such as the antisense Orf1ab RNA concentrations and immunometabolism score, were also linked to the patient-reported anxiety or depression score. Individuals classified as having severe anxiety or depression (with a score of 4 and 5) displayed significantly higher antisense RNA concentrations and lower immunometabolism scores (p<0·05) than those categorised as mild (with scores of 1–3; figure D). In conclusion, the associations among persistent viral RNA, immunometabolism, and patient-reported outcomes provide mechanistic insights for addressing the challenges posed by long COVID.
With an estimated 65 million individuals affected by post-COVID-19 condition (also known as long COVID),1 non-invasive biomarkers are direly needed to guide clinical management. To address this pressing need, we used blood transcriptomics in a general practice-based case-control study. Individuals with long COVID were diagnosed according to WHO criteria, and validated clinical scales were used to quantify patient-reported outcomes.2 Whole blood samples were collected from 48 individuals with long COVID and 12 control individuals matched for age, sex, time since acute COVID-19, severity, vaccination status, and comorbidities (appendix 1 p 2). Digital transcriptomic analysis was performed using the nCounter (Nanostring Technologies, Seattle, WA, USA) platform, as described for critical COVID-19.3 Consequently, 212 genes were identified to be differentially expressed between individuals with long COVID and controls (figure A), of which 70 remained significant after adjustment for false discovery rate correction (appendix 1). Several viral RNAs were upregulated: nucleocapsid, ORF7a, ORF3a, Mpro (a nirmatrelvir plus ritonavir [Paxlovid] target), and antisense ORF1ab RNA. Specifically, the upregulation of antisense ORF1ab RNA suggests ongoing viral replication. SARS-CoV-2-related host RNAs (ACE2/TMPRSS2 receptors, DPP4/FURIN proteases) and RNAs prototypical for memory B-cells and platelets4 were also upregulated (figure A). Multivariable logistic regression identified antisense SARS-CoV-2 and FYN RNA concentrations as independent predictors of long COVID (corrected for age and sex; appendix 1 p 2). Receiver operating characteristic curve analysis showed significant discrimination (area under curve [AUC] 0·94, 95% CI 0·86–1·00) between individuals with long COVID (n=48) and controls (n=12), with 93·8% sensitivity and 91·7% specificity (figure B). Single biomarkers antisense SARS-CoV-2 (AUC 0·78, 0·65–0·90) and FYN RNA (AUC 0·89, 0·79–0·99) were significant predictors with lower sensitivity (52·1% and 72·9%, respectively) but similar specificity (91·7% and 100%, respectively; figure B). Upon summarising transcriptomic results into biological pathways, we found significantly decreased immunometabolism in individuals with long COVID, which was negatively correlated with the blood viral load (appendix 1 p 3). A qualitative analysis of individual SARS-CoV-2 transcript positivity revealed significant differences between individuals with long COVID and controls for antisense (65% vs 25%), ORF7a (60% vs 25%), and nucleocapsid (50% vs 8%) RNAs (figure C). Similarly, the SARS-CoV-2 transcript positivity with respect to the total blood viral load was also significantly different (60% vs 8%). By use of multivariable logistic regression, we found that age and sex were not associated with the distinction between a low and high viral RNA load status. Conversely, the number of comorbidities (odds ratio [OR] 1·61, 95% CI 1·14–2·49) and COVID vaccine doses (OR 0·36, 0·14–0·79) emerged as independent predictors of distinguishing between low and high viral RNA load status (appendix 2). We found that viral and immune parameters, such as the antisense Orf1ab RNA concentrations and immunometabolism score, were also linked to the patient-reported anxiety or depression score. Individuals classified as having severe anxiety or depression (with a score of 4 and 5) displayed significantly higher antisense RNA concentrations and lower immunometabolism scores (p<0·05) than those categorised as mild (with scores of 1–3; figure D). In conclusion, the associations among persistent viral RNA, immunometabolism, and patient-reported outcomes provide mechanistic insights for addressing the challenges posed by long COVID.