Increased levels of circulating cell-free DNA in COVID-19 patients with respiratory failure – Nature.com

Enrolled patients

We enrolled 345 patients with COVID-19 hospitalized at Showa University Hospital between April 2020 to April 2021. SARS-CoV-2 infections were confirmed by real-time reverse transcription polymerase chain reaction (RT-PCR) with samples obtained from nasopharyngeal swabs23,24. Patients with the following criteria were excluded: (i) inability to collect pretreatment serum, (ii) unknown outcomes due to transfer to other hospitals, (iii) age<18years, and (iv) pregnancy. After application of the exclusion criteria, 95 patients were included in the analysis.

Since SpO2 is a key clinical parameter for evaluating the severity of COVID-19 and respiratory failure, we first investigated the relationship between serum cfDNA (cf-mtDNA and cf-nDNA) levels and SpO2 in enrolled COVID-19 patients upon admission. Serum was isolated from the patients blood within 6h of admission. SpO2 was negatively correlated with cf-nDNA (R: 0.256, P=0.012; Fig.1a), whereas no significant association was found between SpO2 and cf-mtDNA (R=0.123, P=0.233; Fig.1b). Given that some patients received oxygen during the SpO2 measurement, we examined the relationship between the SpO2/ fraction of inspiratory oxygen (FiO2) ratio and cfDNA25,26. SpO2/FiO2 ratios negatively correlated with both circulating cf-nDNA (R=0.423, P<0.001; Fig.1c) and cf-mtDNA (R=-0.284, P=0.005; Fig.1d).

Correlation between cfDNA and blood oxygen levels. Blood was obtained from COVID-19 patients within 6h of admission, followed by serum isolation. Total cfDNA isolated from serum was analyzed by qPCR for the quantification of cf-nDNA and -mtDNA. Scatter plots show the correlation between SpO2 and (a) cf-nDNA, and (b) cf-mtDNA; between SpO2/FiO2 ratio and (c) cf-nDNA, and (d) cf-mtDNA.

We also investigated the relationship between circulating cfDNA levels and clinically used biomarkers measured upon admission. cf-nDNA levels were positively associated with circulating neutrophil counts, d-dimer, ferritin, lactate dehydrogenase (LDH), C-reactive protein (CRP), brain natriuretic peptide (BNP), and Krebs von den Lungen 6 (KL-6) levels (Supplementary Fig.1). Similarly, cf-mtDNA levels positively correlated with circulating neutrophil counts, d-dimer, LDH, CRP, and KL-6 levels (Supplementary Fig.2). However, cf-mtDNA levels did not significantly correlate with ferritin and BNP levels (Supplementary Fig.2).

Furthermore, we carried out multiple regression analyses to determine the clinical indicators independently associated with cf-nDNA and cf-mtDNA levels. Among the blood markers that were significantly correlated with cf-nDNA levels (Supplementary Fig.1), only neutrophil count remained independently associated (Supplementary Table 1). Similarly, only d-dimer levels were independently associated with cf-mtDNA levels (Supplementary Table 2).

Given the negative correlation between cfDNA and SpO2 upon admission, we examined the levels of circulating cfDNA based on the severity of COVID-19. The enrolled patients were categorized as moderate, severe, and critical based on the severity scoring system and classification from previous reports (Table 1)27,28,29. Moderate cases included patients with pneumonia who did not require OT; severe cases included those who required OT; and critical cases were individuals admitted to the intensive care unit requiring MV with or without extracorporeal membrane oxygenation (ECMO). Figure2 illustrates the study flow diagram, and Table 1 presents the characteristics of the study participants. Among the 95 patients, 41 (43.2%) received various types of OT, including oxygen supplementation via a nasal cannula, mask, or MV, during hospitalization, whereas 54 (56.8%; moderate group) did not receive any OT (Fig.2 and Table 1). Among those who received OT, 24 (25.3%; severe group) received OT without MV and 17 (17.9%; critical group) required MV. Within the critical group, ECMO was used in four patients (4.2%), and three patients (3.2%) died during hospitalization. The duration from admission to death for the three deceased patients was 29, 35, and 81days. Significant differences were observed among the three groups in terms of age (P<0.001), sex (P=0.048), smoking status (P=0.006), and frequency of comorbidities with COPD (P=0.018) and diabetes mellitus (DM) (P=0.005). The median age of the moderate group was significantly lower than that of the severe and critical groups (P=0.002 and P=0.003, respectively). The moderate group tended to have more females and never smokers, whereas the critical group showed a higher frequency of comorbidities with COPD and DM.

CONSORT diagram showing enrollment of COVID-19 patients, allocation, and outcomes.

We first compared serum cfDNA levels between COVID-19 patient groups. Notably, significant differences were observed among the three groups for both types of cfDNA (cf-nDNA, P=0.016; cf-mtDNA, P=0.012; Fig.3a and b). Critical COVID-19 cases (630.6 [440.51,463.8] copies/L) had higher cf-nDNA levels than moderate cases (349.4 [224.2618.7] copies/L) (P=0.004, Fig.3a). No significant differences in cf-nDNA levels were observed between moderate and severe cases (351.4 [282.4922.0] copies/L) as well as that between severe and critical cases (P=0.389 and P=0.088, respectively; Fig.3a). Similarly, the levels of cf-mtDNA in critical cases (1,073.4 [639.52,572.5] copies/L) were higher than those in moderate cases (543.5 [298.91,035.2] copies/L) (P<0.001, Fig.3b). No significant differences were observed in cf-mtDNA levels between moderate and severe cases (632.9 [298.91,805.9] copies/L) or between severe and critical cases (P=0.360 and P=0.070, respectively; Fig.3b). The cfDNA levels of critical cases (N=17) were further compared between the high positive end-expiratory pressure (PEEP) group (PEEP12 cmH2O) (N=9) and the low PEEP group (PEEP<12 cmH2O) (N=8). Three patients receiving ECMO were included in the high PEEP group. The results for cf-nDNAwere 1138.7 [469.71,463.9]copies/L in the high PEEP group and 561.7 [491.72,085.3]copies/L in the low PEEP group, with no significant difference between the two groups (P=0.958; Supplementary Fig.3a). For cf-mtDNA, the high PEEP group was1,228.6 [519.14,357.5] copies/L,while the low PEEP group was1,073.4 [1009.91405.5] copies/L. There was no significant difference in cf-mtDNA levels between the two groups (P=0.985; Supplementary Fig.3b).

Serum levels of cf-nDNA and cf-mtDNA in COVID-19 patients. cf-nDNA (a) and cf-mtDNA (b) were measured in COVID-19 patients, including those with moderate (N=54), severe (N=24), and critical disease (N=17). Data are presented as medians with boxes indicating upper and lower quartiles, whiskers indicating extremes, and with P values calculated by Wilcoxon Rank-Sum Test.

To investigate whether receiving OT accounted for the increase in circulating cfDNA in COVID-19 patients, all 95 patients were grouped into two categories; those who underwent OT and those who did not (non-OT) (Supplementary Table 3). The OT group was characterized by older age (P<0.001) and comprised a higher proportion of males (P=0.043) than the non-OT group. Significant differences in smoking status were also observed (P<0.001). There were 38 (70.4%) never smokers in the non-OT group and 15 (36.6%) in the OT group. Body mass index (BMI) was also generally higher in the OT group than in the non-OT group, although the difference was not statistically significant (P=0.058). Additionally, the OT group had more comorbidities, such as interstitial pneumonia and DM than the non-OT group (P=0.019 and P=0.001, respectively). The levels of cf-nDNA were significantly higher in the OT group as compared to the non-OT group (549.9 [289.01158.2] copies/L versus 349.4 [224.2618.7] copies/L) (P=0.026; Fig.4a). In the receiver operating characteristic (ROC) analysis, the area under the curve (AUC) of cf-nDNA for OT was 0.634 (P=0.005; Fig.4c), and an optimal cutoff value of 843.5 copies/L distinguished those in the OT group from those in the non-OT group with a sensitivity of 36.6% and specificity of 90.7%. Patients with a cf-nDNA level843.5 copies/L had increased odds of receiving OT (odds ratio [OR] 5.65, 95% confidence interval [CI] 1.8417.29, P=0.001). cf-mtDNA levels were significantly higher in the OT group than in the non-OT group (953.4 [429.31,873.4] copies/L versus 543.5 [298.91,035.2] copies/L) (P=0.021; Fig.4b). The AUC of cf-mtDNA for OT was 0.639 (P=0.029; Fig.4d), and a cutoff value of 945.2 copies/L distinguished those with OT from those without OT with a sensitivity of 53.7% and specificity of 74.1%. Patients with a cf-mtDNA level945.2 copies/L had increased odds of receiving OT (OR 3.31, 95% CI 1.397.85, P=0.006).

Elevated cf-nDNA and cf-mtDNA levels in COVID-19 patients subject to OT. cf-nDNA (a) and cf-mtDNA (b) were measured in COVID-19 patients including those with OT (N=41) and non-OT (N=54). Data are presented as medians with boxes indicating upper and lower quartiles, whiskers indicating extremes, and with P values calculated by Wilcoxon Rank-Sum Test. (c) ROC curves for cf-nDNA to predict OT. The AUC was 0.634 for OT with a cf-nDNA level of 843.5 copies/L. (d) ROC curves for cf-mtDNA to predict OT. The AUC was 0.639 for OT with a cf-mtDNA level of 945.2 copies/L.

Given that established clinical blood biomarkers such as LDH, d-dimer, and ferritin have been associated with disease severity in COVID-19 patients30,31,32,33,34, we tested whether these biomarkers, along with cfDNA, could predict the risk of receiving OT. For every 100 copies increase in cf-nDNA level, patients had increased odds of receiving OT after adjusting for age and sex (OR 1.11, 95% CI 1.001.24, P=0.033) (Table 2). Other markers, including neutrophil count, d-dimer, ferritin, LDH, and CRP levels were also significantly associated with OT after adjustment (Table 2). Notably, LDH was the most strongly associated with increased odds of receiving OT among the markers tested (OR 4.32, 95% CI 2.049.12, P<0.001) (Table 2). Although univariate analyses showed that both KL-6 and cf-mtDNA were associated with OT, these associations were no longer significant in the multivariable analysis after adjustment (Table 2).

To investigate whether receiving MV contributed to an increase in circulating cfDNA levels in COVID-19 patients, we categorized all 95 subjects into two groups: patients who received MV and those who did not (non-MV) (Supplementary Table 4). The MV group was characterized by older age (P=0.020) and a higher proportion of males (P=0.022) than the non-MV group. Significant differences in smoking status were evident between the MV and non-MV groups (P=0.044), with never smokers constituting 47 (60.2%) of the non-MV group and 6 (35.3%) in the MV group. Moreover, patients in the MV group had more comorbidities such as hypertension and COPD than those in the non-MV group (P=0.027 and P=0.005, respectively). The levels of cf-nDNA were significantly higher in patients who received MV than in the non-MV group (630.6 [440.51,463.8] copies/L versus 351.4 [230.4646.5] copies/L) (P=0.006; Fig.5a). In ROC analysis, the AUC of cf-nDNA for MV was 0.712 (P=0.002; Fig.5c), and an optimal cutoff value of 389.4 copies/L distinguished those with MV from those without MV with a sensitivity of 82.4% and specificity of 59%. Patients with a cf-nDNA level389.4 copies/L had increased odds of receiving MV (OR 6.36, 95% CI 1.6724.14, P=0.002). The levels of cf-mtDNA were significantly higher in patients with MV than in the non-MV group (1,073.4 [639.52,572.5] copies/L versus 560.8 [330.31,078.4] copies/L) (P=0.004; Fig.5b). The AUC of cf-mtDNA for MV was 0.720 (P=0.005, Fig.5d), and a cutoff value of 945.2 copies/L distinguished those with MV from those without MV with a sensitivity of 70.6% and specificity of 69.2%. Patients with a cf-mtDNA level945.2 copies/L had increased odds of receiving MV (OR 6.87, 95% CI 2.0123.4, P<0.001).

Elevated cf-nDNA and cf-mtDNA levels in COVID-19 patients subject to MV. cf-nDNA (a) and cf-mtDNA (b) were measured in COVID-19 patients including those with MV (N=17) and non-MV (N=78). Data are presented as medians with boxes indicating upper and lower quartiles, whiskers indicating extremes, and with P values calculated by Wilcoxon Rank-Sum Test. (c) ROC curves for cf-nDNA to predict MV. The AUC was 0.712 for MV with a cf-nDNA level of 389.4 copies/L. (d) ROC curves for cf-mtDNA to predict MV. The AUC was 0.720 for MV with a cf-mtDNA level of 945.2 copies/L.

We also assessed the association between cfDNA levels and clinically established measures with MV. For every 100 copies increase in cf-nDNA level, patients displayed increased odds of receiving MV after adjusting for age and sex (OR 1.14, 95% CI 1.031.27, P=0.008). Similarly, for every 100 copies increase in cf-mtDNA level, patients had increased odds of receiving MV after adjustment (OR 1.06, 95% CI 1.011.12, P=0.008) (Table 3). Similar to the observed prediction for OT, neutrophil count, d-dimer, ferritin, LDH, CRP, and KL-6 levels were significantly associated with MV after adjustment. Notably, among the tested markers, LDH had the highest predictive value (OR, 2.38) for MV (Table 3).

Read more from the original source:

Increased levels of circulating cell-free DNA in COVID-19 patients with respiratory failure - Nature.com