key: cord-0684766-cn4412ys
authors: Chen, P. Z.; Bobrovitz, N.; Premji, Z.; Koopmans, M.; Fisman, D. N.; Gu, F. X.
title: SARS-CoV-2 Shedding Dynamics Across the Respiratory Tract, Sex, and Disease Severity for Adult and Pediatric COVID-19
date: 2021-02-19
journal: nan
DOI: 10.1101/2021.02.17.21251926
sha: c98c7acfa1f31debe1ee27d4a1630dbfe6500613
doc_id: 684766
cord_uid: cn4412ys

Background: SARS-CoV-2 shedding dynamics in the upper (URT) and lower respiratory tract (LRT) remain unclear. Objective: To analyze SARS-CoV-2 shedding dynamics across COVID-19 severity, the respiratory tract, sex and age cohorts (aged 0 to 17 years, 18 to 59 years, and 60 years or older). Design: Systematic review and pooled analyses. Setting: MEDLINE, EMBASE, CENTRAL, Web of Science Core Collection, medRxiv and bioRxiv were searched up to 20 November 2020. Participants: The systematic dataset included 1,266 adults and 136 children with COVID-19. Measurements: Case characteristics (COVID-19 severity, age and sex) and quantitative respiratory viral loads (rVLs). Results: In the URT, adults with severe COVID-19 had higher rVLs at 1 DFSO than adults (P = 0.005) or children (P = 0.017) with nonsevere illness. Between 1-10 DFSO, severe adults had comparable rates of SARS-CoV-2 clearance from the URT as nonsevere adults (P = 0.479) and nonsevere children (P = 0.863). In the LRT, severe adults showed higher post-symptom-onset rVLs than nonsevere adults (P = 0.006). In the analyzed period (4-10 DFSO), severely affected adults had no significant trend in SARS-CoV-2 clearance from LRT (P = 0.105), whereas nonsevere adults showed a clear trend (P < 0.001). After stratifying for disease severity, sex and age (including child vs. adult) were not predictive of the duration of respiratory shedding. Limitation: Limited data on case comorbidities and few samples in some cohorts. Conclusion: High, persistent LRT shedding of SARS-CoV-2 characterized severe COVID-19 in adults. After symptom onset, severe cases tended to have higher URT shedding than their nonsevere counterparts. Disease severity, rather than age or sex, predicted SARS-CoV-2 kinetics. LRT specimens should more accurately prognosticate COVID-19 severity than URT specimens. Primary Funding Source: Natural Sciences and Engineering Research Council.

Our systematic review identified studies reporting SARS-CoV-2 quantitation in respiratory 75 specimens taken during the estimated infectious period (-3 to 10 days from symptom onset 76

[DFSO]) (15, 18). The systematic review protocol was based on our previous study (19) and was 77 prospectively registered on PROSPERO (registration number, CRD42020204637). The 78 systematic review was conducted according to Cochrane methods guidance (20) . Other than the 79 title of this study, we have followed PRISMA reporting guidelines (21). 80

For analyses based on rVL (viral RNA concentration in the respiratory tract) and to account 120 for interstudy variation in the volumes of viral transport media (VTM) used, the rVL for each 121 collected sample was estimated based on the specimen concentration (viral RNA concentration 122 in the specimen) and dilution factor in VTM. Typically, swabbed specimens (NPS and OPS) 123

report the viral RNA concentration in VTM. Based on the VTM volume reported in the study 124 along with the expected uptake volume for swabs (0.128 ± 0.031 ml, mean ± SD) (24), we 125 calculated the dilution factor for each respiratory specimen and then estimated the rVL. 126

Similarly, liquid specimens (ETA, POS and Spu) are often diluted in VTM, and the rVL was 127 estimated based on the reported collection and VTM volumes. If the diluent volume was not 128 reported, then VTM volumes of 1 ml (NPS and OPS) or 2 ml (POS and ETA) were assumed (23, effect on the intercept (regression t-test for " ). Shedding dynamics were compared between 165 cohorts by interaction (regression t-test for # ). The statistical significance of viral clearance for 166 each cohort was analyzed using simple linear regression (regression t-test on the slope). 167

Regression models were extrapolated (to 0 log10 copies/ml, rather than an assay detection limit) 168

to estimate the duration of shedding. 169

To assess heterogeneity in shedding, rVL data were fitted to Weibull distributions (19), and 170 the rVL at a case percentile was estimated using the Weibull quantile function. Each cohort in 171 statistical analyses included all rVLs for which the relevant characteristic (LRT or URT, age 172 cohort, sex or disease severity) was ascertained at the individual level. Cohorts with small 173 sample sizes were not compared, as these analyses are more sensitive to potential sampling error. 174

Statistical analyses were performed using OriginPro 2019b (OriginLab) and the General Linear 175 Figure) . For pediatric cases, the search found only nonsevere 189 infections and URT specimen measurements. Appendix Table 1 summarizes the characteristics  190 of contributing studies, of which 18 had low risk of bias according to the modified JBI critical 191 appraisal checklist. Studies at high or unclear risk of bias typically included samples that were 192 not representative of the target population; did not report the VTM volume used; had non-193 consecutive inclusion for case series and cohort studies or did not use probability-based sampling 194 for cross-sectional studies; and did not report the response rate (Appendix Table 2 While regression analysis compared mean shedding levels and dynamics, we fitted rVLs to 205

Weibull distributions to assess heterogeneity in shedding. Both severe and nonsevere adult 206 COVID-19 showed comparably broad heterogeneity in URT shedding throughout disease course 207 ( Figure 2B) . For severe disease, the standard deviation (SD) of rVL was 1.86, 2.34, 1.89 and 208 1.90 log10 copies/ml at 2, 4, 7 and 10 DFSO, respectively. For nonsevere illness, these SDs were 209 2.08, 1.90, 1.89 and 1.96 log10 copies/ml, respectively. 210 . CC-BY-NC 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)

The copyright holder for this preprint this version posted February 19, 2021. symptom onset. Based on distribution fitting (Figure 2B) , at 2 DFSO, the estimated rVL at the 212 80 th case percentile (cp) for severe disease was 9.54 (95% CI, 8.78-10.4) log10 copies/ml, while it 213 was 8.84 (95% CI, 8.49-9.20) log10 copies/ml for nonsevere illness. By 10 DFSO, this difference 214 reduced: the 80 th -cp estimates were 6.86 (95% CI, 6.20-7.59) and 6.45 (95% CI, 5.91-7.04) log10 215 copies/ml for severe and nonsevere disease, respectively. 216

After stratifying adults for disease severity, our analyses showed nonsignificant differences 217 in URT shedding based on sex and age. For nonsevere illness, male and female cases had no 218 significant difference in mean rVL at 1 DFSO (P for intercept = 0.085) or rate of viral clearance 219 (P for interaction = 0.644) ( Figure 2C) . Similarly, for severe disease, male and female cases had 220 comparable mean rVLs at 1 DFSO (P for intercept = 0.326) and URT dynamics (P for 221 interaction = 0.280) ( Figure 2D ). For nonsevere illness, younger and older adults had no 222 significant difference in URT shedding levels at 1 DFSO (P for intercept = 0.294) or post-223 symptom-onset dynamics (P for interaction = 0.100) ( Figure 2E ). For severe disease, the adult 224 age cohorts showed similar mean rVLs at 1 DFSO (P for intercept = 0.915) and rates of viral 225 clearance (P for interaction = 0.359) ( Figure 2F) . 226

Our analyses showed that high, persistent LRT shedding of SARS-CoV-2 was associated 229 with severe COVID-19, but not nonsevere illness, in adults ( Figure 3A) . At the initial day in our 230 analyzed period (4 DFSO), the mean rVL in the LRT of severe cases (8.42 [95% CI, 7.67-9.17] 231 log10 copies/ml) was significantly greater (P for intercept = 0.006) than that of nonsevere cases 232 (6.82 [95% CI, 5.95-7.69] log10 copies/ml). Between severities, the difference in LRT clearance 233 . CC-BY-NC 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted February 19, 2021. ; https://doi.org/10.1101/2021.02.17.21251926 doi: medRxiv preprint rates was marginally above the threshold for statistical significance (P for interaction = 0.053). 234 However, severe cases had persistent LRT shedding, with no significant trend in SARS-CoV-2 235 clearance in the analyzed period (-0.14 [95% CI, -0.32 to 0.030] log10 copies/ml day -1 , P = 236 0.105), whereas nonsevere cases rapidly cleared the virus from the LRT (-0.41 [95% CI, -0.64 to 237 -0.19] log10 copies/ml day -1 , P < 0.001). For nonsevere cases, the estimated mean duration of 238 LRT shedding (down to 0 log10 copies/ml) was 20.4 (95% CI, 13.2-27.7) DFSO. 239

Accordingly, the distributions of severe and nonsevere LRT shedding bifurcated along 240 disease course ( Figure 3B ). At 6 DFSO, the 80 th cp estimate of LRT rVL was 9.40 (95% CI, 241 8.67-10.20) log10 copies/ml for severe COVID-19, while it was 7.66 (95% CI, 6.65-8.83) log10 242 copies/ml for nonsevere illness. At 10 DFSO, the difference between 80 th -cp estimates expanded, 243 as they were 8.63 (95% CI, 8.04-9.26) and 6.01 (95% CI, 4.65-7.78) log10 copies/ml for severe 244 and nonsevere disease, respectively. 245

Our data indicated that nonsevere illness yielded greater skewing in LRT shedding than 246 severe disease in the analyzed period ( Figure 3B ). For nonsevere COVID-19, the SD of rVL was 247 1.92, 2.01 and 2.09 log10 copies/ml at 6, 8 and 10 DFSO, respectively. For severe disease, it was 248 lesser 1.25, 1.37 and 1.61 log10 copies/ml at 6, 8 and 10 DFSO, respectively. 249

For severe COVID-19, regression analysis showed, in the LRT, comparable mean rVLs at 4 250 DFSO between younger and older adults (P for intercept = 0.745) ( Figure 3C ). Both severe age 251 cohorts also showed persistent LRT shedding in the analyzed period: younger adults (-0.20 [95% 252 CI, -0.32 to 0.042] log10 copies/ml day -1 , P = 0.105) and older adults (-0.13 [95% CI, -0.39 to 253 0.13] log10 copies/ml day -1 , P = 0.316) both had no significant trend in SARS-CoV-2 clearance. 254

Likewise, severely affected male cases had no significant trend in LRT shedding (0.001 [95% 255

. CC-BY-NC 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted February 19, 2021. ; https://doi.org/10.1101/2021.02.17.21251926 doi: medRxiv preprint statistically analyses were not conducted (Appendix Table 3 ). 257

Interestingly, nonsevere cases showed similar SARS-CoV-2 shedding between the URT and 258 LRT, whereas severe cases shed greater and longer in the LRT than the URT (Figure 3, D and 259 E). At 4 DFSO, the URT rVL of nonsevere adults was 6.62 (95% CI, 6.50-6.74) log10 copies/ml, 260 which was not different from the LRT rVL of nonsevere adults (P for intercept = 0.651). In 261 contrast, at 4 DFSO, the URT rVL of severe adults (7.34 [95% CI, 7.01-7.68] log10 copies/ml) 262 was significantly lower than the LRT rVL of severe adults (P for intercept = 0.031). 263 264

For the pediatric cohort, regression estimated, in the URT, the mean rVL at 1 DFSO to be 266 7.32 (95% CI, 6.78-7.86) log10 copies/ml and SARS-CoV-2 clearance rate as -0.32 (95% CI, -267 0.42 to -0.22) log10 copies/ml day -1 (Figure 4A ). Both estimates were comparable between the 268 sexes for children ( Figure 4D) . The estimated mean duration of URT shedding (down to 0 log10 269 copies/ml) was 22.6 (95% CI, 17.0-28.1) DFSO for children with COVID-19. 270

Between pediatric cases, who had nonsevere illness in our dataset, and adults with nonsevere 271 illness, both URT shedding at 1 DFSO (P for intercept = 0.653) and URT dynamics (P for 272 interaction = 0.400) were similar (Figure 4A) . Distributions of rVL were also comparable 273 between these cohorts ( Figure 4B) . Conversely, URT shedding at 1 DFSO was greater for 274 severely affected adults when compared to nonsevere pediatric cases (P for intercept = 0.017), 275 but URT dynamics remained similar (P for interaction = 0.863) (Figure 4C) . we found that adults with severe COVID-19 showed higher rVLs shortly after symptom onset, 281 but similar SARS-CoV-2 clearance rates, when compared with their nonsevere counterparts. In 282 the LRT, we found that high, persistent shedding was associated with severe COVID-19, but not 283 nonsevere illness, in adults. Interestingly, in the analyzed periods, adults with severe disease 284 tended to have higher rVLs in the LRT than the URT. 285

After stratifying for disease severity, we found that sex and age had nonsignificant effects on 286 post-symptom-onset SARS-CoV-2 shedding levels and dynamics for each included analysis 287 (summarized in Table 2 ). Thus, while sex and age influence the tendency to develop severe 288 COVID-19 (2-4), we find no such sex dimorphism or age distinction in URT shedding among 289 cases of similar severity. This includes children, who had nonsevere illness in our study and 290

show similar URT shedding post-symptom onset as adults with nonsevere illness. 291

Notably, our analyses indicate that high, persistent LRT shedding of SARS-CoV-2 292 characterizes severe COVID-19 in adults. This suggests that the effective immune responses 293 associated with milder COVID-19, including innate, cross-reactive and coordinated adaptive 294 immunity (5-9), do not significantly inhibit early, or prolonged, SARS-CoV-2 replication in the 295 LRT of severely affected adults. Hence, uncontrolled LRT replication tends to continue, at least, 296 to 10 DFSO, coinciding with the timing of clinical deterioration (median, 10 DFSO) (2, 49). 297 Furthermore, the bifurcated profiles of LRT shedding concur with the observed severity-298 associated differences in lung pathology, in which severe cases show hyperinflammation and 299 progressive loss of epithelial-endothelial integrity (50-52). 300 . CC-BY-NC 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted February 19, 2021. ; https://doi.org/10.1101/2021.02.17.21251926 doi: medRxiv preprint They reinforce that severe COVID-19 is associated with greater rVLs than nonsevere illness (12-302 14), and suggest that sex and age may not significantly influence prognostic thresholds. In the 303 URT, both nonsevere and severe cases tend to clear SARS-CoV-2 at comparable rates. Thus, 304 time course of disease (e.g., DFSO) should be considered alongside rVL, rather than simply 305 employing rVL at admission. LRT shedding, however, bifurcates considerably between 306 nonsevere and severe COVID-19, meaning that SARS-CoV-2 quantitation from the LRT may 307 more accurately predict severity. While URT specimens are typically used to diagnose COVID-308 19, LRT specimens (our study predominantly analyzed sputum) may be collected from high-risk 309 patients for severity prognostication. 310

While our analyses did not account for virus infectivity, higher SARS-CoV-2 rVL is 311 associated with a higher likelihood of culture positivity, from adults (15, 16) as well as children 312

(36), and higher transmission risk (10). Hence, our results suggest that infectiousness increases 313 with COVID-19 severity, concurring with epidemiological analyses (53, 54). They also suggest 314 that adult and pediatric infections of similar severity have comparable infectiousness, reflecting 315 epidemiological findings on age-based infectiousness (54-56). Moreover, since respiratory 316 aerosols are typically produced from the LRT (57), severe SARS-CoV-2 infections may have 317 increased, and extended, risk for aerosol transmission. As severe cases tend to be hospitalized, 318 this provides one possible explanation for the elevated risk of COVID-19 among healthcare 319 workers in inpatient settings (58); airborne precautions, such as the use of N95 or air-purifying 320 respirators, should be implemented around patients with COVID-19. 321

Our study has limitations. First, while our study design systematically developed a large, 322 diverse dataset, there were few severe female cases with LRT specimens and no severe pediatric 323 . CC-BY-NC 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted February 19, 2021. ; https://doi.org/10.1101/2021.02.17.21251926 doi: medRxiv preprint cases included. Statistical comparisons involving these cohorts were not conducted based on 324 increased sensitivity to sampling bias, as COVID-19 presents broad heterogeneity in rVL. 325

Additional studies should permit these remaining comparisons. Second, our analyses did not 326 assess the influence of therapies or additional case characteristics, including comorbidities. 327

While the relationships between some comorbidities and SARS-CoV-2 kinetics remain unclear, 328 recent studies indicate many potential therapies (e.g., remdesivir, hydroxychloroquine, lopinavir, 329 ritonavir, low-dose monoclonal antibodies and ivermectin) have no significant anti-SARS-CoV-2 330 effects in patients (59-64). Third, the systematic dataset consisted largely of hospitalized 331 patients, and our results may not generalize to asymptomatic infections. 332

In summary, our findings provide insight into SARS-CoV-2 kinetics and describe 333 virological factors that distinguish severe COVID-19 from nonsevere illness. They show that 334 high, persistent LRT shedding characterizes severe disease in adults, highlighting the potential 335 prognostic utility of SARS-CoV-2 quantitation from LRT specimens. Lastly, each study 336 identified by our systematic review collected specimens before October 2020. As widespread 337 transmission of the emerging variants of concern likely occurred after this date (65, 66), our 338 study presents a quantitative resource to assess the effects of their mutations on respiratory 339 shedding levels and dynamics. 340 . CC-BY-NC 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. . CC-BY-NC 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) . CC-BY-NC 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The table summarizes collected case characteristics in the systematic dataset. Adult cases were 543 those aged 18 y or older, while pediatric cases were those aged younger than 18 y. 544

. CC-BY-NC 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)

The copyright holder for this preprint this version posted February 19, 2021. ; https://doi.org/10.1101/2021.02.17.21251926 doi: medRxiv preprint Figure 1 . Study selection. 559 . CC-BY-NC 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) each cohort. C. Regression analysis comparing URT shedding between nonsevere pediatric and 608 . CC-BY-NC 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)

The copyright holder for this preprint this version posted February 19, 2021. ; https://doi.org/10.1101/2021.02.17.21251926 doi: medRxiv preprint

An interactive web-based dashboard to track COVID-19 in real time

The authors thank S. Fafi-Kremer, PharmD, PhD (Strasbourg University