SARS-CoV-2 exhibits extreme differences in early viral loads among specimen types suggesting improved detection of pre-infectious and infectious individuals using combination specimen types (preprint)

ABSTRACT

Background. Screening testing, often via self-collected specimens, remains a key strategy to detect infections early and prevent SARS-CoV-2 transmission, and to enable earlier initiation of treatment. However, which specimen type best detects the earliest days of infection remains controversial. Further, the analytical sensitivity of diagnostic tests must also be considered, as viral loads below a test's limit of detection (LOD) are likely to yield false-negative results. Comparisons of quantitative, longitudinal SARS-CoV-2 viral-load timecourses in multiple specimen types can determine the best specimen type and test analytical sensitivity for earliest detection of infection. Methods. We conducted a COVID-19 household transmission study between November 2021 and February 2022 that enrolled 228 participants and analyzed 6,825 samples using RT-qPCR to quantify viral-load timecourses in three specimen types (saliva [SA], anterior-nares swab [ANS], and oropharyngeal swab [OPS]). From this study population, 14 participants enrolled before or at the incidence of infection with the Omicron variant. We compared the viral loads in specimens collected from each person at the same timepoint, and the longitudinal viral load timecourses from each participant. Using these viral loads, we inferred the clinical sensitivity of each specimen type to detect infected, pre-infectious and infectious individuals (based on presumably infectious viral load levels) using assays with a range of analytical sensitivities. We also inferred the clinical sensitivity of computationally contrived specimen types representing combinations of single specimen types. Results. We found extreme differences (up to 109 copies/mL) in viral loads between paired specimen types in the same person at the same timepoint, and that longitudinal viral load timecourses across specimen types did not correlate. Because of this lack of correlation, infectious viral loads were often observed in different specimen types asynchronously throughout the course of the infection. In the first 4 days of infection, no single specimen type was inferred to achieve >95% detection of infected or infectious individuals, even with the highest analytical sensitivity assays. In nearly all participants (11/14), a rise in ANS viral loads was delayed (as many as 7 days) relative to SA and OPS. We also observed that ANS and OPS had the most complementary viral load timecourses, resulting in optimal inferred performance with a computationally contrived combined anterior nares-oropharyngeal (AN-OP) swab specimen type. The combination AN-OP swab had superior inferred clinical sensitivity the first 8 days of infection with both high- and low-analytical-sensitivity assays. This AN-OP swab was also inferred to significantly improve detection of pre-infectious and infectious individuals over any single specimen type. Conclusion. Our work demonstrates that the viral load in one specimen type cannot reliably predict the viral load in another specimen type. Combination specimen types may offer a more robust approach for earliest detection of new variants and respiratory viruses when viral kinetics are still unknown.