SARS-CoV2 Testing: The Limit of Detection Matters

Abstract

25Resolving the COVID-19 pandemic requires diagnostic testing to determine which individuals 26 are infected and which are not. The current gold standard is to perform RT-PCR on 27 nasopharyngeal samples. Best-in-class assays demonstrate a limit of detection (LoD) of 100 28 copies of viral RNA per milliliter of transport media. However, LoDs of currently approved 29 assays vary over 10,000-fold. Assays with higher LoDs will miss more infected patients, 30 resulting in more false negatives. However, the false-negative rate for a given LoD remains 31 unknown. Here we address this question using over 27,500 test results for patients from across 32 our healthcare network tested using the Abbott RealTime SARS-CoV-2 EUA. These results 33suggest that each 10-fold increase in LoD is expected to increase the false negative rate by 34 13%, missing an additional one in eight infected patients. The highest LoDs on the market will 35 miss a majority of infected patients, with false negative rates as high as 70%. These results 36 suggest that choice of assay has meaningful clinical and epidemiological consequences. The 37 limit of detection matters. 38 39 102 Efficiency was measured from plots of fluorescence intensity vs. cycle number for 50 positive 103 samples chosen at random, yielding an expression for viral load in copies/mL as a function of Ct 104 (Eq. 6, Supplementary Methods). Per this expression, the expected negative cutoff corresponds 105 to 9.2 copies per mL or ~2 virions per RT-PCR reaction volume (0.5mL), supporting the validity 106 of our parameter estimation.107 We used Python (v3.6) and its NumPy, SciPy, Matplotlib, and Pandas libraries to plot linear 108 regression and Theil-Sen slopes with 95% confidence intervals on repeat positives; a 109 normalized cumulative distribution (histogram) of positive results (with reversed x-axis for ease 110 of interpretation); binned histogram by 0.5 log10 units, and linear regression on log10-111 transformed data. 112 Results 113 Of the 27,098 tests performed on 20,076 patients over the testing period, 6,037 tests were 114 positive (22%), representing 4,774 unique patients. Analysis of repeats within 6 or 12 hours of 115 each other (7) demonstrated high repeatability of Ct values over these short time windows (R 2 116 0.70 and 0.63, n=25 and 51, respectively), supporting the validity of this quantitative measure as 117 a basis for assessment of viral load in patients (Fig. 1). We used basic principles of PCR and 118 detailed measurements of PCR efficiency on 50 randomly chosen positive samples to convert 119from Ct values to viral load, in units of copies of viral RNA per mL of viral transport medium. In 120 order to study the patient population upon presentation without confounding by repeat 121 measurements on the same patients, the remainder of the analysis was on the first positive 122 value for the above 4,774 unique patients.

Superspreading Events Without Superspreaders: Using High Attack Rate Events to Estimate Nº for Airborne Transmission of COVID-19

“…19 In sum, the period most relevant for infectivity samples only a narrow slice of the total range of viral loads observed across disease progression, helping to explain why selecting a single viral load value for the five cases results in a modest range for N 0 values (~300-2,000). 17 It should be noted that this range is broadened by measurement differences: the data contains 9 patients from Germany which were analyzed using sputum samples. These samples are systematically higher than those from other 16 patients, which were analyzed using nasopharyngeal (NP) swabs.…”

“…measurements include Arnaout et al (initial positive results from 4,774 patients, mean of ~5.2 log 10 copies/mL) and Jacot et al (initial positive results from 4,172 patients, mean of ~6.5 log 10 copies/mL, and median of 6.77 log 10 copies/mL)[16,17] . Arnaout finds a maximum viral load of 2.5 × 10 9 copies/mL (9.4 log 10 copies/mL) while Jacot finds a maximum viral load of ~2 × 10 10 copies/mL (10.3 log 10 copies/mL), broadly consistent with Kleibocker and Jones.…”

“…Kleiboeker et al do not report a standard deviation, so the figure of ~2.0 log 10 copies/mL was calculated based on the underlying data provided to us by Kleiboeker (private communication). The mean log 10 copies/mL for Arnaout and Jacot were calculated based on digitizing the data in their figures 2(a) and 1(a), respectively[16,17] 5. Jacot et al only measure the initial sample of a given patient, but these patients were tested at different points during disease progression[16] .…”

The Path Forward for COVID-19 Diagnostics

“…However, more conclusive studies could be conducted to examine viral load in the context of the LOD required for diagnostics to be effective, especially with respect to multiple different bodily fluids. This is particularly important since viral loads are known to vary widely across individuals, ranging from well above LODs of diagnostic tests to near or below them [19].…”

Saline collection media and an extraction free workflow enables massively scalable, highly sensitive, and cost-effective SARS-CoV-2 testing

“…However, saline-based specimens have been shown to perform poorly in extraction-free work ow 3,6 . In contrast, the method presented here utilizes TBE dilution, heat treatment, and large sample input volumes to achieve 100% sensitivity and speci city in 60 clinical samples, and a LoD at 1,000 GCE/mL, which is comparable to assays using extraction-based methods 11,13 . The limitations of this study include the relatively small number of available clinical samples that preclude a more thorough analysis of sensitivity compared to extractionbased work ows.…”