High capacity clinical SARS-CoV-2 molecular testing using combinatorial pooling

Abstract

Background: The SARS-CoV-2 pandemic led to unprecedented testing demands, causing major testing delays globally. One strategy used for increasing testing capacity was pooled-testing, using a two-stage technique first introduced during WWII. However, such traditional pooled testing was used in practice only when positivity rates were below 2%.

Methods: Here we report the development, validation and clinical application of P-BEST - a single-stage pooled-testing strategy that was approved for clinical use in Israel.

Results: P-BEST is clinically validated using 3636 side-by-side tests and is able to correctly detect all positive samples and accurately estimate their Ct value. Following regulatory approval by the Israeli Ministry of Health, P-BEST was used in 2021 to clinically test 837,138 samples using 270,095 PCR tests - a 3.1fold reduction in the number of tests. This period includes the Alpha and Delta waves, when positivity rates exceeded 10%, rendering traditional pooling non-practical. We also describe a tablet-based solution that allows performing manual single-stage pooling in settings where liquid dispensing robots are not available.

Conclusions: Our data provides a proof-of-concept for large-scale clinical implementation of single-stage pooled-testing for continuous surveillance of multiple pathogens with reduced test costs, and as an important tool for increasing testing efficiency during pandemic outbreaks.

Fig. 1. Two-stage pooling vs. single-stage pooling.

a In traditional two-stage Dorfman pooling, each sample is added to a single pool. Samples belonging to negative poos are classified as “negative” and results can be reported after a single round of testing. Positive pools indicate that one or more of the pooled samples are positive, thus all corresponding samples are re-tested individually and only then, results can be reported. b In combinatorial single-stage pooling, each sample is added to several pools, according to a specific pooling design. A decoding algorithm is used to detect all positive samples and their Ct value after a single round of testing.

Fig. 2. Classification example.

a A schematic design of five samples and five pools, their viral load measurements, and the classification of each sample as “positive” ( + ), “negative” (-) or “suspected” (?). The numbers in gray shaded squares correspond to the true, although unknown viral load of each sample. b The same case as in panel A, with S2 as a weak positive sample.

Fig. 3. Ct value estimates via P-BEST.

a A scatter plot of the Ct value of positive samples as measured individually (x-axis) vs. their P-BEST estimated value (y-axis). Shown is the Pearson correlation coefficient. b Boxplots of the the Ct value (as measured individually) of samples classified by P-BEST as “suspected”. Samples are divided into 3 groups according to the reason for their “suspected” classification, i.e., invalidated pools (blue), weak pools (orange), and overcrowding (green). Each dot represents the Ct value of a single sample. Black lines represent the median, boxes denote the 25th and 75th percentiles, and error bars represent 1.5 times the interquartile range.

Fig. 4. P-BEST in clinical diagnostics during 2021.

a The number of P-BEST runs performed stratified by pooling design. b The average weekly positivity rate in Israel in 2021 based on Israeli MOH published data. c Weekly number of samples tested by P-BEST color-coded by pooling design. d Weekly number of runs performed using P-BEST color-coded by pooling design. The same as C for the number of runs.

Fig. 5. Tablet-directed pooling.

a Description of the equipment required for manual pooling. b The number of samples and pools in each manual design and their efficiency. Recommended positivity rate corresponds to the rate below which the number of “suspected” samples reported by the method is negligible. The run time corresponds to total pipetting time for sample pooling using a manual pipette. c An illustration of the first three iPipet pipetting operations using two 96-well plates on top of a standard tablet.