Safe and effective pool testing for SARS-CoV-2 detection

Abstract

Objectives: The global spread of SARS-CoV-2 is a serious public health issue. Large-scale surveillance screenings are crucial but can exceed test capacities. We (A) optimized test conditions and (B) implemented pool testing of respiratory swabs into SARS-CoV-2 diagnostics.

Study design: (A) We determined the optimal pooling strategy and pool size. In addition, we measured the impact of vortexing prior to sample processing, compared a pipette-pooling method (by combining transport medium of several specimens) and a swab-pooling method (by combining several swabs into a test tube filled with PBS) as well as determined the sensitivities of three PCR assays. (B) Finally, we applied high-throughput pool testing for diagnostics.

Results: (A) In a low prevalence setting, we defined a preferable pool size of ten in a two-stage hierarchical pool testing strategy. Vortexing of swabs (n = 33) increased cellular yield by a factor of 2.34. By comparing Ct-values of 16 pools generated with two different pooling strategies, pipette-pooling was more efficient compared to swab-pooling. Measuring dilution series of 20 SARS-CoV-2 positive samples in three PCR assays simultaneously revealed detection rates of 85% (assay I), 50% (assay II), and 95% (assay III) at a 1:100 dilution. (B) We systematically pooled 55,690 samples in a period of 44 weeks resulting in a reduction of 47,369 PCR reactions.

Conclusions: For implementing pooling strategies into high-throughput diagnostics, we recommend utilizing a pipette-pooling method, performing sensitivity validation of the PCR assays used, and vortexing swabs prior to analyses. Pool testing for SARS-CoV-2 detection is feasible and effective in a low prevalence setting.

Keywords: Pool testing; SARS-CoV-2; Surveillance.

Fig. 1

Hierarchical pool testing for SARS-CoV-2 detection. A: Illustration of the two-stage hierarchical pool testing strategy. B: The reduction of PCR-tests compared to individual testing (continuous lines are nonlinear regression curves, outer dotted lines are 95% or 99% test sensitivity, respectively) and C: The expected number of tests for different pool sizes are shown. Data visualized in Figures B and C are generated using the shiny app of Christopher Bilder, which is based on an algorithm to compute the expected number of tests when performing two-stage hierarchical pool testing [5,8]. D: The mean positivity rate per week of tests performed at the University Hospital of Cologne and in Germany (as published [11]) are shown.

Fig. 2

Validation of the pooling method and determining PCR sensitivity. A and B: β-globin concentration in individual specimens before and after vortexing (n = 33). A Mann-Whitney test was performed. C and D: Sample preparation time was measured for four different operators preparing n = 10 samples in 6 replicates. E and F: Swab-pooling and pipette-pooling are illustrated, and processing time was measured for four operators preparing n = 6 pools with a size of 10 each (paired t-test was performed). G: Ct-values are displayed for n = 16 single positive specimens (ctrl) as well as for each positive specimen in a pool prepared either by the pipette or swab pooling method, respectively, and tested in assay I. Negative test results are highlighted by the triangle shape. H: Ten-fold dilution series of n = 20 SARS-CoV-2-positive samples, tested with three PCR assays. I: The amplification factor was calculated for dilution series containing five Ct-values. J: The mean and standard deviation of Ct-values and K: Detection rate of n = 20 undiluted samples are shown. L: Lowest detectable SARS-CoV-2 copy number as determined using dilution series of cell culture supernatant extrapolated to approved standards measured in all assays. M: Ct-values of n = 25 positive samples combined with a stock of negative specimens in a 1:5, 1:10, 1:20 and 1:50 dilution, respectively, tested in assay III. Ctrl: control, **p ≤ 0.01, ***p ≤ 0.001, ****p ≤ 0.0001.

Fig. 3

Performance of high-throughput pool testing for SARS-CoV-2 detection. A: Pool testing started on April 9, 2020. The number of pooled samples per day and the percentage of reduced PCR-tests compared to individual testing (blue line) are displayed. B: The number of samples tested in pools and C: The number of pools tested during a period of 44 weeks are shown. D: Correlation of Ct-values of n = 128 positive pools and the respective individual positive sample. Correlation was performed only if a pool and the respective positive individual sample was analyzed with the same assay. E: Violin plot of adjusted Ct-values (Ct_a) of n = 175 individual positive samples detected in pools. The reduced number of data points is due to a software problem, so that some data could not be retrieved. Dotted lines represent quartiles and the dashed line the median.