Pool testing of nasopharyngeal specimens by QIAstat for detection of SARS-CoV-2
NPFS specimens were tested by QIAstat either in a pool of 5 specimens (pool-5) or in a pool of 10 specimens (pool-10) (Table 1). For pool-5 testing, a total of 10 pooled runs (n = 50 specimens) that gave positive results and 11 pooled runs (n = 55 specimens) that gave negative results by QIAstat were individually assessed by standard RT-qPCR. For pool-10 testing, specimens in 14 positive pools (n = 140 specimens) and 12 negative pools (n = 120 specimens) by QIAstat were individually assessed. QIAstat test results for respiratory pathogens other than SARS-CoV-2 were disregarded. By individual testing, a total of 6 pools had > 1 positive samples in the pool-10 group (S1 table). Among these pools, two were undetectable by QIAstat pool-10 testing. In both pool groups, all of the positive pool results were correct, but 3 of 10 negative pool-5 results and 5 of 12 negative pools in pool-10 results were falsely negative after individual assessment. The sensitivity, specificity and accuracy of QIAstat pool-5 and pool-10 against standard individual testing were not significantly different (Table 1). A total of 10 positive results were missed by the QIAstat pool testing approach: the RT-qPCR CT values by the standard method for these specimens ranged from 32.6 to 38.3. To investigate whether lower analytical sensitivity of QIAstat was responsible for these discrepant results, we individually tested some of these specimens by QIAstat and noted that the positive samples, that were missed by pooled testing, were also negative by individual QIAstat testing, although they were positive by two other methods (S2 table). The mean CT values for these samples by standard RT-qPCR and Xpert were 35.1±1.5 and 35.1±1.9 respectively. For samples that were positive by both pool testing and individual testing approach, the RT-qPCR CT values were positively correlated (Pearson correlation coefficient r=0.9719 and 0.9181; p<0.001; for pool-5 and pool-10, respectively) (Figs 1A and 1B). Mean CT value change (DCT) because of pooling 5 and 10 specimens were not significantly different from each other (p=0.5354) (Fig 1F).
Pool testing of nasopharyngeal specimens by Xpert for detection of SARS-CoV-2
For pool-5 testing of NP swabs by Xpert, a total of 31 pooled runs (n = 155 specimens) that gave positive results and 19 pooled runs (n = 95 specimens) that gave negative results by Xpert were individually assessed by standard RT-qPCR. For pool-10 testing, specimens in 11 positive pools (n = 110 specimens) and 9 negative pools (n = 90 specimens) by Xpert were individually assessed. By individual testing, a total of 4 pools had > 1 positive samples in the pool-5 group (S1 table). All pool test results, except one, were correct in each of the pool-5 and pool-10 groups. The sensitivity, specificity and accuracy of Xpert pool testing in NP specimens against standard, individual testing were >95% in both group and were not affected by pool size of 5 or 10 specimens (Table 1). The samples that gave a false negative result in pool-5 had a PCR CT value of 39 by the standard test, and the sample that gave a false positive result in pool-10 had a PCR CT value of 42.5 by Xpert. For samples that were positive by both methods, the RT-qPCR CT values were positively correlated (Pearson correlation coefficient r= 0.9719 and 0.9583; p<0.0001; for pool-5 and pool-10, respectively) (Fig 1C and 1D). Mean CT value change (DCT) because of pooling 5 and 10 specimens were not significantly different from each other (p=0.4765) (Fig 1F). However, the mean CT value difference of QIAstat vs Xpert pool testing (DCT) from standard testing were significantly different (p<0.01) (Fig 1F).
Pool testing of saliva specimens by Xpert for detection of SARS-CoV-2
For testing of pooled saliva specimens by Xpert a total of 8 pooled runs (n = 80 specimens) that gave positive results and 7 pooled runs (n = 70 specimens) that gave negative results were individually assessed by standard RT-qPCR. All except two (87 %) pooled test results were correct. The sensitivity, specificity and accuracy of saliva pool testing by Xpert against standard, individual testing were 87.5%, 99.3% and 98.7%, respectively (Table 1). For samples that were positive by both methods, the RT-qPCR CT values were positively correlated (Pearson correlation coefficient r=0.8535, p<0.05) (Fig 1E). Mean CT value change (DCT) because of pooling 10 saliva specimens was not significantly different from pool-10 testing of NP swab specimens (p=0.6507) (Fig 1F).
Correlation and comparison of CT values by QIAstat and Xpert tests in individual and pooled specimens
For a direct comparison of CT values, we tested a subset of known positive specimens (n=6; CT <30 by standard testing) by both QIAstat and Xpert tests both individually and in a pool of 10 NPFS specimens. The CT values obtained by individual versus pool-testing were positively correlated by both QIAstat (r=0.9868, p<0.001) and Xpert (r=8631, p<0.05) testing (Fig 2A and B). The mean CT value changes (DCT) because of pooling 10 NPFS specimens were -3.683 (95% CI, -5.106 to -2.261) and -3.333 (95% CI, -5.861 to -0.8054) by QIAstat and Xpert, respectively, which was not significantly different from each other (p=0.7316) (Fig 2C). However, CT values obtained by Xpert were significantly (p<0.01 by paired, Student’s T-test) lower or stronger than those of QIAstat tests, with mean difference of -4.250 (95% CI, -6.795 to -1.705) and -4.6 (95%CI, -6.703 to 2.497) for individual and pooled tests, respectively (Fig 2D).