Evaluation and Comparison of the Hologic Aptima SARS-CoV-2 Assay and the CDC 2019-nCoV Real-Time Reverse Transcription-PCR Diagnostic Panel Using a Four-Sample Pooling Approach

To expand testing capacity during the severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) pandemic, numerous molecular assays have been granted emergency-use authorization (EUA) for testing individual patient samples but not for sample pooling. Although pooling multiple patient samples may expand the testing capacity while reducing the burden of supply of test reagents, consumables, and kits, pooling inherently results in a dilution that may reduce sensitivity. The lack of detection of weakly positive samples may have significant consequences in efforts to curb SARS-CoV-2 transmission. At this time, only two pooling assays have received EUA (1). Quest Diagnostics applied four-sample pooling, stating 100% sensitivity (2). While it is unclear how many weakly positive samples were included, based on linear regression analysis, they state that samples with cycle threshold (C_T) values of >37 are expected to be missed (2). The second EUA, by Poplar Healthcare, leveraged the Hologic Aptima SARS-CoV-2 assay with seven-sample pooling, stating 100% sensitivity (3). However, the highest C_T value tested was only 34.6. Griesemer et al. evaluated five- and nine-sample pooling approaches using the CDC 2019 novel coronavirus (2019-nCoV) real-time reverse transcription-PCR (RT-PCR) diagnostic panel, emphasizing weakly positive samples (C_T = 33 to 39) (4). They reported that 4 of the 24 nine-sample pools were missed, but all were detected in a five-sample pool.

We evaluated four-sample pooling using the CDC 2019-nCoV real-time RT-PCR diagnostic panel (CDC assay) and the Hologic Aptima SARS-CoV-2 transcription-mediated amplification (TMA) assay, with 25% of samples having values within 2 to 3 C_Ts of the assay’s limit of detection, according to FDA guidance (1). Frozen, residual nasopharyngeal swabs (BD flexible minitip flocked swab or Puritan PurFlock Ultra flocked swab) collected in viral transport medium (VTM) (Xpert VTM, Copan or equivalent VTM, and Remel M4, M4RT, or M6) from patients presenting to Pittsburgh-based UPMC medical facilities were used. Samples were originally tested using the Cepheid SARS-CoV-2 EUA (Cepheid) assay, which served as the reference method, and averaged C_T values of N1 and E targets were used to classify samples into group 1 (C_T < 34; n = 23), group 2 (C_T = 34 to 36; n = 4), and group 3 (C_T ≥ 37; n = 8), according to FDA guidance (1). For pools, samples were thawed and mixed, and 500 μl each of four samples were pooled. Thirty-five positive pools, each consisting of 1 positive and 3 negative samples, and 20 negative pools (4 uniquely negative samples) were made. For TMA, 500 μl of the pool was added to the lysis buffer tube according to the instructions for use (IFU) (5). For the CDC panel, 200 μl of pools was used for extraction on the bioMérieux NucliSens easyMag system according to the IFU (6, 7). Positive samples were retested individually on both platforms; negative samples were not.

All 20 negative pools were negative by both assays, resulting in 100% specificity (data not shown). Relative light unit (RLU) values of all negative pools ranged between 301 and 328. Testing of positive samples individually confirmed positive results (Tables 1 and 2). For positive pools, the CDC assay correctly detected 34/35, with a sensitivity of 97.4%. Sample 34 was missed, with an average C_T of 37.7 (Table 1). However, the seven other samples having C_T values of >37 were positive when pooled. Therefore, the sensitivities of the CDC assay were 100% (27/27) for group 1/group 2 samples and 87.5% (7/8) for group 3 positive samples. C_T correlation between individual and pooled was strong, with an R² value of >0.97 (Fig. 1). Overall, the average differences between individual and pooled C_T values were 2.2 ± 1.23 (N1) and 2.2 ± 1.54 (N2), with the pooled samples having weaker C_T values (data not shown). When combining targets, the average difference in C_T values was 2.4 ± 1 (data not shown). Correlation of individual results from Cepheid and CDC C_T values reveals an overall R² of 0.95 (Fig. 2), with an average difference of 1.2 ± 1.3 C_Ts (data not shown).

The TMA assay correctly detected 32/35 positive pools, with a sensitivity of 91.4%. The three discrepant samples had C_T values falling within group 3 (Table 2). Samples 24 and 26 had RLU values within or close to the negative RLU range observed, where the value for sample 32 was slightly higher. The other five group 3 positive samples were detected by TMA with RLU values of >600 (Table 2). Notably, an absolute RLU value is not a sole indicator of a positive or negative result. Like the CDC assay, the TMA assay was 100% sensitive (27/27) for pooled group 1/group 2 samples but had reduced sensitivity (62.5%; 5/8) for group 3 positive samples.

A sample pooling approach for the detection of SARS-CoV-2 may be a solution to expand testing capacity. Previous studies have evaluated sample pooling using real-time PCR assays, showing a range of sensitivities (2–4, 8). Our evaluation revealed that CDC and TMA sensitivities remained 100% for C_T values of <37 but dropped for C_T values of ≥37. Limited data exist for pooling on TMA; however, this automated, high-throughput platform is well poised for pooling. The TMA assay had an acceptable performance, with a reduction in sensitivity seen strictly with group 3 positive samples. Two considerations when comparing data from pooling studies are the numbers of samples used to create the pool and the differences in assay methodologies. Indeed, Griesemer et al. showed reduced sensitivity when they attempted to use a nine-sample pooling approach (4). Additionally, differences in assay methodologies may also play a role in performance differences. For example, unlike the CDC assay, the TMA assay does not utilize a full extraction step prior to signal amplification, so a 6% difference in sensitivity is surprisingly good.

Overall, this study supports that a four-sample pooling strategy on either the CDC or TMA platform retains >90% sensitivity for the detection of SARS-CoV-2. Validation of sample pooling is just one of many steps in implementing such a strategy. The selection of an appropriate patient population for pooling is an important factor. In our study, we aimed to use such an approach for asymptomatic individuals requiring SARS-CoV-2 screening, where the prevalence was <1%. While four-sample pooling is likely a conservative approach with our intended low-prevalence population, we chose to limit the number of samples pooled as four-sample pooling would also be applicable in higher-prevalence settings, up to about 5 to 6%. Other considerations that remain a challenge are tracking individual specimens in pools, retrieving individual samples for confirmatory testing, automated and streamlined protocols, workflow, reporting, and billing. Laboratory personnel remains a crux to expanding laboratory testing. Solutions to these barriers and proper infrastructure, resources, and support will be required for long-term adoption and implementation.

This project was undertaken as a laboratory quality improvement initiative and as such was not formally reviewed by the University of Pittsburgh Institutional Review Board.

S.L.M. and S.E.V. have no disclosures.