key: cord-0713890-zbwf54pj authors: Wunsch, Marie; Aschemeier, Dominik; Heger, Eva; Ehrentraut, Denise; Krüger, Jan; Hufbauer, Martin; Syed, Adnan S; Horemheb-Rubio, Gibran; Dewald, Felix; Fish, Irina; Schlotz, Maike; Gruell, Henning; Augustin, Max; Lehmann, Clara; Kaiser, Rolf; Knops, Elena; Silling, Steffi; Klein, Florian title: Safe and effective pool testing for SARS-CoV-2 detection date: 2021-10-28 journal: J Clin Virol DOI: 10.1016/j.jcv.2021.105018 sha: 03fa437ee1c57685afc44287468441f71098b8ba doc_id: 713890 cord_uid: zbwf54pj 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 pipette-pooling method (by combining transport medium of several specimens) and swab-pooling method (by combining several swabs into a test tube filled with PBS) as well as the sensitivity 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. The SARS-CoV-2 pandemic is a serious public health problem of unprecedented magnitude in recent times. In particular individuals at older ages or with comorbidities are at a high risk to require hospitalization and intensive care [1] . Therefore, it is essential to control person-toperson transmission in order to protect vulnerable individuals and limit the number of severe cases. Until herd immunity is achieved by vaccination, nonpharmaceutical interventions need to be applied. Many countries could successfully contain the spread of COVID-19 through social distancing or lock-down measures, contact tracing, quarantine, and large-scale testing in the ongoing pandemic [2, 3] . In order to control viral transmission when lifting lock-down strategies, large-scale testing and surveillance are critical interventions. These approaches are based on frequent tests of individuals e.g. by rapid antigen-based tests or reverse transcription-real-time PCR to detect SARS-CoV-2 in swab specimens. However, large-scale surveillance screenings can exceed test capacities of diagnostic laboratories. Pooling swab specimens for PCR-testing can increase test capacities and limit the consumption of reagents [4] . Therefore, swab samples are combined and tested in a single PCR reaction. If this pool test is positive, the remaining sampling material of the included specimens can be retested separately to detect the infected individual. If the pool test is negative, all individuals are declared as not infected [5] [6] [7] (two-stage hierarchical pool testing). Pool testing is highly efficient in a setting of low disease prevalence and the availability of highly sensitive test methods [6] . It can be applied to enable surveillance screenings of asymptomatic individuals in public institutions e.g. hospitals, schools or retirement homes, which carry a high risk for superspreading events and severe disease courses. When pool testing is established, test conditions need to be optimized including (a) the pooling strategy and pool sizes, (b) sample preparation and pooling method, (c) the quality of SARS-CoV-2 5 detection by PCR. In this study, we determined and implemented the optimized pool testing procedure into the diagnostic routine for SARS-CoV-2 detection. 6 Pooling efficiency was computed using a web tool published by Bilder and colleagues [6, 8] . Calculations were performed assuming a PCR-test sensitivity of 99% or 95% and a test specificity of 99%. The expected number of tests was computed for different pool sizes as described [6] . = 100 (1 − expected number of tests ) % Oropharyngeal or combined nasal/oropharyngeal swabs were collected and transferred into To determine the cellular content of the same n=33 specimens before and after vortexing, human -globin-gene quantification was performed as published [9] . For the pipette pooling, an aliquot of the transport medium from each of ten storage tubes were combined into one test tube. For the swab pooling, transport medium was removed, PBS added to a first tube containing the swab, vortexed, and transferred into a second swab tube followed by vortexing. After the PBS had traveled through all ten swab tubes, it was transferred into a test tube. 16 different pools were conducted using each of the two pooling methods. Preparation time was measured. To simulate various pool sizes, 25 positive specimens with various Ct-values were diluted 1:5, 1:10, 1:20 and 1:50 in negative samples, respectively. To compare detection rates of three PCR systems, ten-fold dilution series of n=20 SARS-CoV-2-positive samples were simultaneously tested in three assays. After arrival in our laboratory, samples were vortexed 5 seconds, and preselected to be tested individually or in pools. Samples of symptomatic individuals and recently positive tested persons were excluded from pool testing. Within 44 weeks, 55,690 samples were tested in pools using the pipette pooling method. For correlation analysis, a spearman's rank correlation was used. For comparing -globingene concentrations, a Mann-Whitney test was performed. To assess statistical differences in Ct-values comparing pooling methods or PCR assays, a multiple comparison one-way ANOVA was used. For comparing preparation times and for matched-pair analysis, a paired t-test was used. The amplification factor was calculated as published [10] , Kruskal-Wallis test was performed. GraphPadPrism 7.0 (GraphPad Software, Inc.) was used for analysis. Figures were created using Adobe Illustrator 18.1 (Adobe Inc.). Pool testing can be performed using different strategies. In this study we conducted two-stage hierarchical pooling procedures ( Figure 1A ). Pool testing efficiency depends on the disease prevalence. Bilder and colleagues [6, 8] proposed an algorithm to compute the expected number of tests when performing two-stage hierarchical pool testing ( Figure 1B, C) . As the disease prevalence increases, the reduction of PCR-tests declines due to the retesting of individual samples of positive pools. However, the pooling efficiency of smaller pool sizes declines more slowly compared to pooling 20 or more samples. At the time pool testing was initiated, the positivity rate at the University Hospital of Cologne was 3.88% ( Figure 1D ). However, by excluding samples of symptomatic individuals and recently positive tested persons, the positivity rate of pooled samples was below 0.1%. 10 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. Pool testing requires optimal sample conditions in order to minimize false negative results. Pre-analytic factors can influence the test results. We could not detect a significant difference To test feasibility of the pipette-and swab-pooling method, four operators processed n=6 pools applying both methods, respectively ( Figure 2F ). The mean processing time was 3 minutes, 47 seconds for a swab-based pool (95% CI: (2 min,59sec.)-(4min,36sec.)) and 1 minute, 55 seconds for a pipette-based pool (95% CI: (1min,33sec.)-(2min,16sec.)). In order to investigate the sensitivity, we generated 16 different pools with each of the two pooling methods, by merging one SARS-CoV-2-positive sample with nine negative samples, respectively ( Figure 2G ). which yielded a negative test result (triangle shape in Figure 2G ). To compare the detection rates of three PCR systems used in our diagnostic laboratory, ten-fold dilution series of n=20 SARS-CoV-2-positive samples were simultaneously tested in three assays (I, II, and III, referring to the Roche LightCycler® 480II, the Hologic Panther Fusion®, and Roche Cobas®6800 System). Ct-values for e-gene amplification were analyzed as they yielded similar Ct-values compared to the second viral target, respectively (Supplementary Figure 1C) . As shown in Figure 2H Figure 2I ). The lowest detectable copy number was 200 copies for assay I, 2,000 for assay II, and 20 copies for assay III as determined using two approved standards ( Figure 2L ). To determine the detection-rate for different pool sizes, 25 Ctrl: control, **p ≤0.01, ***p ≤ 0.001, ****p ≤0.0001 The above experiments suggested the following as the optimal pool test conditions for SARS-CoV-2 detection: (a) pooling 10 samples using the two-stage hierarchical strategy; (b) vortexing the swab specimens before pooling; (c) applying the pipette-pooling method, and (d) utilizing assay III for PCR-testing. We set up a pool testing facility, implemented features for pool testing into the laboratory software, and systematically pooled up to 488 samples per day ( Figure 3A) . In order to limit the number of positive samples run in pools, we preselected samples supported by algorithms of the laboratory software. Patients that had been tested positive for SARS-CoV-2 before or showed COVID-19-like symptoms were tested 16 individually. Pool testing was performed for surveillance screenings of patients and staff of as well as for every patient admitted to the University Hospital of Cologne. The mean percentage of reduced PCR-tests was 85.77%. Decreased savings of PCR reactions were due to retesting caused by technical issues or positive tested pools. Within 44 weeks, 55,690 samples were tested in pools and only 4.7% (n=2,640 samples) had to be retested individually ( Figure 3B ). As Figure 3C shows Large-scale testing and surveillance screenings enable the rapid detection of clusters of infections and help preventing superspreading events and uncontrolled transmission of the virus until herd immunity by vaccination is reached. However, test capacities are limited and PCR-tests are cost-intensive. Pool testing is a feasible option to enable high-throughput screenings without overwhelming capacities of diagnostic laboratories. To our knowledge, this is the first systematic investigation addressing various aspects of pool testing for SARS-CoV-2 detection. However, reports and a review on this topic have recently been published [12] [13] [14] [15] [16] [17] [18] [19] [20] [21] [22] [23] [24] [25] [26] [27] . Eberhardt and colleagues suggest forming subgroups if a pool yields a positive result [15] . The samples of the positive subgroup are then tested individually. This strategy can further increase efficiency, as fewer tests need to be performed, but it also requires additional hands-on-time and a sensitive, rapid detection method to ensure that the test result for the individual specimen is not delayed. Pooling strategies need to be time efficient to rapidly detect SARS-CoV-2-infected individuals as well as suitable for high-throughput screenings. Following these considerations, we decided to use two-stage hierarchical pool testing. In a low prevalence setting with restricted test resources and a neglectable time aspect, increasing pool sizes and forming subgroups is a reasonable option. Another approach is the combinatorial pool testing strategy [19, 20, 28] . Here, samples are assigned into multiple pools which enables the detection of infected individuals in a single round of testing. Pre-analytical handling can substantially impact test sensitivity, however, limited data on this topic are available. Test results are influenced by improper transport conditions, variations of the sampling device (flocked vs. cotton swabs), the transport media [29, 30] as well as the anatomical structure of the pharynx (Mallampati score). We could not observe 20 differences in Ct-values comparing oropharyngeal and combined nasal/oropharyngeal specimens. This is in line with findings of Woelfl et al., describing no differences in viral loads or detection rates when comparing nasopharyngeal and oropharyngeal specimens [31] . In addition, high viral concentrations and detection rates in salvia compared to nasopharyngeal swabs were reported [24, 32] and saliva samples can be used for pool testing [33] . Pre-analytics can potentially influence the inhibition rate which in our setting seems to be a rather rare multifactorial event, possibly due to high quality sampling material, trained staff etc. Data on inhibition rates are of great interest but are exceeding the focus of this study. Compared to individual testing, pool testing requires some additional processing as well as documentation steps. By integrating a special feature for pool testing into our laboratory software, we were able to minimize additional hands-on-time. The already labelled tubes are scanned and assigned to a pool. If a pool tests positive, a request for re-testing of the individual samples is automatically generated by the lab software. The individual samples are stored according to the pool numbering, so that samples for re-testing can easily be identified. However, in order to save personnel capacities, it might be advisable to use a pipette robot, performing both pooling and documentation. A critical point in the context of pool testing is the time aspect. To ensure, that the test result of the individual sample is available and communicated on the same day, we have set a time limit up to which we perform pool testing. Samples arriving later that day are tested individually. Pooling of individual samples using the swab-method is more time-consuming and has additional limitations regarding handling and risk of contamination compared to the pipettemethod. However, recent developments in PCR diagnostics allow a rapid detection of SARS-CoV-2 in swab specimens [34] , and swab-pooling is valuable if pools (a collection of multiple 21 swabs in a tube) are prepared directly after collection in schools, old people's homes, hospitals, etc. We developed a feasible pooling procedure that can readily be implemented in diagnostic routines. The data communicated here will contribute to the process of finding a consensus pool testing strategy enabling larger test capacities to effectively combat the SARS-CoV-2 pandemic. Clinical course and risk factors for mortality of adult inpatients with COVID-19 in Wuhan, China: a retrospective cohort study Preparing for a responsible lockdown exit strategy Inferring change points in the spread of COVID-19 reveals the effectiveness of interventions Pooling nasopharyngeal/throat swab specimens to increase testing capacity for influenza viruses by PCR Pool size selection when testing for SARS-CoV-2 Tests in short supply? Optimal retesting configurations for hierarchical group testing A Shiny app for pooled testing Human papillomavirus 16 load in normal and abnormal cervical scrapes: an indicator of CIN II/III and viral clearance Amplification efficiency: linking baseline and bias in the analysis of quantitative PCR data Daily Situation Report of the Robert Koch Institute, Updated status for Germany 24 Pooling of samples for testing for SARS-CoV-2 in asymptomatic people Sample pooling as a strategy to detect community transmission of SARS-CoV-2 Challenges and issues of SARS-CoV-2 pool testing Challenges and issues of SARS-CoV-2 pool testing Challenges and issues of SARS-CoV-2 pool testing Large-scale implementation of pooled RNA extraction and RT-PCR for SARS-CoV-2 detection Evaluation of COVID-19 RT-qPCR test in multi-sample pools Efficient prevalence estimation and infected sample identification with group testing for SARS-CoV-2 Efficient high-throughput SARS-CoV-2 testing to detect asymptomatic carriers Evaluation of sample pooling for diagnosis of COVID-19 by Real time PCR-A resource saving combat strategy Severe Acute Respiratory Syndrome Coronavirus 2 (SARS-CoV-2) Screening With Specimen Pools: Time to Swim, or Too Deep for Comfort? Evaluation of sample pooling for screening of SARS CoV-2 Active Surveillance of Asymptomatic, Presymptomatic, and Oligosymptomatic SARS-CoV-2-Infected Individuals in Communities Inhabiting Closed or Semi-closed Institutions A robust pooled testing approach to expand COVID-19 screening capacity Sample pooling as a strategy for community monitoring for SARS-CoV-2 Effectiveness of sample pooling strategies for SARS-CoV-2 mass screening by RT-PCR: a scoping review Using viral load and epidemic dynamics to optimize pooled testing in resource-constrained settings Comparison of flocked and rayon swabs for collection of respiratory epithelial cells from uninfected volunteers and symptomatic patients Dry cotton or flocked respiratory swabs as a simple collection technique for the molecular detection of respiratory viruses using real-time NASBA Virological assessment of hospitalized patients with COVID-2019 Saliva or Nasopharyngeal Swab Specimens for Detection of SARS-CoV-2 Pooling saliva to increase SARS-CoV-2 testing capacity Rapid one-step detection of SARS-CoV-2 We would like to thank all members of the Institute of Virology and our colleagues of the Department I of Internal Medicine and the Institute of Medical Statistics and Computational Biology, University Hospital of Cologne for supporting our work. commercial, or not-for-profit sectors. We declare no competing interests. ☒ The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.☐The authors declare the following financial interests/personal relationships which may be considered as potential competing interests: