Effectiveness of sample pooling strategies for diagnosis of SARS-CoV-2: Specimen pooling vs. RNA elutes pooling

Abstract

Purpose

The pandemic of SARS-CoV-2 or COVID-19 has hugely created an economic imbalance worldwide. With the exponential increase in the number of cases and to keep in check on the community transmission, there is high demand and acute shortage of diagnostic kits. The pooled-sample strategy turns out to be the promising strategy intended to determine the optimal testing for specimens with limited resources and without losing the test sensitivity and specificity. The study was performed with standard molecular biology graded lab equipment, FDA-approved COVID-19 RNA extraction, and SARS-CoV-2 tests kits.

Materials and methods

The study aims to comparatively analyze the pooling strategy of the naso-oropharyngeal specimen sample and RNA extracted from the same patient samples in the pool of 3,5, and 8 with no significant loss in test usability. Another primary focus of the study was detection of low or borderline SARS-CoV-2 positives in the pooling strategy. A total of 300 samples (240 positives and 60 negatives) were tested for 3, 5, and 8 pools of specimen samples and RNA elutes.

Results

The comparative analysis determined the sensitivity for three and five pool strategy to be above 98% and eight pool strategy to be 100%.

Conclusion

The RNA elutes pooling strategy concordance rate is better than that of specimen pooling with 100% specificity. Thus, in the substantial crisis of resources with the global pandemic, pooling approaches for SARS-CoV-2 can be practical in a low prevalence rate of 5%.

1. Introduction

The severe acute respiratory syndrome-related coronavirus (SARS-CoV-2) has been threat to global health care system, in terms of patient care, control and surveillance. To support effective public health measures, the ability to detect SARS-CoV-2 infection is top priority [1]. In the laboratory, COVID-19 analysis consists of three primary steps: viral inactivation and lysis, viral RNA extraction, and reverse transcription (RT)-PCR. With rapid spread and rising testing demand, a shortage of test reagents, particularly RNA extraction kits, has emerged as a major bottleneck as the pandemic spread [2].

The WHO has recommended the RT-PCR test as the gold standard and laboratory confirm method for SARS-CoV-2 infection. Although the disease spread slowly at first, by early 2020, and then exponentially, with the second and third waves. SARS-CoV-2 had an estimated reproduction number (R0) of 2.87, ranging from 2 to 6.3 in different countries, implying rapid secondary infection transmission [3]. Extensive testing required for detecting COVID-19 is proving difficult in the diagnosis of SARS-CoV-2. These require efficient resource allocation as well as prompt and accurate diagnosis. Hence pooled-sample testing techniques are required to screen a large population and speed up testing [4,5].

Robert Dorfman coined the term “pooling samples” in 1943, when he used serum samples to screen for syphilis while recruiting for World War 2. Sample pooling can save money and time [6]. Although the pooling of respiratory samples for RT-PCR has been reported to be effective in several diseases including HIV, it is not widely used in diagnostics. The method entails combining multiple samples into a single sample. If the result is positive, each sample in the pool is examined individually again; if the pool is negative, no further testing is required. [7, 8].

Pool testing can be implemented prior determining the pool size; as the RT-PCR test provide high sensitivity in parameters such as infection prevalence, sensitivity, specificity, PPV, negative predictive value (NPV), and LOD were successfully applied [9]. As a result, the goal of the study was to compare the pooling strategy of the naso-oropharyngeal specimen sample and extracted RNA from the same samples. The pooling technique was also used to examine the pattern of expression of low or borderline SARS-CoV-2 positives.

2. Materials and methods

Anonymous samples, excluding the personal identifiers like age or gender, were used for pooling process to assess increased testing competence and efficacy. Ethics approval has been obtained by theinstitutes scientific committee with letter no. No./MCR/ISC/FR/18 (65) dated September 16, 2021.

2.1. Sample collection and processing

The combined nasal and oral swabs were collected and transported in Viral Transport Media (VTM) by maintaining the proper cold chain. A total of 300 samples were tested, with 300 ​μl of each sample processed for RNA extraction using the Huwel Nucleic Acid Extraction Kit following the manufacturer's protocol [10].

2.2. Reverse transcriptase-polymerase chain reaction assay

Extracted RNA elute of 10 ​μl was used for RT-qPCR assay to detect SARS-CoV-2 RNA using nCoV Real-Time Detection kit based on Taqman probe real-time fluorescent PCR technology. The thermocycling protocol was followed as: 50 ​°C for 15 ​min; 95 ​°C for 3 ​min; 45 cycles of 95 ​°C for 5 ​s; 60 ​°C for 40 ​s. Roche thermocycler 96 Real-Time PCR system was used for PCR.

For detection of SARS-CoV-2: E and ORF1ab (RdRp) genes VIC and FAM channels were used respectively and interpretation was made according to the kit protocol. Samples with no gene expression or Ct's above 35 were negative (N), and the positive (P) samples were categorized based on their cumulative threshold value (Ct-value) as SP- strong positive (16–22), MP- moderate (23–29), and WP- (30–35) as weak positive.

2.3. Scheme of pooling

Two-stage Dorfamn's pooling algorithm was adopted. In the first step, samples were pooled in disjoint groups and tested and in second step, the pool tested positive were individually tested and no additional tests were perfomed for negative pools [7].

The VTM samples and RNA elute were pooled. The pooling strategy was developed in accordance with ICMR guidelines, as pooling of 5 samples was approved in regions with a prevalence rate of 2–5%.

https://www.icmr.gov.in/pdf/covid/strategy/Advisory_on_feasibility_of_sample_pooling.pdf.

Total 300 samples (240 positive and 60 negative) were tested individually and in pools. The strategy collectively formed 200 pools: 100 pools of 3 samples, 60 pools of 5 samples, and 40 pools of 8 samples (Fig. 1 ).

Fig. 1.

A diagrammatic model of the pooling strategy identifies all conceivable combinations impacting Ct score and weak positives to control the infection spread (WP= weak positive, SP= strong positive, MP= moderate positive, P= positive, and N= Negative).

The pool of 3 was further categorized into 10 different combinations of P and N in set of 10 as- (1) 2MP:1WP, (2) 1MP:2WP, (3) 3WP, (4) 3SP, (5) 1SP:2WP, (6) 1SP:1:WP:1MP, (7) 1WP:2N, (8) 2P:1N, (9) 2N:1P, and (10) 3N.

The pool of 5 was categorized into 10 combinations of P and N in set of 6 as follows- (1) 2MP:3WP, (2) 1MP:4WP, (3) 1 SP:2WP:2MP, (4) 5P, (5) 2P:3N (6) 1WP:4N, (7) 1MP:4N, (8) 1SP:4N, (9) 3WP:2N, and (10) 5N.

Lastly, the pool of 8 was categorized into 5 pools each of 8 combinations: (1) 1SP:7WP, (2) 8WP, (3) 4WP:4MP, (4) 1SP:7N, (5) 4MP:4N, (6) 4WP:4N, (7) 2WP:6N, and (8) 8N. The main idea was to identify all possible combinations affecting the Ct's, conduct an in-depth analysis of the Ct, and identify the missing WP samples due to pooling.

2.3.1. Sample pooling

Three samples 100 ​μl and five samples 60 ​μl each were pooled in different combinations of WP, MP, and SP and subjected to manual RNA extraction. The RT-qPCR assay was used to detect SARS-CoV-2 in a volume of 10 ​μl of extracted RNA.

2.3.2. RNA pooling

Similar to sample pooling, RNA was pooled in various combinations. The eluted RNA was pooled in volumes of 3.5, 2, and 1.25 ​μl in pools of 3, 5 and 8 RNA respectively.

These pooled samples and RNA extracts were tested, and the results of individual samples and pooled specimens were compared. The performance characteristics of pooled analysis such as sensitivity, specificity, PPV, and NPV were also analyzed as described in Fig. 2 [11].

Fig. 2.

Tabulated form of the method used to predict the Positive Predictive Value (PPV) and Negative Predictive Value (NPV) as described by Zimmerman [11].

2.4. Statistical analysis

The SPSS 16.0 software was used to analyze statistical differences. All experiments performed in triplicates were expressed as mean ​± ​s.d. All the samples were double-tested individually to confirm their status.

3. Results

The retention of sufficient sensitivity is a requirement of pooled RNA extraction and RT-PCR of any nucleic acid-based test. The stratified testing includes a 5% experimental prevalence rate, an assay with a lower limit of detection (LOD) of 1–5 ​μl RNA copies/μl, an assay sensitivity of 98%–100% specificity, a two-staged pooling algorithm, and a pool size range of 3, 5, and 8 samples.

This study's pooling strategy differs from others in that it compares the efficiency and cost-effectiveness of specimen samples and RNA elutes (as per the guidelines of ICMR). The presumption was to pool multiple specimens and process them as a single RNA in a single PCR test without significant loss of sensitivity. When compared to individual testing, the presented viewpoint of pooling strategy of both sample and RNA elutes results in a six-fold cost reduction per sample.

3.1. Impact of Ct in 3 sample pooling

The obtained Ct value for 100 pools of 3 specimens showed 93% concordance, with seven pools missing the MPs and WPs. “Concordance” being the similarity between the findings of individual samples and the pooled results. In comparison to individual Cts, pools containing two or more SPs in any combination saw a 2.3–2.9% increase in Ct value. Another discovery was that any pool with a single SP or MP detected as positive with a 0.2–0.4 Ct differential had 100% concordance; pools with one P and other Ns have a 90% concordance rate, lowering Ct and only one false positive pool was discovered among all of the negative samples Table 1 .

Table 1.

Comparison of pool data and concordance.

Pooling strategy	No. of samples		No. of pool	No. of positive pool	No. of deviated pool	Concordance (%)
Specimen pooling	300	Pool of 3	100	90	7	93
		Pool of 5	60	55	6	90
RNA elutes	300	Pool of 3	100	90	5	95
		Pool of 5	60	55	5	92
		Pool of 8	40	32	4	90

The RNA elute pool, on the other hand, had a concordance rate of 95%, with only 5% missing the WPs in the combination of MP and WPs. Pools with two or more positives in any combination increased by 2–3 Cts, compared to individual Cts. Pools with one P and the rest Ns have a 99% concordance rate with Ct decreases.

3.2. Impact of Ct in 5 sample pooling

The results of five samples were comparable to the three-sample pool. Pools containing two or more SP or MP show an increase in overall Ct of 1.35–2.2. Because of dilutions with other weak or negative samples, WPs results were implicated. Pools with MP and WP combinations were found negative (Ct ​> ​35) when compared to deconvoluted ones, and 1 false-positive pool was reported.

A 92% concordance was observed in RNA elute pooling by same method. Five pools containing WP:N combinations were detected negative. These borderline WPs had a 2–3 Ct elevation, resulting in Ct ​> ​35 or negatives with a sigmoid graph (see Table 1).

3.3. Impact of Ct in 8 sample pooling

RNA elutes were tested using a strategy of pooling 8 samples in 8 different combinations and 40 such pools were studied having 5 pools in each combination. The study reported a 90% concordance. Borderline WPs with Ct of 34–35 were missed out in the pooling but deconvoluted samples show comparable results on re-testing. Several practical factors, such as reproducibility of RNA extraction and RT-PCR, freeze-thawing cycles etc, as well as theoretically, an increase in Ct by 3 cycles affect the results of 8 samples pooling. 90% concordance delineated 4 negative pools containing WPs and negatives; while pools with SP and MP were reported positive with ∼4–6 Ct raise. MPs were not missed, but an increase in Ct value was observed in this strategy (see Table 1).

An elevation in Ct value was observed when the average Ct values of pool strategy compared with individual SARS-CoV-2 positive samples. Low Ct value denotes high viral load in the collected sample and vice versa (Table 2 ). The Ct value for pool of 3 ranges from 15.34 to 35.00 for the E gene and 17.23–35.00 for the RdRP gene. For the pool of 5, Ct ranges from 15.12 to 35.00 for the E gene and 16.89–35.00 for RdRP. And for the pool of 8, Ct ranges from 14.12 to 35.00 for the E and 15.89–35.0 for RdRP.

Table 2.

Comparing the Ct value of genes for SARS-CoV-2 detection in differential pooling strategies.

Pooling strategy	E-gene		RdRP gene
	Pooling	ID	Pooling	ID
POOL 3	25.0	26.3	26.1	27.1
POOL 5	25.0	26.1	25.9	26.8

ID – Individual sample.

3.4. Effect of dilution

The dilution of samples in specimen pooling increased the likelihood of missing WPs (Ct 34–35), whereas RNA pooling reduced the likelihood of missing those. This sensitivity loss is observed when dealing with WPs, whereas SPs and MPs are detectable even when pooling 8 samples. With a prevalence rate of 5%, the pooling strategy, whether RNA or specimen, proves to be effective in mass testing, surveillance, and resource conservation.

3.5. Calculation of specificity, sensitivity, positive and negative predictive value

In pool testing, the sensitivity is the likelihood that an individual sample will be declared positive. The sensitivity was calculated using the distribution of Ct values in any individual sample's routine test. The ability of pooling to identify the accurate non-infected samples or true negatives is termed as specificity for pool testing. The percentage of cases with positive test findings that are already patients is known as positive predictive value and the percentage of cases yielding negative test results that are already healthy is known as the negative predictive value (Fig. 2).

When all discordance were considered, 3 specimen pool strategy sensitivity, specificity, PPV, and NPV were 98.9% (95% CI: 25.4 to 26.6 Ct), 93.3%, 93%, and 99% respectively. For 5 specimen pool strategy, sensitivity, specificity, PPV, and NPV were 98.1% (95% CI: 26.3 to 27.4 Ct), 90.7%, 90%, and 98.3% respectively (see Table 2).

When all discordance were considered, 3 RNA elute pool strategy sensitivity, specificity, PPV, and NPV were 98.9% (95% CI: 24.4 to 25.8 Ct), 95.2%, 95%, and 99% respectively. For 5 RNA elute pool strategy sensitivity, specificity, PPV, and NPV were 100% (95% CI: 24.5 to 26.0 Ct),92.3%, 91.66%, and 100%. Lastly, 8 RNA elute pool strategy sensitivity, specificity, PPV, and NPV were 100% (95% CI: 26.4 to 27.5 Ct), 90.3%, 89.33%, and 100% respectively (see Table 2).

4. Discussion

This novel study shows that pooling can help with faster PCR results even when the prevalence of SARS-CoV-2 is as low as 5%. RNA pooling had a higher sensitivity and specificity than specimen pooling, with a sensitivity of around 99% and above. High-throughput pooling strategies and RT-PCR-based methods have previously been used to diagnose large populations and current situations such as pandemics [[12], [13], [14]]. Strategy of merging the samples is a cost-effective method to increase the speed of infection screening.

According to the study, the Ct value shifts to a lower range when two or more positive samples are present in a pool. Dilution can also be seen with an increase in Ct value in some WPs as the ratio of negative samples in the pool increases [15]. Similar observations were reported by others Abdalhamid et al., Yelin et al., Gupta et al. on comparing the individual and pooled COVID-19 positive samples and RNA elute [1,14,16].

Numerous studies using various pool sizes could detect even a single positive sample in the pool. Positive samples from up to 32 pools were accurately detected by a large pool size of up to 63 samples per pool. [14,17,18]; however, 10% of false negativity was observed by Yelin and co-workers. Lohse et al ., and Deckert et al., reported positive samples in a pool of 30 and 25 samples, respectively with low to moderate prevalence. A single positive sample can be detected in studies with a small sample size (5–10 samples/pool) and a low to moderate prevalence rate [16]. Hogan discusses a similar study to detect community transmission that screened 292 samples in 9–10 pools with only one false result [19]. A noticeable fact in this study was that the 1 positive sample used was either SP or MP, with a Ct of less than 29.

A significant reduction in total tests was observed depending on pool size and prevalence. With a prevalence rate of <5%, the pooling strategies of 3,5 and 8 reduced the number of tests per individual to <0.33, <0.2, and <0.13, respectively. Similar studies by Abdalhamid et al., Sinnot- Armstrong et al., and Shai-Narkiss et al., also reported a significant reduction of 53–93% in the total test done with a prevalence rate of ≤5% [4,16,[20], [21], [22]].

Strength of this study lies in the considerations given to possible combinations in a pool where there is no room for omitting WP samples, which many studies overlook. Maximum precision of results without affecting the sensitivity and specificity of the individual sample is required for the pooling strategy. Our pooling strategy not only precisely validates the samples, but also generates new statistics for two distinct strategies.

5. Conclusion

According to the results of the above pooling strategy, as the pandemic continues, the pooling strategy can be actively implemented in settings with a 5% prevalence rate. The advantage of sample pooling strategies is that they lower molecular diagnosis costs and allow for faster patient screening, but they have slightly lower specificity when compared to RNA elutes, which have 100% specificity. With ongoing vaccination and lower prevalence rates in various Indian states, a pooling strategy is the most efficient way to increase testing. Pooling RNA elutes proves more specific and sensitive. RNA pooling has a higher concordance rate than specimen pooling and produces fewer false-positive results. PCR is at least ten times more expensive than RNA extraction. Finally, all ideas for a pandemic-fighting pooling strategy with validation in their settings, PCR kits, and RNA extraction based on the prevalence rate are worth developing.

CRediT author statement

Vijaylakshmi Jain: Conceptualization, Methodology, Investigation, Visualization, Writing-Original Draft. Nikita Sherwani: Conceptualization, Supervision. Niza Monga: Writing, Review & Editing, Formal analysis. Aparna Sahu: Review & Editing.

Funding

This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.

Ethics approval

Ethics approval has been obtained by the college scientific committee with letter no. No./MCR/ISC/FR/18 (65) dated September 16, 2021.

Conflict of Interest

The authors declare no conflict of interest.

Appendix A. Supplementary data

The following is the Supplementary data to this article:Mutimedia component 1.