Improvement of SARS-COV-2 screening using pooled sampling testing in limited RT-qPCR resources.

BACKGROUND: COVID-19 is a worldwide pandemic representing the most challenging global health crisis currently. Screening tests availability are a problematic task due to resource-limited abilities of some countries using RT-qPCR technique for SARS-COV-2 detection.
OBJECTIVE: To cope with these health emergencies, in particular with this COVID-19 pandemic, states with low molecular diagnostic resources must optimize their capacity in molecular tests. We aimed to design a simple and effective strategy to improve inputs in the RT-qPCR tests as we attempted to check the financial advisability of using such an approach by calculating reduction rate of the test unit cost.
METHODS: The used RNA was taken from suspected Covid-19 positive people. Nasopharyngeal swabs were collected at Pasteur Institute Diagnostic Center, Constantine, Algeria, 2020. We have optimized a screening strategy by grouping 16 individuals per pool, without reducing the sensitivity of RT-qPCR.
RESULTS: A 1/16 dilution of a positive sample was a practical limit that does not require the use of robotic systems or mathematical modeling to construct the pools. The financial analysis of our strategy has shown that the costs can be reduced to 90 %. The pooled testing strategy that was proven in this study could be recommended to help COVID-19 containment in countries with low potential screening infrastructures using RT-qPCR technique by reducing the number of tests required to identify all positive subjects.

Introduction

COVID-19 Pandemic is a global challenging health crisis, affecting social and economic sectors. It weakened entire nations, mainly low technological resources nations (Nkengasong and Mankoula, 2020; Schellekens and Sourrouille, 2020). Diagnostic tests using RT-qPCR technology have become the benchmark for SARS-CoV-2 diagnosis essential for pandemic control according to the World Health Organization. Diagnostic kits market witnesses a huge pressure at the worldwide level, particularly in developed countries due to the massive screening of their population. Nonetheless, this situation deprived many other countries to afford these tests for screening implementation, which is not necessarily massive, as is the case in Algeria. In addition to the difficulty of supplying KITs, there is another hindrance specific to RT-qPCR technology as it is not widely used in our country.

To cope with this situation and to bridge the weak number of the used tests, we implemented a pooled sample testing strategy aiming to increase the national capacities of COVID-19 screening by optimizing the low number of thermal-cyclers and RT-qPCR available Kits. In its simplest form, the pooling test works by constituting a set of individual samples, in the case where the pool is COVID-19 negative, all the individuals are identified as negative and if the pool is positive, further tests would be necessary to identify infected individuals in the group (Dorfman, 1943; Olivier Gossner, 2020). This approach has been successfully used in the countries with high technological potential to deal with the lack of screening means (Mutzel et al., 2020; Shani-Narkiss et al., 2020), as it has been proposed as a massive screening strategy (De Salazar et al., 2020; Bilder and Tebbs, 2012).

Nevertheless, the approaches proposed in the literature do not take into account the specific realities to countries with limited access to molecular diagnostic tools, namely RT-qPCR Kits and thermal cyclers. To that end, we tested in local conditions several actions in recently published group tests and tried to answer key questions about this approach:

What is the size of a pool so that a single positive individual remains detectable (Eis-Hübinger et al., 2020)? When should the samples be grouped, would it be appropriate to do so after or before the RNA extraction step (Powers, 2011)?

In the other hand, we aimed to check the financial advisability of using a such approach by calculating reduction rate of the test unit cost. The answers provided by this work lead to a single objective to transpose this strategy in the requirements of countries with the same technological level as Algeria offers them the possibility to expand their potential for COVID-19 fighting using RT-qPCR.

Methods

The used RNA was taken from suspected Covid-19 positive people. Nasopharyngeal swabs were collected at Pasteur Institute Diagnostic Center, Constantine, Algeria, 2020. This study was approved by the “Ad’hoc ethical and review board committee of the national center of biotechnology research”. Given the deidentified nature of testing, individual patient consent was not required for this study.

At first, we tested RT-qPCR ability to detect a single positive sample diluted in a group of negative samples, without increasing the number of amplification cycles under standard conditions.

RNA extraction

All RNAs were extracted from nasopharyngeal samples; 200 μL of the universal transport media were used according to the conditions recommended by the manufacturer of the KIT (DAAN Gene Co., Ltd of Sun Yat-sen university China references of KIT DA0591).

Pooling

We used 12 RNAs extracted COVID-19 positive and 36 other RNAs from confirmed COVID-19 negative individuals. The COVID-19 status of all these samples has been previously and individually confirmed by RT-qPCR. Each of the 12 positive RNAs was grouped with negative RNAs according to different dilution factors (1/4, 1/8, 1/16 and 1/32). The RNAs were grouped at equal volumes (10 μl) in a 1,5 ml tube, then vortexed for 30 s and centrifuged at 800 g for 1 min.

RT-qPCR reaction

Was done under the conditions described by the KIT supplier, BGI biotechnologies (Wuhan Co Ltd China Catalog MFG030010). The detection threshold written on the used Kit, was of 100 copies / ml. RT-qPCR was performed on an ABI 7500 Applied Biosystems 7500 Real-Time PCR System device (ThermoFisher scientific USA) according to the program recommended by the KIT manufacturer. We used the same RT-qPCR conditions for the individual samples as for the pooled ones.

Ct values

Each obtained value in individual amplification was compared to that observed after the pooling. Ct manufacturer values were retained; the result was considered as positive if the value of the obtained Ct is ≤ 38. The machine background noise baseline (ABI7500-ThermoFisher scientific USA) was set as the individual samples obtained results and was not changed for the results analysis after pooling.

Results

Inclusivity

The observed Ct values following the amplification of the target ORF1ab-gene (KIT BGI) before or after pooling for each of the individuals are compared with each other and a statistical study was carried out to assess the performance of the detected RT-qPCR. Positive individuals were diluted to ¼, 1/8, 1/16, and 1/32 with negative Sars-Cov 2 ARN samples. All the Ct values after individual tests are below the positivity threshold recommended by the KIT manufacturer, namely < to 38. All the Ct values of the internal control of the amplification reaction (the β-actin gene) than either for individual or pooled samples were <31.

The Ct values in the 1/4 pool were one unit higher than those of the individual samples, while for the 1/8 and 1/16 dilution groups are 2–3 units higher, however, all the Ct values remain within the positivity threshold. In the group of 1/32, an increase in the observed Ct values was over 3 units making some samples negative 33 % (4/12).

The case of sample 2

At dilution ¼, the result displayed by the machine was not determined for the target gene and the internal control gene, therefore we suspect an internal concern with the reaction for this case.

Data performance analysis

In order to evaluate the performance of practical diluted samples for the massive PCR testing, statistical analyses have been achieved. The main test of this part was to find the stopping threshold of the diluted sample (Ct value). As described in method section; the mixture fraction started from dilution 1/4 to 1/32. We consider the obtained data as binary classification (Table 1 ).

Table 1

Results of tested mixtures before and after 1/4, 1/8, 1/16 and 1/32 dilutions.

Code	Positive sample before dilution	Dilution
		1/4	1/8	1/16	1/32
1	1	1	1	1	1
2	1	ND	1	1	0
3	1	1	1	1	0
4	1	1	1	1	1
5	1	1	1	1	1
6	1	1	1	1	1
7	1	1	1	1	1
8	1	1	1	1	1
9	1	1	1	1	1
10	1	1	1	1	0
11	1	1	1	1	0
12	1	1	1	1	1

1 = positive, 0 = negative ND no determined.

For the samples in each dilution, we first assigned the number 1 to positive test observation; in this case, the observation is true positive, because each mixture contains one positive sample (Ct Value <38). Second, we assigned the number 0 to negative observation result, in this case the observation is false-negative (Ct Value >38), since the mixture contain a positive sample. The performance metrics can be generated by drawing values from a 2 × 2 contingency table; the confusion matrix in Table 2 summarizes the results of different fractions.

Table 2

Statistical results and confusion matrix of diluted samples from 1/4 to 1/32.

Dilution	Precision	F-measure	Error	Recall		Tested as
					Position	Positive	Negative
1/4	100.00 %	0.96	8.33 %	91.67 %	Positive	11	1
					Negative	0	0
1/16	100.00 %	1	0.00 %	100.00 %	Positive	12	0
					Negative	0	0
1/32	100.00 %	0.80	33.33 %	66.67 %	Positive	8	4
					Negative	0	0

The performances of the results were all evaluated using the standard parameters for classification including Sensitivity and Precision (Fawcett, 2006; R Development Core Team, 2008).

Recall = Sensitivity = True Positive/ True Positive + False Negative

Precision = Confidence = True Positive / True Positive + False Positive

Models with high sensitivity have fewer false-negatives, and models with high specificity have fewer false-positives (Positive test value = precision). Recall and precision can be reported by parameters that measure the combination of both, like F-measure, which is the harmonic mean of recall and precision:

The classification metrics give a good idea about the overall tests performance. In the diluted mixtures from 1/4 to 1/16, the three parameters were good, with high sensitivity values varying from 91.97 % to 100 % and F-measure 0.96. The advantage of PCR-based diagnostic is precision. In our study the precision values of all tests were perfect with a 100 % value and were explained by the absence of false-positive observations for all tests. Concerning observation 2 in dilution 1/4, the test was not determined and the same result for internal control was displayed. This error was about 8.33 %, but the real error, in which the false-negative error happen is equal to values divided by the total number of fraction 1/4 to 1/16, so it was equal to 2.77 %. However, in 1/32 fraction dilution, the number of false-positive increased to 4 observations, by default, the sensitivity decreased to 66.67 %, in conclusion we recommend a dilution limit threshold of 1/16 fraction dilution (Table 3 ).

Table 3

Sensitivity of different fractions.

DILUTION	1/4	1/8	1/16	1/32
Sensitivity	91.67 %	100.00 %	100.00 %	66.67 %

We carried out a blind test of four (04) RNAs extracted from asymptomatic samples non-confirmed as Covid-19 positive. We diluted each of the samples in the 16th with RNAs from individuals tested negative for COVID-19. This test was carried out under the same conditions as above. The Ct values obtained for the four (04) tested samples did change the initial state of positivity. Nevertheless, an increase of 1–2 units of Ct values was observed in the pooled samples compared to those obtained individually.

Financial considerations

To determine the impact of financial considerations of this procedure; and whether the pooling method is a good procedure with important economic issues, we studied out the expected cost for financing this approach and evaluation of the total cost for the selected pool holding 16 patient samples. We noticed then a meaningful cost reduction comparing to individual test. We have proceeded with statistical language R (Van Domelen et al., 2018) and the function in the pooling package (De Winter, 2013), which enabled us to plot the relationship between total costs vs. pool size. Pooling works great in the two-sample t-test scenario (Shipitsyna et al., 2007), as it reduces the variance of each observation from to , where g is the pool size. To better illustrate the case, we supposed that we would study "costs vs. pool size" for d = 0.1775, where d is a numerical value designating the real modification in the group means, and the biomarker has a variance. We supposed that the global cost for 1000 tests RT-qPCR is about 15,000 in arbitrary values and the evaluation of total cost for pooling test is illustrated in Fig. 1 .

Fig. 1

Visualize Total Costs for Pooling Design as a Function of Pool Size. The figure shows a high-profile application of pooling method, in which the price is downscaled by reducing the number of assays by nearly 92 %.

Discussion

The growing demand for SARS-CoV-2 tests by RT-PCR has deprived some countries of access to a massive screening strategy, which has demanded a new approach so that the available laboratories can accommodate a large number of samples to analyze. Another limitation to our study was the low number of samples as well as the poor amount of literature as the virus appeared and mutated recently.

Nucleic acid testing allows the uses of sample pooling strategies, which have already been used in previous studies for a large number of pathogens (Mallapaty, 2020). Literature review on pooled test strategies in the case of COVID-19 discloses two main approaches: one purely mathematical (statistical), which seeks the precise estimation of the optimal size of the group beyond the search for the frequency of the positive target in this group (Bao et al., 2020).

This approach was made on a provisional basis and models situations that could be large and impossible to transpose to the specific features of certain countries. The other approach exposed methodologies with a large volume of samples, (using for example high-throughput automata) which could not be transposed to the reality of certain low-income countries. We believe that our results can help small laboratories meet the growing demand for SARS-CoV-2 tests and can be used as a full-scale experiment for their needs. The limited diagnostic test resources of the SARS-CoV-2 forced some laboratories to optimize their procedure to deal with the pandemic. We then proposed a simple and practical approach with groups of small numbers of individuals of 8 or 16 patients per test. The obtained results show that using RT-qPCR tests on pooled samples of 16 individuals, the reliability and reproducibility of the data (Ct values) is of 100 %. The experimental proposed protocols in our approach are easily transposable in an environment with low technological resources; in terms of laboratory consumables and equipment, available in developing countries such as Algeria.

Protocols for RNA RT-PCR testing of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) became available early in the pandemic, yet the infrastructure of testing laboratories is stretched and in some areas it is overwhelmed (Sardanelli et al., 2020) while other developing countries are still struggling against poor laboratory equipments.

Sardanell et al. data suggest that pooling of up to 30 samples per pool can increase test capacity with existing equipment and test kits to detect positive samples with sufficient diagnostic accuracy. They mentioned that borderline positive single samples might escape detection in large pools; these samples were noticed typically in convalescent patients 14–21 days after symptomatic infection. The pool size could accommodate different infection scenarios and be optimised according to infrastructure constraints as is the case in our country (Yelin et al., 2020).

Three studies (Lohse et al., 2020a;Deckert and Kyei, 2021; Abdalhamid et al., 2020) observed that larger pool size, of around 25–30 samples, can accurately detect a positive sample. Yelin et al. (Lohse et al., 2020a) arranged the negative samples in the pools in ascending numbers and one positive sample in each pool. Here, positive samples were consistently detected up to a pool size of 32. However, false negativity was detected in 10 %.9 (Lohse et al., 2020a).

Five studies noticed an increase in the threshold cycles (Ct) value upon pooling compared with the Ct detected during SARS-Cov-2 individual testing. Yelin et al. (Lohse et al., 2020a) observed that as the number of negative pooled samples increases (from 1:2 to 1:32), the amplified RNA reaches the threshold later, as expected from a diluted sample; but majority pools (up to 32-sample pools) reached the threshold, only one of the ten tested replicates, did not cross the threshold in pools of 32.9. (Lohse et al., 2020a;Deckert and Kyei, 2021; Abdalhamid et al., 2020; Deka and Kalita, 2020; Garg et al., 2020).

Garg et al. (Lohse et al., 2020b) reported that pools with atypical/low positive cases showed Ct value above 30, in most of the cases which on deconvoluted testing has lower Ct value below 30. In another study on specimen pooling, it was observed that pooling did not affect the sensitivity of detecting SARS‐CoV‐2 when the PCR cycle threshold (Ct) of the original specimen was lower than 35. Nonetheless, Lohse et al. (Mutesa et al., 2021) observed lower Ct values in some re-tested positive individual samples. They hypothesized that “the lower Ct values of pools than that of single samples were because of the carrier effect of the higher RNA content.” This is the major disadvantage of pooled testing, nevertheless, adding a few additional PCR cycles could be considered as a means to increase the detection rate of low-viral load samples.

Mutesa L et al. (27) have demonstrated pooling viability for group sizes up to 100 samples, showing that cost savings of a factor of nearly 20 can be achieved. They quantified the loss of sensitivity due to dilution and discussed a number of ways in which it may be mitigated for example, through frequently repeated group tests. These strategies could enable the use of larger pool sizes, bringing even greater cost savings at low prevalence. The most striking aspect of their approach is how rapidly the cost of testing a population can fall, pooled test sensitivity permitting, as the viral prevalence decreases. This should incentivise decision-makers to act firmly to drive the prevalence down through mass screening, contact tracing and isolation. As the viral prevalence is reduced, all aspects of this strategy become more and more affordable (27).

Conclusion

The pooled testing strategy that was proven in this study could be recommended to help COVID-19 containment in countries with low potential screening infrastructures using RT-qPCR technique by reducing the number of tests required to identify all positive subjects. Nonetheless further studies are required in order to agree and support our findings.

Limitations

The low number of samples as well as the poor amount of literature as the virus appeared and mutated recently.

Ethical approval

This study was approved by the “Ad’hoc ethical and review board committee of the national center of biotechnology research”. Given the deidentified nature of testing, individual patient consent was not required for this study.

Data availability

The data that support the findings of this study are available from the corresponding author upon request.

Author statement

All authors disclose any financial and personal relationships with other people or organizations that could inappropriately influence our work, as there is no interest to declare.

Declaration of Competing Interest

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.