A comparison of SARS-CoV-2 wastewater concentration methods for environmental surveillance

Abstract

Wastewater¹ surveillance of SARS-CoV-2 may be a useful supplement to clinical surveillance as it is shed in feces, there are many asymptomatic cases, and diagnostic testing can have capacity limitations and extended time to results. Although numerous studies have utilized wastewater surveillance for SARS-CoV-2, the methods used were developed and/or standardized for other pathogens. This study evaluates multiple methods for concentration and recovery of SARS-CoV-2 and seeded human coronavirus OC43 from municipal primary wastewater and/or sludge from the Greater Seattle Area (March-July 2020). Methods evaluated include the bag-mediated filtration system (BMFS), with and without Vertrel™ extraction, skimmed milk flocculation, with and without Vertrel™ extraction, polyethylene glycol (PEG) precipitation, ultrafiltration, and sludge extraction. Total RNA was extracted from wastewater concentrates and analyzed for SARS-CoV-2 and OC43 with RT-qPCR. Skimmed milk flocculation without Vertrel™ extraction performed consistently over time and between treatment plants in Seattle-area wastewater with the lowest average OC43 C_q value and smallest variability (24.3; 95% CI: 23.8-24.9), most frequent SARS-CoV-2 detection (48.8% of sampling events), and highest average OC43 percent recovery (9.1%; 95% CI: 6.2-11.9%). Skimmed milk flocculation is also beneficial because it is feasible in low-resource settings. While the BMFS had the highest average volume assayed of 11.9 mL (95% CI: 10.7-13.1 mL), the average OC43 percent recovery was low (0.7%; 95% CI: 0.4-1.0%). Ultrafiltration and PEG precipitation had low average OC43 percent recoveries of 1.0% (95% CI: 0.5-1.6%) and 3.2% (95% CI: 1.3-5.1%), respectively. The slopes and efficiency for the SARS-CoV-2 standard curves were not consistent over time, confirming the need to include a standard curve each run rather than using a single curve for multiple plates. Results suggest that the concentration and detection methods used must be validated for the specific water matrix using a recovery control to assess performance over time.

Keywords: Environmental surveillance; SARS-CoV-2; Wastewater.

Graphical abstract

Fig. 1

Effective volume assayed for and percent recovery for each method. A) The effective volume assayed is the proportion of the original wastewater sample assayed by RT-qPCR. The BMFS, with and without Vertrel, have the largest average effective volume assayed per reaction. PEG precipitation and skimmed milk flocculation, with and without Vertrel, have the smallest average effective volume assayed per reaction. B) The percent recovery is calculated using the standard curves for the RT-qPCR assay generated with each experimental run and the Cq value for each undiluted sample. Skimmed milk flocculation has the highest average percent recovery, followed by skimmed milk flocculation with Vertrel extraction. Both BMFS methods had the two lowest average percent recoveries. All methods tested had some non-detections in the undiluted sample except PEG precipitation and BMFS with Vertrel extraction. C) Percent recovery descriptive statistics by method.

Fig. 2

OC43 RT-qPCR control charts. Average Cq values were calculated for each method by averaging across treatment plants and time. The UWL and LWL, or upper and lower warning limits, for each method were calculated by adding or subtracting, respectively, the standard deviation from the average Cq. The UCL and LCL, or the upper and lower control limits, for each method were calculated by adding or subtracting, respectively, three times the standard deviation from the average Cq. Anything detected at or above a Cq of 40 was considered a non-detection. All samples that had non-detections by RT-qPCR in the undiluted samples reported here had detection in the 10⁻¹ dilution. BMFS without Vertrel extraction (A) has a lower average Cq compared to BMFS with Vertrel extraction (B), but has a substantially larger range of data. Skimmed milk flocculation without Vertrel extraction (C) and with Vertrel extraction (D) have similar average Cq's, control limits, and warning limits. Ultrafiltration (E) had a similar average Cq to both skimmed milk methods, but had a larger variability in the data and fewer detections in the undiluted samples. PEG precipitation (F) had a low average Cq and variability around the mean, but only one treatment plant was tested with this method and it is therefore not directly comparable.

Fig. 3

Limits of detection and quantification for SARS-CoV-2 assays. The limit of detection and limit of quantification for each SARS-CoV-2 assay is dependent on the water matrix. Greater inhibition of the assays were seen when standard curves were prepared using skimmed milk wastewater extracts as the diluent (B, D, F) than when standard curves were prepared using nuclease free water as the diluent (A, C, E), as seen by decreased efficiency and higher limits of detection. In panels D and F, the single vertical line indicates the LOD and LOQ are on top of each other.

Fig. 4

Percent positivity for SARS-CoV-2: samples were considered positive if the C_q ≤ 40. Direct sludge extraction had highest percent positive for N1 and N2 but had fewer samples. N3 had the highest percent positive in skimmed milk flocculation with Vertrel™ extraction. All three assays had roughly 30% positive with skimmed milk flocculation.

Fig. 5

Presumptive positive for SARS-CoV-2 by method. A sample was considered “presumptive positive” if at least one of the SARS-CoV-2 assays had a C_q ≤ 40 in either the non-diluted or 10⁻¹ dilution reaction.

Fig. 6

SARS-CoV-2 detection by wastewater treatment plant. Percent positivity includes undiluted and 10⁻¹ diluted RT-qPCR assays for each method and treatment plant. Detection of SARS-CoV-2 varied by method for each treatment plant.