key: cord-0696497-eggi4lgg
authors: LaTurner, Z. W.; Zong, D. M.; Kalvapalle, P.; ReyesGamas, K.; Terwilliger, A.; Crosby, T.; Ali, P.; Avadhanula, V.; Hernandez Santos, H.; Weesner, K.; Hopkins, L.; Piedra, P. A.; Maresso, A.; Stadler, L.
title: Evaluating recovery, cost, and throughput of different concentration methods for SARS-CoV-2 wastewater-based epidemiology
date: 2020-11-30
journal: nan
DOI: 10.1101/2020.11.27.20238980
sha: 5a7f99472427faae8c59b8571a1fdd3648f467f4
doc_id: 696497
cord_uid: eggi4lgg

As the COVID-19 pandemic continues to affect communities across the globe, the need to contain the spread of the outbreaks is of paramount importance. Wastewater monitoring of the SARS-CoV-2 virus, the causative agent responsible for COVID-19, has emerged as a promising tool for health officials to anticipate outbreaks. As interest in wastewater monitoring continues to grow and municipalities begin to implement this approach, there is a need to further identify and evaluate methods used to concentrate SARS-CoV-2 virus RNA from wastewater samples. Here we evaluate the recovery, cost, and throughput of five different concentration methods for quantifying SARS-CoV-2 virus RNA in wastewater samples. We tested the five methods on six different wastewater samples. We also evaluated the use of a bovine coronavirus vaccine as a process control and pepper mild mottle virus as a normalization factor. Of the five methods we tested head-to-head, we found that HA filtration with bead beating performed the best in terms of sensitivity and cost. This evaluation can serve as a guide for laboratories establishing a protocol to perform wastewater monitoring of SARS-CoV-2.

CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. Community Survey (U.S. Census Bureau, 2019). Composition data were reported from samples 158 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) preprint

The copyright holder for this this version posted November 30, 2020. ; https://doi.org/10.1101/2020.11.27.20238980 doi: medRxiv preprint taken between October 5, 2020 to October 6, 2020. All data was provided by the City of 159 Houston (Houston Public Works and Houston Health Department). 160 161 At the end of the sampling period, all samples were transported on ice to a central processing 167 The different concentration methods are depicted in Figure 1 . Concentration occurred the day 168 following sample collection (Oct. 7). PEG concentration began the day of sample collection to 169 allow the samples to sit overnight (Oct. 6). Technical replicates were performed in triplicate for 170 each concentration method. The direct extraction, HA filtration with bead beating, and 171 ultrafiltration concentration methods were completed at Rice University. The resulting 172 concentrates were immediately transported to Baylor College of Medicine on ice for extraction. 173 HA filtration with elution and PEG methods were completed at Baylor College of Medicine. 174 Concentration methods were split between labs to reduce processing burden and because each 175 lab had more experience with their respective methods. Concentrates were stored at 4°C until 176 extraction.

. CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) preprint to the bottle. The solution was then inverted, gently shaken by hand, and allowed to precipitate 221 overnight at 4°C. The following day, the sample was centrifuged for 30 minutes at 16,900 g and 222 .

CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. Microcentrifuge tubes containing the sample were weighed to calculate total volume of 243 concentrate.

. CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. samples. Primer and probe information can be found in Table S1 . Reaction mixes were prepared 265 on ice according to the composition outlined in Table S2 for N1 and N2, and Table S3 for

CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) preprint

The copyright holder for this this version posted November 30, 2020. ; https://doi.org/10.1101/2020.11.27.20238980 doi: medRxiv preprint BCoV. RNA template for BCoV was diluted 50x to attain a concentration within the 267 quantifiable range of the ddPCR equipment. Thermocycling conditions are outlined in Table S6 . 268 After thermocycling, samples were held at 4°C for no longer than 12 hours until being read on 269 the QX200 Droplet Reader (18644003, Bio-Rad). Droplet data was analyzed on the QuantaSoft Primer and probe information can be found in Table S1 . Reaction mixes were prepared on ice 277 according to the composition outlined in Table S4 . RNA template for pMMoV was taken from 278 the 50x dilutions created for measuring BCoV to conserve undiluted extract. Thermocycling 279 conditions are outlined in Table S7 . qPCR data was analyzed on the QuantStudio Design and CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) preprint

The copyright holder for this this version posted November 30, 2020. ; https://doi.org/10.1101/2020.11.27.20238980 doi: medRxiv preprint 2.6.3 Limit of Quantification (LoQ) 288 The LoQ for ddPCR was defined as 3 positive droplets per 10,000 total droplets generated by the The LoQ for RT-qPCR was determined to be 0.69 gene copies/µL RNA template for a reaction 294 setup with 4 µL of RNA template. This was the concentration of the lowest standard used in our 295 calibration curve. pMMoV was the only gene target measured through RT-qPCR, and all values 296 were significantly above this limit. Concentration factors (Table S8 -S13), average percent 297 recovery, and unit conversions were then used to convert this raw LoQ to an effective LoQ 298 associated with each concentration method (Eqn. 1).

Measurements below the LoQ may indicate presence of gene targets but are not reliably accurate 300 measurements of the concentration. 

We first evaluated the effectiveness of five concentration methods at detecting and quantifying 305 the SARS-CoV-2 RNA in wastewater samples. We obtained 24 hour composite wastewater 306 samples from six different WWTPs in Houston covering a range of influent flow rates, 307 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) preprint

The copyright holder for this this version posted November 30, 2020. ; https://doi.org/10.1101/2020.11.27.20238980 doi: medRxiv preprint population sizes, and wastewater compositions (Table 1) . We also created a negative control 308 sample that contains only DI water that had been spiked with the BCoV surrogate. We applied 309 each of the five different concentration methods to each sample in triplicate, extracted the RNA 310 of the SARS-CoV-2 virus, and quantified the concentration of CDC target N1 and N2 using 311 digital droplet PCR (ddPCR). We then back calculated the concentration of the virus in terms of 312 copies of the virus per liter of wastewater (Figure 2A, B) . 313 The concentration methods use different mechanisms to concentrate SARS-CoV-2. HA filtration CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) preprint

The copyright holder for this this version posted November 30, 2020. ; https://doi.org/10.1101/2020.11.27.20238980 doi: medRxiv preprint All five of the tested concentration methods were able to detect the presence of SARS-CoV-2 331 RNA, but varied significantly in viral RNA titer. Direct extraction yielded the highest apparent 332 concentration of SARS-CoV-2 RNA across all wastewater samples. In contrast, PEG had 333 consistently lower signal for SARS-CoV-2 RNA. Both methods involving HA filtration (with 334 bead beating and with elution) yielded similar concentrations that were about a half log lower 335 than direct extraction. In all cases, the resulting concentration of N1 and N2 per liter of 336 wastewater was highly dependent on the concentration method used, more so than from which 337 WWTP the sample came from. This suggests that the true concentration of SARS-CoV-2 RNA 338 from all six WWTPs was roughly the same. It is worth noting that these measurement systems 339 may be detecting free SARS-CoV-2 RNA along with intact SARS-CoV-2 virus particles, and 340 that each concentration method may have a different ability to measure that free SARS-CoV-2 341 RNA.

Direct extraction yielded the highest concentrations of genome copies per L of wastewater due to 343 the fewest losses associated with concentration and the largest concentration factor applied.

When the results are shown in terms of the raw data, copies of N1 or N2 per µL of RNA 345 template, direct extraction had the lowest raw concentrations of viral RNA and many data points 346 were below the LoQ (as shown by the line in (Figure 2C, D) . Furthermore, the only method that 347 yielded results that were consistently above the LoQ for all WWTPs and both targets was HA 348 filtration with bead beating. Because the LoQ is constant in terms of copies of RNA per µL of 349 RNA template, the method that yields the highest raw concentration of RNA per µL of template 350 is the one that produces quantifiable signal. Therefore, when evaluating concentration methods, a 351 key metric for consideration is not the genome copies per liter wastewater, but the copies per µL 352 of RNA template. In the set of methods we evaluated, we found that HA filtration with bead 353 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) preprint

The copyright holder for this this version posted November 30, 2020. ; https://doi.org/10.1101/2020.11.27.20238980 doi: medRxiv preprint beating performed the best in terms of having the highest raw genome copies per microliter of 354 RNA template and was thus consistently able to quantify SARS-CoV-2 RNA in all wastewater 355 samples tested.

One important decision in SARS-CoV-2 RNA concentration that may impact sensitivity, 357 reproducibility, and variability is the volume of input wastewater. As input volume increases you 358 are left with a higher concentration of virus RNA in a concentrate of the same volume. is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) preprint The startup costs reflect equipment that was unique to each concentration method. Standard lab 386 equipment required by all concentration methods, such as pipettes and PPE, were not included in 387 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) preprint

The copyright holder for this this version posted November 30, 2020. ; https://doi.org/10.1101/2020.11.27.20238980 doi: medRxiv preprint these calculations. Additionally, RNA extraction costs were not included, except in the case for 388 the price of a bead beater, which is necessary to quantify RNA when using the HA filtration with 389 bead beating method. The largest factor contributing to startup costs in all cases was centrifuges.

The differences in startup costs was generally due to differences in cost of the specific centrifuge 391 required. A centrifuge is not necessary to perform the HA filtration with bead beating method, Table S19-S23. 410 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this this version posted November 30, 2020. ; https://doi.org/10.1101/2020.11.27.20238980 doi: medRxiv preprint 411 Throughput time was also assessed as different groups conducting WBE for SARS-CoV-2 have 412 varying numbers of samples and unique requirements for turnaround times needed to report data. 413 All concentration methods included a centrifugation step to remove solids in their throughput 414 time. The PEG method had the lowest throughput, largely because of the precipitation step, 415 taking at least 3 hours longer to process a batch of samples than the other methods evaluated. In 416 our study, PEG was precipitated overnight to lower the burden on lab workers, but our Table S24 . Table 2 . 432 PEG had the highest concentration factor of 300, leading to the lowest LoQ, while direct 433 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this this version posted November 30, 2020. ; https://doi.org/10.1101/2020.11.27.20238980 doi: medRxiv preprint extraction has the highest LoQ because there was no concentration occurring. It is important to 434 note that this LoQ does not take into account losses incurred by the different concentration 435 methods. The optimal method will balance a low LoQ with a high recovery leading to raw 436 concentrations of gene targets being significantly above the raw LoQ. We accounted for recovery 437 by averaging percent recovery of BCoV for the different methods and incorporating that average 438 into an effective LoQ. It should be noted that BCoV has yet to be identified as an optimal 439 surrogate for SARS-CoV-2 but can still provide valuable information in this context. The Table 2 ]. The effective LoQ is directly related to the sensitivity of 445 the concentration method, which is a critical factor in being able to reliably quantify SARS-CoV-446 2 in wastewater samples, especially when there is relatively low community prevalence.

Lowering the effective LoQ relative to direct extraction is the essential reason why we include a 448 concentration step, because it makes lower concentrations of virus more reliably quantifiable. CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) preprint

The copyright holder for this this version posted November 30, 2020. ; https://doi.org/10.1101/2020.11.27.20238980 doi: medRxiv preprint to the expected concentration, it is possible to determine the overall recovery of the process 479 control and loss of the process control during sample processing. If the process control has been 480 validated, and is known to behave in a way similar to the target component, then these losses can 481 be incorporated into the measured concentration of the target component to estimate a "true" 482 measure of the target component in the sample. Process controls can also be used simply as 483 positive controls to ensure that nothing went awry during sample processing and analysis. hemagglutinin-esterase (HE) glycoprotein (Saif, 2010) . Apart from structural similarity, BCoV is 501 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) preprint

The copyright holder for this this version posted November 30, 2020. ; https://doi.org/10.1101/2020.11.27.20238980 doi: medRxiv preprint also easily obtainable in an attenuated form from a common cattle vaccine and poses a low 502 health risk to humans. Interestingly, DI water controls spiked with BCoV showed significantly different levels of 518 recoveries compared to wastewater samples concentrated via the same method. One potential 519 explanation for this could be rupture of the viral particles due to a large difference in osmotic 520 pressures across viral capsid/envelope in DI water. Rupture may not occur in wastewater 521 samples due to a significant amount of dissolved compounds reducing differences in osmotic 522 pressure inside and outside of the viral particle. Thus, the disparate concentration methods may 523 have different effects on ruptured versus unruptured virus.

. CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) preprint

The copyright holder for this this version posted November 30, 2020. ; https://doi.org/10.1101/2020.11.27.20238980 doi: medRxiv preprint

There are many factors that affect wastewater concentration of SARS-CoV-2 RNA between 526 excretion in feces and quantification in the lab that potentially confound translation of 527 concentration to community prevalence. While a process control, like BCoV, can be used to 528 account for factors during the measurement process (i.e. between sampling and quantification), a 529 normalization factor attempts to account for factors during the measurement process and 530 additional upstream factors, like dilution in the sewer system. The variability of measurements between each concentration method was determined using the 565 coefficient of variance (CV) for BCoV and pMMoV ( Table 2 ). The lower the CV, the lower the 566 variability and the higher confidence one can have in a measured value. Further, if CV is low 567 enough, it may be reasonable to reduce replicates (i.e. from triplicates to duplicates) to save on 568 cost and throughput. The CV was measured for each WWTP for a particular method and then 569 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) preprint

The copyright holder for this this version posted November 30, 2020. ; https://doi.org/10.1101/2020.11.27.20238980 doi: medRxiv preprint averaged for all WWTPs in the method to get the CV of the whole method. The PEG method 570 showed the highest average CV, while the HA filtration with bead beating method had a 571 generally low CV for both BCoV and pMMoV. As of now, it is not clear what causes the 572 difference in variability between each concentration method. We chose not to directly 573 incorporate CV of N1 and N2 gene targets because a large proportion of measurements were 574 below the LoQ. Therefore variability in N1 and N2 would be more attributed to the 575 quantification procedure that was used to establish the LoQ as opposed to variability caused by 576 the particular concentration method. Between BCoV and pMMoV, the BCoV CV is likely a 577 better indicator of a potential SARS-CoV-2 CV due to BCoVs higher structural similarity to 578 SARS-CoV-2 than pMMoV. more work needs to be done to determine the best process controls and normalization factors for 591 SARS-CoV-2. The WBE for SARS-CoV-2 community needs to identify process controls and 592 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) preprint CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) preprint

The copyright holder for this this version posted November 30, 2020. ; https://doi.org/10.1101/2020.11.27.20238980 doi: medRxiv preprint . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) preprint

The copyright holder for this this version posted November 30, 2020. ; https://doi.org/10.1101/2020.11.27.20238980 doi: medRxiv preprint 

WWTP reported in gene copies/μL RNA template. WWTP are A-E, DI is deionized water, and NTC is no template control.

. CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) preprint

The copyright holder for this this version posted November 30, 2020. ; https://doi.org/10.1101/2020.11.27.20238980 doi: medRxiv preprint

Development of 638 wastewater pooled surveillance of SARS-CoV-2 from congregate living settings

Genome 643 sequencing of sewage detects regionally prevalent SARS-CoV-2 variants

COVID-19 surveillance in Southeastern Virginia 647 using wastewater-based epidemiology

Computational analysis of SARS-CoV-2/COVID-19

surveillance by wastewater-based epidemiology locally and globally: Feasibility, 651 economy, opportunities and challenges

Identification of multiple potential viral 681 diseases in a large urban center using wastewater surveillance

684 Evaluation of SARS-CoV-2 viral RNA in fecal samples

Preliminary Study of Sars-Cov-2 Occurrence in Wastewater in the Czech Republic

Reproducibility and 692 sensitivity of 36 methods to quantify the SARS-CoV-2 genetic signal in raw wastewater: 693 findings from an interlaboratory methods evaluation in the U

Bovine Respiratory Coronavirus

First detection of SARS-CoV-2 RNA in wastewater in North 706 America: A study in Louisiana

Wastewater Analysis of SARS-CoV-2 as a Predictive Metric of 712

Positivity Rate for a Major Metropolis

Applicability of polyethylene glycol precipitation 715 followed by acid guanidinium thiocyanate-phenol-chloroform extraction for the detection 716 of SARS-CoV-2 RNA from municipal wastewater

Rapid Scaling Up of Covid-19 Diagnostic Testing in 720 the United States -The NIH RADx Initiative

Detection of SARS-CoV-2 in raw 725 . CC-BY-NC-ND 4.0 International license It is made available under a and treated wastewater in Germany -Suitability for COVID-19 surveillance and 726 potential transmission risks

Virological 731 assessment of hospitalized patients with COVID-2019

Wastewater Are Higher than Expected from Clinically Confirmed Cases

Rapid 739 bacteriophage sedimentation in the presence of polyethylene glycol and its application to 740 large-scale virus purification

Survivability, Partitioning, and 743 Recovery of Enveloped Viruses in Untreated Municipal Wastewater