Building-level wastewater monitoring for COVID-19 using tampon 1 swabs and RT-LAMP for rapid SARS-CoV-2 RNA detection 2 3

Building-level wastewater monitoring for COVID-19 using tampon 1 swabs and RT-LAMP for rapid SARS-CoV-2 RNA detection 2 3 Aaron Bivins1,2, Megan Lott3, Marlee Shaffer1, Zhenyu Wu1, Devin North1, Erin K. Lipp3, Kyle 4 Bibby1,2* 5 6 1 Department of Civil & Environmental Engineering & Earth Sciences, University of Notre Dame, 7 156 Fitzpatrick Hall, Notre Dame, IN 46556 8 2 Environmental Change Initiative, University of Notre Dame, 721 Flanner Hall, Notre Dame, IN 9 46556 10 3 Department of Environmental Health Science, University of Georgia, 150 East Green Street, 11 Athens, GA 30602 12 13 *kbibby@nd.edu 14 15 Abstract 16 Community-level wastewater monitoring for severe acute respiratory syndrome coronavirus 2 17 (SARS-CoV-2) RNA has demonstrated useful correlation with both coronavirus disease 2019 18 (COVID-19) case numbers and clinical testing positivity. Wastewater monitoring on college 19 campuses has demonstrated promising predictive capacity for the presence and absence of 20 COVID-19 cases. However, to date, such monitoring has largely relied upon composite or grab 21 samples and reverse transcription quantitative PCR (RT-qPCR) techniques, which limits the 22 accessibility and scalability of wastewater monitoring. In this study, we piloted a workflow that 23 uses tampons as passive swabs for collection and reverse transcription loop-mediated isothermal 24 amplification (RT-LAMP) to detect SARS-CoV-2 RNA in wastewater. Results for the developed 25 workflow were available same day, with a time to result following tampon swab collection of 26 approximately three hours. The RT-LAMP 95% limit of detection (76 gene copies reaction-1) was 27 greater than RT-droplet digital PCR (ddPCR; 3.3 gene copies reaction-1). Nonetheless, during a 28 building-level wastewater monitoring campaign conducted in the midst of weekly clinical testing 29 of all students, the workflow demonstrated a same-day positive predictive value (PPV) of 33% 30 and negative predictive value (NPV) of 80% for incident COVID-19 cases. The NPV is comparable 31 to that reported by wastewater monitoring using RT-qPCR. These observations suggest that even 32 Preprints (www.preprints.org) | NOT PEER-REVIEWED | Posted: 17 May 2021 doi:10.20944/preprints202105.0381.v1

accessibility and scalability of wastewater monitoring. In this study, we piloted a workflow that 23 uses tampons as passive swabs for collection and reverse transcription loop-mediated isothermal 24 amplification (RT-LAMP) to detect SARS-CoV-2 RNA in wastewater. Results for the developed 25 workflow were available same day, with a time to result following tampon swab collection of 26 approximately three hours. The RT-LAMP 95% limit of detection (76 gene copies reaction -1 ) was 27 greater than RT-droplet digital PCR (ddPCR; 3.3 gene copies reaction -1 ). Nonetheless, during a 28 building-level wastewater monitoring campaign conducted in the midst of weekly clinical testing 29 of all students, the workflow demonstrated a same-day positive predictive value (PPV) of 33% 30 and negative predictive value (NPV) of 80% for incident COVID-19 cases. The NPV is comparable 31 to that reported by wastewater monitoring using RT-qPCR. These observations suggest that even 32 with lower analytical sensitivity the tampon swab and RT-LAMP workflow offers a cost-effective 33 and rapid approach that could be leveraged for scalable same-day building-level wastewater 34 monitoring for COVID-19. samples and RT-qPCR to detect one infection among 60 skilled nursing facility residents 73 (Spurbeck et al., 2021). Another building-level study at a skilled nursing facility using wastewater 74 found to be similar among asymptomatic, pre-symptomatic, and symptomatic patients (Lavezzo 101 et al., 2020;Walsh et al., 2020) and asymptomatic and mild COVID-19 cases have been observed 102 to shed SARS-CoV-2 RNA in stool (Park et al., 2020), wastewater monitoring offers an opportunity 103 to screen for COVID-19 cases among building-level populations and identify cases via follow-up 104 clinical testing (Oran and Topol, 2020). 105 106 While wastewater surveillance offers a compelling tool for building-level COVID-19 detection at 107 universities, most of the reported monitoring efforts have depended on composite samplers to 108 achieve representative samples over a defined time period (usually 24 hours). These samplers 109 can be expensive and difficult to place in building service lines. Other studies have used grab 110 samples, but such samples are "snapshots" and may not afford a reliably representative sample. 111 A few SARS-CoV-2 wastewater monitoring efforts to date, however, have used the Moore Swab, 112 a gauze bundle left suspended in sewers to sorb wastewater and enteric pathogens. This type of 113 passive sampling was first used to detect Salmonella Paratyphi in 1948 (Barrett et al., 1980) and 114 has also been used to detect Vibrio cholerae (Barrett et al., 1980) and enteric viruses (Tian et al.,115 2017) in wastewater. More recently, Moore swabs in combination with RT-qPCR were used to 116 monitor wastewater at a university and were able to detect one to two COVID-19 cases in a 117 building (Liu et al., 2020). The same study found that when used alongside grab samples, the 118 Moore Swab allowed a greater sensitivity for SARS-CoV-2 RNA in wastewater from a hospital 119 treating COVID-19 patients (Liu et al., 2020). Another evaluation of passive samplers (gauze, 120 electronegative filter, and cotton buds) alongside traditional sampling techniques (flow-weighted 121 and time-average composite, and grab samples) found that passive samplers were at least as 122 sensitive over 24-hour deployments and a positive correlation between SARS-CoV-2 RNA 123 concentrations in wastewater and those from passive samplers (Schang et al., 2020). 124 125 Passive samplers, such as the Moore Swab, could make wastewater monitoring possible without 126 the use of expensive composite samplers. However, detection and quantification of SARS-CoV-127 2 RNA in wastewater samples has also required the use of RT-qPCR techniques, which depend 128 on specialized PCR equipment such as thermal cyclers. Reverse transcription loop-mediated 129 isothermal amplification (RT-LAMP) (Notomi, 2000) offers the potential to detect SARS-CoV-2 130 RNA in wastewater samples without the use of such equipment. RT-LAMP has been validated for 131 rapid testing of clinical samples including serum, urine, saliva, oropharyngeal swabs, and 132 nasopharyngeal swabs for SARS-CoV-2 RNA (Ganguli et al., 2020;Schermer et al., 2020). A 133 colorimetric RT-LAMP kit developed by New England Biolabs using multiplexed primers targeting 134 the N and E regions (N2 and E1) of the SARS-CoV-2 genome had accuracy greater than 90% 135 compared to RT-qPCR and a 95% limit of detection of 59 copies per reaction when used to test 136 heat treated saliva samples (Lalli et al., 2021). Multiplexing primers and the addition of guanidine 137 chloride was found to increase the sensitivity five-to tenfold for colorimetric LAMP with the N2 138 and E1 primers yielding the best performance among seven primer sets . A 139 preprint reported the use of RT-qLAMP with primers targeting the ORF1a, E, and N genes to test 140 wastewater samples for SARS-CoV-2 RNA without extraction in wastewater volumes up to 9.5 141 µL (Ongerth and Danielson, 2020). 142

143
During the current study, we piloted the application of colorimetric RT-LAMP to detect SARS-144 CoV-2 RNA in wastewater from tampon swabs and primary influent from WWTPs in northern 145 Indiana and northeast Georgia using a variety of extraction and processing techniques. We 146 assessed the sensitivity, specificity, and limit of detection of RT-LAMP for wastewater samples 147 compared to RT-qPCR and reverse transcription droplet digital PCR (RT-ddPCR). We then used 148 tampon swabs and RT-LAMP for rapid monitoring of building-level wastewater at the University 149 of Notre Dame over six weeks in conjunction with ongoing public health measures to assess the 150 positive and negative predictive value of these measures. Indiana. All such samples collected at WWTPs are referred to as "primary influent" throughout. In 158 addition to primary influent, a number of wastewater samples were collected from the wastewater 159 systems at the University of Notre Dame (ND) and neighborhoods within the Athens area, 160 including the University of Georgia (UGA). All samples from wastewater collection systems are 161 referred to as "raw sewage" throughout. Raw sewage samples were collected using two 162 techniques: 24-hour time-based composite samples (for the main sewage discharge manhole at 163 ND) and tampon swab passive samplers (detailed further below). In all cases, immediately after 164 collection, both primary influent and raw sewage samples were stored and transported on ice or 165 at 4°C until processed. 166 167

Tampon Swab Samplers 168
Tampons were used as low-cost and readily available Moore swabs for passive sampling of raw 169 sewage in the wastewater collection system. At UGA, 100% organic cotton tampon swabs (OB 170 Brand Organic Tampons, Super) were deployed into the wastewater collection system for 24 171 hours at each sampling location. After recovery, swabs were placed in sterile WhirlPak bags 172 (Nasco, Fort Atkinson, WI) and saturated with 20 mL of sterile PBS. Saturated swabs were hand 173 massaged for two minutes to elute viruses and then the sorbate was squeezed from the swab 174 and collected in a sterile 50 mL centrifuge tube for immediate extraction. 175 176 At ND, with the assistance of utilities personnel, tampon swabs (Tampax Pearl, Super) were 177 deployed into the wastewater collection system weekly for six weeks from approximately 8:00 am 178 to 11:00 am at nine different locations selected to isolate individual residential halls (RH) 179 (anonymized as RH 1 to 9). During the monitoring period, these RHs housed 1,627 students 180 accounting for 25% of the on-campus residents. Upon retrieval from manholes, swabs were 181 placed into sterile WhirlPak bags and stored on ice. In the lab, swabs were hand squeezed while 182 in the WhirlPak bag to remove most of the sorbate and then aseptically placed into a 60 mL luer-183 lock syringe (ML60, Air-Tite Products Co, Virginia Beach, VA). The sorbate remaining in the 184 WhirlPak bag was then poured into the syringe and pressed into a 50 mL centrifuge tube using 185 the syringe plunger typically resulting in 25 to 35 mL of sorbate. After the first press, a volume of 186 PBS/Tween20 solution (10 mM sodium phosphate, 0.15M NaCl, 0.05% Tween 20) was pipetted 187 into the syringe (typically 15 to 25 mL) such that the total volume of absorbate resulting from each 188 swab was 50 mL and pressed through the swab into the centrifuge tube. The resulting 50 mL of 189 sorbate was then immediately concentrated or extracted as described below. The membrane was then aseptically rolled into a 2 mL Garnet bead tube (Qiagen, Hilden, 201 Germany) and frozen at -80°C until homogenization prior to extraction.

COVID-19 Clinical Surveillance at ND 307
During the period of wastewater monitoring at ND, COVID-19 safety protocols were in place 308 including universal masking, physical distancing, daily health checks, and asymptomatic and 309 symptomatic COVID-19 testing. COVID-19 testing methods included saliva-based PCR tests, 310 primarily for asymptomatic surveillance, nasal swab PCR tests, and rapid antigen tests. All 311 undergraduate and professional students participated in mandatory weekly surveillance testing. significance determined by likelihood ratio test (Fox, 1997) and fit assessed using Tjur's R-341 squared (Tjur, 2009). Comparisons between two groups (e.g. inhibition between sample types) 342 were made using Mann-Whitney tests and between multiple groups (e.g. inhibition between 343 extraction methods and positivity rate between sorbate fractions) using Kruskal-Wallis tests with 344 Dunn's post test (Dunn, 1964

No Extraction Inhibition Rate 400
We attempted extraction-free RT-LAMP on five tampon swab sorbate and four 24-hour composite 401 samples of WWTP influent. The inhibition rate among the five undiluted passive samples was 402 100%. The inhibition rate for undiluted composite samples was 100% when using 7 µL or 4 µL of 403 input. After 1:10 dilution, no inhibition was observed for 7 µL of input. Given the dilution required 404 to remedy inhibition and the resulting 10x increase in the 95% LOD, we abandoned extraction-405 free RT-LAMP as a reliable detection method. 406 407

Heat Extraction Inhibition Rate 408
After heat extraction, 100% of swab sorbate samples (n=5) were inhibited and remained so even 409 after 1:10 dilution. Among the five solid fraction samples, 100% were inhibited after heat 410 extraction, and 40% remained so even after 1:10 dilution. Given the high rate of inhibition, we 411 abandoned heat extraction as a reliable method for detection in wastewater via RT-LAMP. during the six days following wastewater testing ( Figure 2A). As shown in Figure 2B, NPV was 487 greater with a maximum of 78% on the day of wastewater testing to a day six minimum of 38%. 488 The PPV of wastewater testing could be adversely affected by positive RT-LAMP results 489 attributable to convalescent COVID-19 cases returning to residence halls after isolation. As shown 490 in Figure S9, there were six instances where RT-LAMP replicates were positive despite no 491 incident COVID-19 cases, but with returning convalescent cases in the prior seven days. In these 492 six instances, it required four or more convalescent cases before 2 of 3 RT-LAMP replicates were 493 positive, suggesting that a cutoff value of 67% positivity (2 of 3 replicates) could increase the PPV 494 of the wastewater method. As shown in Figure 2A, PPV is increased to 33% when 2 of 3 positive 495 RT-LAMP reactions are required to classify a sample as positive. This change in cutoff value 496 leaves the NPV largely unchanged ( Figure 2B). If the detection of convalescent COVID-19 cases 497 by wastewater sampling is considered a true positive (e.g., the true detection of SARS-CoV-2 498 RNA shed into the wastewater system), then the PPV improves to 56% on day 0 up to 75% by 499 day three after wastewater monitoring ( Figure S11). 500 501 4. Discussion 502

Reliable RT-LAMP Workflow and Analytical Performance 503
To develop more accessible wastewater monitoring techniques, we piloted and characterized the 504 performance of a monitoring protocol that makes use of tampon swabs and RT-LAMP to detect 505 SARS-CoV-2 RNA in building-level wastewater. The 95% LOD for a single RT-LAMP reaction 506 was 20 times higher than the RT-ddPCR N1 assay 95% LOD. Several studies have found that 507 SARS-CoV-2 RNA shedding in feces can outlast nasopharyngeal shedding in up to 50% of 508 COVID-19 patients (Elbeblaw, 2020;Jones et al., 2020;Wang et al., 2020). In such cases, the 509 higher RT-LAMP LOD could be advantageous by allowing for convalescent cases to go 510 undetected, while newly incident COVID-19 cases could still be detected. RT-LAMP 511 demonstrated an overall sensitivity of 57% compared to PCR methods, and a specificity of 100% 512 compared to one-step RT-ddPCR. Unfortunately, we were not able to replicate the findings of an 513 earlier pre-print study as all of our attempts to test wastewater directly were inhibited (Ongerth 514 and Danielson, 2020). Our attempts at heat extraction were also consistently inhibited despite the 515 success with saliva and other clinical samples (Mahmoud et al., 2021). We found that regardless 516 of the wastewater type (influent composite or swab sorbate) the use of an extraction kit for testing 517 by RT-LAMP was important to produce uninhibited results. The optimized tampon swab and RT-LAMP workflow yielded a same-day PPV of 33% and an 529 NPV of 80% in six weeks of wastewater monitoring. Accounting for the detection of convalescent 530 cases improves the PPV to 56%. The PPV we observed was much lower than the 82% reported 531 during another study leveraging PEG precipitation and RT-qPCR, but the NPV we observed (80% 532 versus 88.9%) was comparable (Betancourt et al., 2021). The specificity of the tampon swab and 533 RT-LAMP method for COVID-19 cases was 80%, which is better than the 52% specificity reported 534 for an ultracentrifugation and RT-qPCR method that did not distinguish new infections from 535 convalescent (Colosi et al., 2021). Thus, the tampon swab and RT-LAMP approach may offer a 536 specificity and NPV comparable to more sophisticated monitoring methods. Several 537 epidemiological modeling studies have suggested that an optimal strategy for managing COVID-538 19 on college campuses should include high-frequency screening tests that are highly specific 539 (Lopman et al., 2021;Paltiel et al., 2020). Our observations indicated that the NPV and PPV for 540 tampon swab and RT-LAMP monitoring were maximized with wastewater monitoring daily to 541 every three days. 542 we could not avoid using a kit-based RNA extraction, the method does not require a composite 576 sampler or thermal cycler for RT-qPCR, relying instead on tampons for sampling and basic lab 577 equipment including centrifuges, microcentrifuges, vortexes, and single temperature incubators 578 for swab processing and RT-LAMP testing. The per sample analytical cost was comparable 579 between RT-ddPCR ($35) and the NEB RT-LAMP kit ($31); however, we estimate that a self-580 assembled RT-LAMP kit using the same primers could halve the per-sample cost once optimized. 581 Even with the off-the-shelf kit, the per capita consumables cost for the entire workflow was 582 approximately $0.25. 583 584

Limitations 585
There are limitations that should be considered in generalizing the findings of this study. First, our 586 comparison of RT-LAMP and RT-qPCR/ddPCR leveraged samples from only two monitoring 587 sites, ND and UGA. Although we made use of raw sewage and WWTP influent samples, 588 wastewater, and therefore RT-LAMP performance, can be variable among sites. For comparison 589 with clinical surveillance, we monitored wastewater at nine ND residence halls. We note that while 590 COVID protocols during the sampling period did not allow guests into the residence halls, it is not 591 possible to completely exclude the possible shedding of SARS-CoV-2 RNA into the residence 592 hall wastewater by non-residents. The predictive performance was variable between halls and 593 weeks and the study was not designed to further investigate these differences. The tampon swabs 594 were only deployed for a three-hour interval between 8:00 am and 11:00 am. This period 595 accounted for roughly 20% of daily domestic water use, but the performance of the workflow could 596 potentially be improved with longer deployments of the tampon swabs, assuming this does not 597 lead to increased rates of inhibition. We independently monitored the wastewater from residence 598 halls during a large and robust clinical surveillance program that featured weekly testing of every 599 single student. In the midst of such a clinical surveillance effort, the predictive performance of 600 wastewater monitoring is likely to be conservative. Nonetheless, our experience suggests that 601 tampon swabs in combination with RT-LAMP could afford a specific, rapid, cost-effective, and 602 accessible screening method for building-level wastewater monitoring. As vaccination efforts 603 continue to progress, such a monitoring method may offer a scalable approach for non-intrusive 604 screening of at-risk populations. 605 606

Conclusions 607
• RT-LAMP sensitivity was 57%, specificity was 100%, 95% LOD was 76 gene copies per 608 reaction compared to SARS-CoV-2 RNA detection by RT-ddPCR. 609 • Tampon swabs combined with RT-LAMP were successfully used to detect SARS-CoV-2 610 RNA in building-level wastewater with results available the by 3 pm the same day. 611 • Over six weeks of monitoring the swab and RT-LAMP wastewater test demonstrated 612 80% negative predictive value and 33% positive predictive value compared to clinical 613 COVID-19 testing. 614 • The consumables cost of wastewater monitoring over six by tampon swab and RT-615 LAMP was less than $2 per person and could likely be further reduced through a self-616 assembled LAMP kit.