The Predictive Power of Poop: COVID-19 Wastewater Survelliance to Inform Public Health Decisions

good morning or good afternoon everyone and welcome to genome webinars i'm julia carro managing editor at genomewipe and i'll be your moderator today the title of today's webinar is the predictive power of poop covert 19 wastewater surveillance to inform public health decisions in colorado this webinar is sponsored by bio-rad laboratories our speakers today are dr susan delong associate professor in the department of civil and environmental engineering at colorado state university dr carol willis professor of microbiology immunology and pathology also at colorado state university and dr heather pitkok chief medical research officer also at colorado state university you may type in a question at any time during the webinar and you can do this through the q a panel which appears on the right side of the webinar presentation we will ask all our speakers you have questions after the presentations have concluded also if you look at the bottom tray of your window you will see a series of widgets that can enhance your webinar experience so with that let me turn it over to susan please go ahead with the presentation thank you so much julia so i'm going to start us off by framing the motivation for wastewater epidemiology so wastewater epidemiology is actually an established field this idea that we can monitor viruses in wastewater to get a sense of the prevalence of a disease in a population and this has been done previously mostly with polio early on in the pandemic work was done in europe to look at sars cov2 in wastewater and this is a published graph from a study in france which shows that the levels of sars cov2 in wastewater track clinical data in this case fetal cases in france as well as paris you can also see that in red the untreated sewage samples the line uh that curve there for the levels of sarsko v2 and wastewater um that actually the data the the detection of sars to kobe 2 and wastewater precedes the clinical cases so it was demonstrated early on in the pandemic that monitoring wastewater could be an early indicator giving us two to ten days of advance notice that an outbreak is about to occur in the community which is really the key motivation for using this tool and the reason the word predictive is in the title of this talk so the reason that this works is that people who are infected with sars qov2 will shed the virus in their feces very early on in infection you can see in this graph the log rna copy is of the genetic material of the stars virus as a function of day since onset of symptoms and early on when symptoms occur the fecal shedding is very high and it does occur prior to symptom onset and importantly both pre-symptomatic so before symptoms set in as well as asymptomatic people who never have symptoms of this disease will shed the virus so we can get an indication of the levels the general levels of infection in a community without people needing to feel sick go to the doctor get a test wait for that test result to come back so that's another reason that this is an early indicator and a helpful parameter to monitor for public health decision making so in this talk today we're going to talk about two use cases for monitoring sars cov2 or the virus causing covet 19 in wastewater large population monitoring and small population monitoring large population monitoring involves taking samples at wastewater treatment plants at the influence of wastewater treatment plants and this allows us to monitor in entire communities with a relatively small number of samples so here you can see that we might monitor between twenty thousand and two million people with wastewater treatment plant samples we're doing that in the state of colorado at 20 locations twice a week and the samples are sent to our testing laboratory at csu and the data is used to plot trends in viral copy number and help epidemiologists at the state make decisions about social distancing status the second use case is small population monitoring and in this case we might monitor a single building or a facility we're doing this on the colorado state university campus monitoring dormitories and i'll talk about our study design there this is a much larger number of samples because of course we're monitoring smaller populations of individuals and the use case here is a little bit different we can use this information to direct follow-up testing to try to identify the infected individuals so i'm going to dive in now to how we actually go about collecting the samples and how we developed our sampling plan so first i'm going to talk about the monitoring that we're doing at colorado state university for building level monitoring so early on we decided to monitor all of our dormitories and residences student residences on campus so in in thinking about this building level monitoring strategy we wanted to think about which locations would be the most strategic and beneficial with respect to controlling outbreaks on campus and we started with a preference for residences over administrative or other types of buildings on campus and this is really due to the timing of people's contribution of fecal matter to the sewer we we sort of know that that's a generally a morning activity and because as i said early on in the talk this virus is mostly shed in feces and wastewater we really want to capture that signal to get a sense of the prevalence of infection in our community so we targeted residential buildings rather than buildings where people work on campus this also allows us to direct follow-up testing because we can link individuals with the buildings they live in and message those individuals to participate in follow-up testing that will be discussed later in this talk so once we decided to monitor dormitories what we needed to do was an in-depth study of the sewer system at csu and this was done in collaboration with engineers that are part of our facilities team at csu you can see um sort of a grayed out image of a building there and in very faint green lines there are um there's a map of the sewer system and so the goal is to find manholes that can be linked with an individual building or a small group of buildings so that that facilitates selection of smaller groups of people for follow-up testing if we detect the virus in the wastewater there's also a lot of complexities to to this decision-making process we wanted to avoid mixed-use buildings to the extent possible so some of some of our buildings have dining commons that also complicates the contribution to the to the sewer we wanted to limit sewer contributions from off-campus buildings so that we knew we were monitoring mostly our our students and we wanted to avoid contributions from unidentifiable contributors so if a manhole also receives water from the gym or the library it will complicate determining how to direct follow-up testing and then there's also other practical considerations that i highlight here so you can see a picture of one of my team members in tyvek suit and a fluorescent vest there and there's a nice manhole on a grassy strip safe to access for the team that's an ideal location some of the ideal manholes that we identified were actually in the street um and and this is just an example from from not a location on campus but we don't want to be sending our our student samplers into the middle of the street this is just a quick picture of the wastewater sampling this is jason graham taking a sample and that's how it's done at the wastewater treatment plant so it's much more straightforward a little bit more about the building level samplers i wanted to provide a few tips and perspective here we use composite sampling pumps these pumps collect samples over a 24 hour period so rather than a grab sample where you just go and take a sample at a certain time point we wanted to collect samples over 24 hour period to make sure our samples were representative we use compositing pumps that are hung in the sewer this keeps them out of sight safe and actually prevents them from freezing in the winters because the temperature in the sewers is more moderate you can see there a picture of one of the strainers in the middle photo that hangs in and collects the waste water our our samplers are programmed to collect water every 15 minutes over that 24 hour period here's just a few more pictures of the team in the field learning how to program the samplers and showing how the samplers are hung in the sewer a few more sample examples of field work i just like to highlight here the importance of safety safety training that our team is is trained in ergonomics to make sure that you know there are no injuries associated with lifting and lowering these heavy samplers slips trips and falls biosafety hazards there are a lot of considerations in deploying a team to do this kind of sampling to ensure that it can be done safely so once we've collected the sample from a community or a building and we are taking a 24-hour composite sample those samples go over to the lab where we have a protocol to remove solids concentrate the virus extract viral rna and do the quantification which our next speaker will discuss in detail i wanted to talk a little bit about some of the sample collection challenges as they relate to the the downstream analysis so some things to make are everyone aware of flow rates vary in sewers over time and the fluorides can be very low the composite samplers that we purchased are off-the-shelf commercial samplers and they're not really designed for building level sampling and so we had to take into some considerations low flow and i'll show you some modifications in a moment it's also worth noting the complexities associated with sewer sheds as well as buildings we're really interested in black water water that comes from the toilet and sewer sheds have all different contributions kitchen waste at dorms and dining commons laundry strong cleaning agents can cause issues and then if we're looking at a sewer shed we might have contributions from industrial wastewater um and then another challenge is by variation in the viral extraction efficiency so i want to talk about some strategies that we use to address these challenges and another one that's not uh noted on the slide here is for building level samples there's a lack of homogenization so the samples are not super well mixed um so we need ways to account for that so we developed some strategies within the sampling framework to innovate on strainer design and then we have controls that can be used to measure concentrations of fecal matter which helps address the issue of various inputs beyond just black water from residences as well as the lack of homogenization and then we have a control that helps us to um track the percent recovery for every sample that we analyze so just a quick uh some quick photos of strainer designs uh we had to develop new strainers what that had lower diameter so a one-quarter inch strainer will work much better for building level sampling because the flow is so low the depth of the water is low in the middle you can see a strainer that's designed to limit clogging we've definitely had challenges with issues of various things that end up in sewers clogging our strainers and then on the right is a low flow strainer which only has holes on the bottom and that little wing plate helps it to not roll in the sewer and this can work in really low flow situations um here's a slide that presents some of the controls that we're using we are we've used two different endogenous viruses as a control for the level of fecal matter which can be used to normalize f plus bacteriophage is associated with gut microbiome and so it's an endogenous virus associated with fecal matter this has worked really well for state samples we've also used the pepper mild model virus which which is another indicator of fecal matter those we can track those two viruses to help us understand the level of fecal material in a particular sample as opposed to other contributions wastewater contributions like laundry water and then for our percent recovery spike we really wanted to use a coronavirus we wanted a virus that had the same surface chemistry and would behave in a similar way in our extraction protocol to sars cov2 and for that we were able to use this bovine coronavirus vaccine and we spiked that into all of our samples prior to extraction of the virus the the stars cov2 virus for analysis and so these are some of the strategies we've used to address the complexities that originate on in the sewer system and on the sampling side and with that i'm going to turn it over to dr carroll willis to talk about the analysis part thank you susan so when the lab receives the wastewater samples there are a number of things that we need to think about um when we were developing the assay in order to measure the levels of the coronavirus in it there are some good things about working with sewage believe it or not um one of the main advantages is there's plenty of it and that made it quite easy for us during the summer to access large quantities to develop our essay and the city of fort collins provided us with a lot of a lot of sewage samples over the summer another advantage is that it's non-invasive and it requires no irb approval to take these samples because we can't track anything back to the individual that contributed as susan mentioned it gives us a nice snapshot of the population snap snapshot of the population and of course you can't underestimate the power of poop jokes especially for those working in the lab and the cons of of working with waste water are there though and we we do take those very seriously for example um coronavirus is not the only virus that's present in wastewater there are a lot of very nasty viruses hepatitis a rotavirus norovirus and these viruses are specifically designed to transmit through feces and so the the viruses that are present are actually still infectious so we have to handle our samples pretty carefully another disadvantage is that the samples are pretty dilute there's a lot of other a lot of water going through the sewers obviously and so that we need 40 mils of sample to get enough virus to detect so we have to concentrate that because we can't we can't deal with a 40 ml sample and just put it into a pcr reaction and wastewater of course has a lot of yucky stuff in it and these these chemicals and molecules can inhibit our reactions that are designed to detect the virus and so we need to account for that and as susan mentioned that samples aren't homogeneous which means that we get we do get a lot of variation from one sample to the next um and from one day to the next and between different between samples from different locations and all of that means that although our experiments have run very carefully there is a lot of noise in in this type of data so susan mentioned the steps that we go through basically we spike in a recovery virus we separate out the particulate matter we concentrate the virus we extract rna and then we quantify using a pcr approach there were a number of things that went into consideration when we were developing our approach for quantifying the virus sorry um we we had a list of qualities that we wanted our assay to have the first one here being fast was really important because we're trying to predict what's going to happen in the future we can't wait two weeks to turn around our results we needed a 24-hour turnaround from receipt of the samples in the lab to when we're producing the data we wanted the assay to be pretty easy um this is because we did not want to have to go find highly trained technicians in the middle of a pandemic we we needed to be able to hire somebody that was basically competent in molecular biology needed to be able to run these assays um as i mentioned it's important that the assay was resistant to inhibition and we wanted it to be sensitive so that we could detect the virus as soon as it appeared and accurate so that we could quantify it unlike with a clinical test where you just need to know whether the person is positive or negative we actually need an accurate quantification of the number of copies of the virus in each sample ideally we wanted to be cheap to set up and run the sample run these assays um and it turned out to be really important that the the supply chain for all the reagents and supplies we needed to run the essay needed to be robust but um and this has been a real problem with a lot of the clinical diagnostic testing and the the supplies for real-time pcr have sometimes been hard to come by and we we've managed to avoid most of those issues and finally we want to be able to rely on our existing equipment and expertise rather than having to buy a lot of new equipment and find additional people to perform the experiments so for the clarification and concentration steps the clarification getting rid of the solids um we we've used filtration and centrifugation these are both um fast and and easy and cheap um both work equally well so these have been our go-to methods for just getting rid of the lumps that are in the sample and then for the concentration we decided to go with an innova in overprep concentrating pipette which was recognized recommended to us by a wastewater treatment facility um who were using this this approach already um this pipette is really really neat we're basically we stick this probe into our sample it sucks up the 40 ml of liquid and then we elute it off of the filter into 400 microliters and this takes two minutes um so we basically take 40 ml of wastewater and we turn it into 400 microliters and we'll also get rid of some of the other like smaller real really small particles in a two-minute period and so this is the the reasons we picked this is that not only does it work really well it's fast and it's really easy to use this instrument is is it tells you how to use it every step the rna extraction we we tried um trisol extraction and the kyogen prep in the end we went with the kaijen um uh like uh kit primarily because it was easy it was it was there were there were less problems with um the the dna not being able to be amplified or rna not being able to amp be amplified afterwards but both of these approaches worked pretty well um trisol has the advantage of being cheaper and the supply chain is more robust but we haven't had any trouble with the supply chain for the other reagents either so for the quantification step um this this is probably the the most important step for us this is what gives us the data um we were very fortunate in that we already had on-site um a bio-rad qx 200 digital droplet pcr system um this had been purchased in 2015 it had been in use on campus in my molecular quantification core for five years my lab was very experienced in using this they've been they've been using it for measuring gene expression for in various samples for a while um so that that was one um one real strength of of our approach was we already had this this uh equipment available um but we also focused on this because we we do have real time machines available as well but we decided to go with digital pcr also because there's less demand for the reagents and the the supply chain has been robust we've had no problems getting the supplies we need to run the assays um digital droplet pcr is particularly resistant to inhibition which is a big advantage that has over real-time pcr it doesn't really get affected by all of the yucky stuff that comes through that might come through with the rna from the wastewater and then another real advantage is that it gives absolute quantification when we put the rna into the sample the data that we get out at the end tells us the exact copy number of virus in that in that specific sample without any need for a standard curve and that's because we're we're actually reporting copy numbers and not just positive and negative this was actually really helpful digital pcr is also more sensitive than real-time pcr we can so we were we're able to detect the virus at lower levels than you would do with real-time pcr and it's also very accurate and finally um there are reagents available that allowed allowed us to perform the reverse transcription turning the rna into dna in the same step as the pcr reaction so that saved us some pipetting um and some time and it allowed us to turn around around our data more quickly so this is an overview of the uh process that we go through when we're when we're after we've isolated the rna from the wastewater the reactions are just regular pcr reactions with a fluorescent um probe they get set up in individual tubes so each each sample has two reactions set up for it one for the virus the coronaviruses and one for the control viruses and then each of those reactions gets partitioned into 20 000 droplets and so droplets sometimes get a a copy of the rna and sometimes they don't get a copy of the rna and this is just done by mixing it with oil in a droplet generator and so then that these individual droplets go through thermocycling cycles of heating and cooling that allows the dna to be amplified and at the end of the reaction any any some any droplet that received a copy of the viral rna will have a fluorescence um will have a fluorescent signal whereas droplets that didn't receive a copy of the viral rna will have no fluorescence and we put the reaction through a droplet reader which is essentially like a flow cytometer for those of you who know about flow cytometry um and that looks at each droplet and measures the amount of fluorescence in two different channels a red channel a green channel and a blue channel and from that amount of fluorescence we can we can count the number of positive droplets and using a mathematical equation connected to the poisson distribution from that the software that would is provided with the machine can determine the exact copy number of the virus in in each sample and so this is what the the data looks like on the right you can see there's the gray drop the gray band at the bottom is uh the negative droplets and then the blue and the green are the positive droplets in two different um two different channels so the green channel is for in in this particular reaction is the bovine coronavirus and the blue channel is is showing the sars coronavirus probes and so you can see that the sars coronavirus probes are labeled with a different fluorescent molecule and so they fluoresce in a different channel and we can detect them separately even though they're in the different in the same tube one thing to notice is that the amount of the positive blue droplets varies across the different samples so that each each uh column there is a different wastewater treatment plant or manhole and you can see the blue sun the blue droplet number will vary and that's because the amount of coronavirus varies from different look in the in different locations whereas at the bottom the bovine corona virus you can see that the the number of green droplets is pretty similar across the board and although one or two of them might have lower numbers of droplets this is because we put the same amount of bovine coronavirus into each reaction we know how much is there um and so we we should get similar concentrations of bovine coronavirus across the board although sometimes if our recovery is bad for example if there are chemicals in the wastewater that affect the survival of the virus then we can detect that using the bovine coronavirus control so we perform a reaction um that contains the coronaviruses and a second reaction um to detect the um in endogenous controls of cole phage and the peppermild mosaic virus okay so um susan already talked a little bit about the control so i won't go into them in a huge amount of detail but what i wanted to point out was on the right here i'm showing the copies per liter of the f-plus and the peppermild mosaic virus and you can see that the f-plus for example in the ug lane um in the ug sample can be pretty low and sometimes we don't detect it at all and we think that that's because in in that particular location there are chemicals or something going into the into the water that are affecting the survival of that virus and our ability to detect it um in this case the peppermill mottle virus is the signal from that is more consistent and the other thing to notice is that especially for the peppermint model virus the the signal is pretty consistent across the board with different for different sites um whereas the f seems to be a little bit more variable but we found that f plus is is great for um our state samples and and the peppermint model virus seems to be working a little better for the dorm samples and as i we also do controls for contamination so a no template control where we don't put any sample in um we should never get any signals so that's that's just to make sure that all of our reactions are clean and we're not mixing between different reactions and then we do positive controls within activated virus just to make sure that all of our reactions are working so this is just an overview of the whole process so we take our wastewater that comes into the lab we add spike of the bovine coronavirus vaccine we we filter or centrifuge we ultrafiltrate to concentrate the virus we extract rna we we assay in two separate reactions to detect four different viruses using real digital pcr and then at the end we get out these droplet numbers um that will give us an indication of the levels of virus in the sample our limit of detection is about three thousand five thousand copies per liter and the limit of quantification is about thirteen thousand to fifteen thousand copies per liter um and these all of these criteria get reported on a dashboard either to the csu rapid response team which heather will be telling you about in a minute um or to the colorado department of public health and the various wastewater treatment facilities and that this data gets uploaded within 24 hours of us receiving the sample almost all of this except for the data analysis is done in one day um pretty much the the people in the lab will put the the drop the pcr reactions on to run before they go home at about nine to ten nine or ten o'clock at night and then they'll either come back and read them um later or i will go in and read them in the morning and then we don't we analyze i analyze data or um a student graduate student will analyze the data first thing in the morning and so we we are able to get through this whole process within 24 hours of reading of receiving the samples and report those results back within 24 hours so as fast or faster than a clinical essay which only tests one person whereas we're testing hundreds to thousands of people i'm going to tell you a little bit more about what we do with the data in a minute but before i go on onto that i wanted to talk first about variant detection um so probably you're aware of that that there are a large number of variants of a number of variants of concern arising around the globe and these have different properties than than the original coronavirus um some of the some of these properties are particularly concerning the mutation in the spike gene n501y um is seen in the uk south african and brazilian variants and it's been shown to help the virus spread more easily um and then on the other hand the e484k mutation is seen in in south african brazilian and in some uk variants and that may affect the the ability of the body body to um recognize that virus after it's after it's been vaccinated um so the antibody response to the wild type virus may not recognize the e484 virus and so these these variants are concerning because we cannot effective they have different properties and so all of the mathematical models and our responses are based on the wild type virus and these new viruses have different properties and so we need to adjust our models as these as these uh variants appear um and we also it's we possibly will need to be more aggressive with our quarantine and other responses as if these become widespread and so we want to be able to detect these in wastewater and we've been able to show that this is possible um this this on the right here the end top right the n1 is the regular um primer probe set that we use to detect the corona virus and then on the on the left um n501y is detected by an assay developed by bio-red that has two different color probes one the blue fluorescent probe detects the mutant virus so the n501y variant and the green probe detects the wild type virus virus and you can see um in the data on the on the bottom um it's not all that clearly labeled but you can you can see that the um blue virus is only present in um one of those five samples on the left um and where whereas the wild type is present in all of the five samples on the left and the samples on the right are just controls to make sure that we're we're detecting what we say we're detecting so that fifth sample the the fifth sample from the left which is right in the middle um we sent that for sequencing and we were able to demonstrate that the presence of multiple variants consistent with it being the uk variant was straight were present in that um particular sample and so um based on this evidence we're pretty confident that the uk variant was present in that sample and we can move on to to plan accordingly for seeing probably more of that variant as time per se proceeds okay so what do we do with the data once we have it and as susan mentioned at the beginning we can we're monitoring large populations and multiple wastewater treatment plants across the state of colorado and small populations um as in the dorms um at csu um which heather's going to tell you about more soon and so the the with the large populations and what we're looking at is trends in viral copy number um and the data that we're providing is informing public health decisions about social distancing um and health systems and and the economy um along with issues being inc used in conjunction with all the all of the information about clinical case numbers um so we started treating we started receiving wastewater samples at the beginning of august and we've been testing twice a week since then um so uh multiple um wastewater treatment plants across the state of colorado are sending us our they're sending us their samples um twice a week um and this is what the data that we we provide looks like so in the black the black line here is our data um over time and then the blue line is the clinical cases and you can see the clinical case load tracks attracts nicely with the the number of uh copies of virus that we're detecting in the wastewater and this is just two different locations but importantly the the actual correlation between um between the amounts of virus in the wastewater and the clinical cases is really is really quite um compelling and so an r squared value closer to one um a perfect r squared value as a perfect um correlation is indicated by this number being one um and so you can see for this this particular wastewater treatment plant sample we're getting a really good correlation between the clinical cases in in orange and then and the signal that we're receiving in the wastewater in blue and even for um those wastewater treatment plants with less um correlation the correlation is still very good and has predictive value and so that this one's the number is not quite so good but it's still definitely giving us a um really useful information um and so when we provide the the copy numbers of the virus um with it's being taken into account and informing the public health decisions down the road um on on the state level for municipalities um and for the local um areas and so i'm going to stop now because that's the bit that i've been most um connected with and i'm going to hand off to heather who's going to tell you about the response that we have at the university level to the data that we're providing thanks so much uh carol um so i'm gonna shift the focus to the uh the small population monitoring and talk a little bit about why we chose to do this and how we analyzed the data and responded to it and what the impact has been so the the why of doing it is early identification of asymptomatic people so the the hope was that by identifying them and putting them into isolation early and then putting their close contacts into quarantine early that we could not only decrease the number of cases but also decrease transmission to people who might have more severe symptoms and be at higher risk there are challenges and these have been touched on before but in in terms of predicting how many infected individuals are contributing to a sample so some of the variability that dr delong spoke about in terms of homogenization of the sample there's also individual variability between the people who are infected because fecal shedding early infection has not been fully characterized and there is variability from one person to the next um the relationship between fecal shedding and respiratory infectivity in terms of knowing that timing is not fully known and fecal shedding can persist even after the quarantine or isolation period is over and that person has returned to their assigned housing and so early on in the semester it was fairly easy to say that if we saw a signal that was an active infection but as the semester went on we started having to make determinations about whether the signal we were seeing was due to people returning to housing and no longer being infected but still shedding fecal matter um or whether it represented new infections that needed to be identified and those people removed from general circulation so um there was again variation in flow rate chemical composition all the things that have been discussed um so we had some significant constraints as we um contemplated bringing this this uh technology online there was no road map to follow and there still is no road map um we didn't even have publications at that point about micro monitoring um approaches um all there were uh good publications about testing at the treatment facility level but we didn't know whether that could be applied to these smaller settings we knew what the limited detection was but we didn't know what the threshold for action should be and at what level we might be able to find cases or and because the timing wasn't clear we didn't know if we would see waste water signals that then could not be identified in the people just because they had not yet uh had a virus in their nasal passages and thus the nasal swabs wouldn't show anything um we didn't know how long the virus would stick around for and and what to do if the if we couldn't find the infection um so all in all we really didn't know if this was going to be a successful and productive thing to do but we didn't have a lot of other options early in the pandemics so in the fall of uh 2020 we just decided that this was something that we had to do and we had to make decisions based on little to no information so we decided to start with the lowest threshold possible which was that limit of detection between three to three thousand to five thousand and then calibrate it to our results and this um definitely was more art than science particularly later on when we started having people who are no longer infective contributing to the signal we looked for trends we found that resolving or increasing trends were much more useful than any individual measurement and we accepted that occasionally there were signals that resolved and we never identified a case and we didn't know if that was because the person was a visitor to one of the dormitories a worker who left um and didn't get tested or um or what the reason was and so trial and error was definitely an element in how we did this as as dr willis mentioned we formed a rapid response team in order to optimize outcomes and there was a very large degree of expertise that was represented by that group so we had clinical scientific technical and statistical and economic experts all contributing their their help to determining what did the data mean then we had legal and regulatory experts helping us understand what the boundaries were of what we could and couldn't do about it and then we just had a lot of of people on the operational side the student affairs size side greek life housing and dining communications helping us to understand how to effectively communicate strategies incentivize students faculty and staff to be compliant and help us keep the campus healthy and safe so um csu worked on developing multiple capabilities and um ddpcr was a was a a big tool for all um or i should say most of the efforts that we launched so the wastewater testing you've heard in detail we also developed a pooled saliva assay which i'll tell you more about which was very helpful in being able to expand the number of people that we could screen in a time efficient and cost efficient manner and then we were also outsourcing molecular testing to a clia certified lab that was using an fda authorized test in order to confirm the results so currently um the way that wastewater monitoring is occurring at csu is that dr willis sends us results letting us know that wastewater was was positive for a cyrus kovi to signal the residents of the facility where that signal is located are notified to go and get tested early so now everyone who has a campus president must be tested but they're asked to test early in the week and then again that's analyzed by digital droplet pcr first um in a in a pooled approach and then by our outsourced commercial partner who is doing the um authorized testing so for example if we found that there were six positives within a building after doing that screening we would then look again to that wastewater result trying to make sure that it's coming back down once those positives have been moved into quarantine or isolation and if it doesn't come down then we continue to notify those those residents and make sure that they do get tested and at the same time we look for an increasing result in the quarantine slash isolation facility to make sure that um we're seeing the movement of that signal so as i mentioned small population monitoring is is much more about overall trends than about any single result and what we what we've seen is it's very um reliable if we see multiple increasing signals or we see multiple disk decreasing signals where we don't always find a result that's commensurate with the size of the signal is when we have like a signal a single really large signal we've almost always find cases when that's the case but we don't know we can't correlate it to the number of people that we actually um find so um we have multiple different people that are contributing to the response to wastewater monitoring and um and all of them are very integral to this process um in terms of the saliva testing it is a pooled approach where uh pools of eight are um are pulled together and tested um it's done at a socially distanced facility um there's been an aerosol evaluation of that facility and because it's a gym with very high ceilings and a large volume of distribution it has been found to be similar to being physically distanced by six feet apart outdoors on average the process takes seven minutes so it's very very quick people don't have to stay around in that building for long periods of time and we only allow asymptomatic people to go get the saliva screening anybody who has symptoms has to go directly to get the fda authorized clia certified nasal swap what we find is there is a degree of correlation between the signals that we're finding in the actual cases that we're finding it's nowhere near that tight correlation that is possible when you're doing it with a large population at a treatment facility but the flip side of that is because it's a small population it's much easier to actually have an actionable item as a as an outcome of the result so um we found that we've had a very high degree of return on investment for this project and and for the overall project the wastewater monitoring has been very helpful in maintaining the health of the um residence facilities where people are living in very tight quarters sharing bathrooms sharing eating spaces etc it's really contributed to keeping the campus open to in-person classes through the goal date in the fall of november 20th which was a big achievement um and we are have ample evidence that asymptomatic cases have been identified and proactively put into isolation so that we were able to reduce the overall number of cases on campus saliva screening really increased the number of people that we could effectively screen and we were able to notify individuals who contributed to a posit a positive sample because the process allows for identification of the individual and tell that person that they needed an fda authorized test so that turned out to be a huge cost-saving measure because we're eliminating the expensive fda authorized test for people who are negative and only sending people who are very likely to be positive to get screened to get those tests so um the cost of analysis was reduced by 86 um and um because of the sensitivity of ddpcr the pools of eight worked very well without substantially increasing the sensitivity of the assay overall and csu was able to move to screening anyone who has a campus presence not just the people in residences in the spring semester and the finally the cost per positive test or the cost per case of covid19 was reduced by approximately 87 which is a huge savings so in summary asymptomatic pre-symptomatic and symptomatic people are all um shedding a virus that is detectable in wastewater ddpcr can be used to detect it and this is part partially due to its sensitivity and and sars cov2 levels can be followed in wastewater they correlate well with infection rates in very large populations however due to the large number of unknowns of variability which have more of an impact in a smaller population it's really not possible to directly correlate levels in wastewater with numbers of infected individuals trends over time an absence of detectable stars covetous small populations can aid assessment regarding need and timing of individual testing and so we had some very um important keys to success that we were able to identify as a result of all of these projects and one is that early buy-in from stakeholders that included seed money and in-kind donations were really important uh to the success of getting these uh projects started in a in a timely way that allowed the university to to um have uh mitigations in place um right at the start of the fall semester um the availability of relevant expertise was really important in understanding what we were finding partnerships of public private and academic institutions where key exchange of information was really important prior investment in droplet digital pcr technology allowed the campus to be poised to get these projects up and running quickly and the robust supply chain for ddp on pcr reagents was really important to be able to maintain the project um when there was a lot of competition for other types of pcr supplies and then uh you know we depended on free and rapid sharing of information and resources by others and we tried to pass it on for those who came to us to ask questions about how to do this and finally the shared interests and motivations not only within our own community but with public private and academic institutions and other communities was very important to the success any project like this involves the work of just a countless number of people and this is not even you know the full list of everyone but uh we want to acknowledge everyone who contributed to the to the success of this project so thank you great yeah thank you very much heather and thank you to all three of you for this for this really nice presentation as a quick reminder if you have a question type it into the q a panel and we would also ask our ask our participants to take a brief moment after the webinar has ended to take our exit survey and give us your feedback uh let's see we had a ton of questions we will certainly not get to all of them but let me see what we could address in the time that we have um [Music] there were several questions around the procedure that you had after finding positives in a certain building in a dorm what what was your next step would you test everybody in the dorm would you shut the door down what was the protocol so um we only shut two dorms down once um and that was in response to that first time we had those extremely large signals because we didn't know what they meant um so the procedure was that if we saw a signal in a dorm we would ask the residents to go get tested and housing and dining was wonderful in helping to track the people as they tested to help assist with the people who were in quarantine or isolation by providing food transportation support and then if we and then we would look at the signal and then if the signal decreased substantially or resolved um we would assume that we had found the positives that were to be found and um and we started again the next time that we had data if the signal was not resolving we assumed that there were some members who had not been compliant that still were infected and so housing and dining would go back and tell them that they needed to go into quarantine um our last resort was that if they did not get tested they would be transferred to quarantine but we never actually had to do that because as soon as people were told to start packing um they found that they had time to get tested after all okay great yeah thank you um you mentioned that the wastewater testing did not require an irb just did the saliva screening require one so the saliva screening is not a research study so it doesn't require irb we did just like with the wastewater testing in order to do our due diligence we did send a question to the irb and say you know this is a public health activity where you know if we were to ever publish the results we would um we would get it through the irb but as a public health activity our understanding is that we need consent from people but we don't need irb review and do you agree and they did agree with that we did send um uh the consent uh to the office of uh um uh of general counsel for the university to make sure that we were appropriately um addressing all the needs uh both legal and we had a regulatory expert to help us also review that consent to make sure that it it was within the standards but everyone who contributed a sample did consent okay great um how sensitive was the wastewater assay i mean could you get down to a single infected individual do you have any idea how sensitive it was i'll i can tell you that one so we know we we know we can reliably detect three thousand to five thousand so that copies per liter um my feeling is that that in a small um small population like a dorm i think probably one one we would be able to detect them but they have to poop at the right time so okay we don't have continuous sampling if you p if the person is pooping between the times that we sample we we wouldn't necessarily catch them and that's one of the reasons we're sampling three times a week because whoever is um shedding into their feces will will be doing that for more than one one day um and so if we miss them one day we should get them the next day um so we can't guarantee that we would detect one person but we feel that if if that one person um did poop at the right time we should be able to catch them okay we also had quite a large number of very detailed technical questions maybe i can just ask a couple um how much initial volume did you use for the wastewater analysis is it only 40 milliliters from one liter of entire sample so the entire sample i think is like gallons like it's not so it's we take like i think it's 85 mils every 15 minutes and so we have a very big sample like we and from that they shake it up and then we get three tubes at 40 mils each we process one so if one 40 ml sample goes into the um into gets goes through all the processing and then one goes into the freezer and we keep one as back up um and throw that away if we don't need it okay and uh somebody wants to know whether you've seen any effect of pasteurization of the wastewater i don't know if you did that on the ability to detect the saskof2 virus by a digital drop like pcr so initially with the protocol that we started with did involve heating the heating the sample before it was used um we decided against doing that one because we were worried it was going to hurt the virus and the other another because it just took an hour which is another hour onto our processing time um my but in the um saliva processing lab i know they do heat for an hour to 70 degrees and it they still have no problems detecting it and i have read that there's not a huge effect of of the pasteurization process and it does kill all the other nasty viruses so if you didn't have a bs a biological safety cabinet maybe that would be one way to get around the risk of of handling this material i think like it would be okay to heat it okay maybe one more general question do you think once the pandemic is over this kind of wastewater analysis could still be useful to detect other types of outbreaks for example i i can take that one i think um it already has been used for other types of infectious diseases as dr delong mentioned and we are very interested in looking into a sort of not just other outbreaks but even a multiple pathogen detection approach we think that could be a really interesting strategy to pursue okay great and um are you going to publish your data or have you already published it in any way like a pre-print or something carol did you want to take that one we're we're working on um publishing some parts of it we did develop the essay um in collaboration with the company which makes some at some of the um publishing parts uh more difficult but we hope to publish some of it and um if you have questions for specific questions you're welcome to contact us directly yeah the other thing that we've been considering is um is potentially publishing the entire approach that the university has taken um just because you know we did feel like it was quite successful in keeping the campus open to our target date okay great thank you i'm sorry we've actually run out of time even though we still have a lot of questions so i'm hoping that somebody will be able to follow up with you via email if we did not answer your question today um so let me thank our speakers susan delong carol bullis and heather pitkook and our sponsor biored laboratories and as a reminder please look out for the survey after you log out to provide us with your feedback and finally if you missed any part of the seminar or you would like to listen to it again we will send you a link to an archived version via email so with that thank you very much for attending this genome what we now you