WEBVTT Kind: captions Language: en 00:00:00.800 --> 00:00:06.560 Thanks for tuning in for this. I'm going to  talk today about the durability of immunity   00:00:06.560 --> 00:00:12.480 following affection by SARS-CoV-2 and this  is a collaboration led by me with a bunch   00:00:12.480 --> 00:00:17.360 of members of my lab and some other folks who  have worked on it with me or are listed here. 00:00:21.280 --> 00:00:28.400 So start from the beginning the durability of  immunity upon natural infection has been called by   00:00:28.400 --> 00:00:34.480 many the greatest unknown factor of the COVID-19  epidemic. Just to give you some examples of this,   00:00:35.440 --> 00:00:42.640 on the left you see a nature video talking about  the big questions six months on. The major one   00:00:42.640 --> 00:00:46.560 of which that they highlighted was what is the  durability of immunity once you get infected?   00:00:47.200 --> 00:00:52.160 Secondly on the right is a news article from  STAT seven months later what we know about   00:00:52.160 --> 00:00:56.240 COVID-19 and the pressing questions that remain  and in there, you'll find that one of the ones   00:00:56.240 --> 00:01:01.520 they highlight to the greatest degree is what  is the durability of immunity of COVID-19? 00:01:01.520 --> 00:01:08.560 Now most of the studies that have spoken about  this topic have done longitudinal observation   00:01:09.120 --> 00:01:11.840 looking at the decline of  some kind of antibody response   00:01:12.960 --> 00:01:17.760 after infection by SARS-CoV-2.  The difficulty with doing that   00:01:17.760 --> 00:01:25.840 on SARS-CoV-2 and getting a result is that the  decline of antibody level occurs fairly slowly   00:01:25.840 --> 00:01:29.280 and as you can see in this one of the first  papers that came out. Essentially there's   00:01:29.280 --> 00:01:34.000 not much decline here there's just the increase  consequent to infection in these three different   00:01:34.000 --> 00:01:38.880 IgG types and then sort of a leveling off and in  a few individuals there's a little bit of decline,   00:01:38.880 --> 00:01:43.040 but really on average you're not seeing the  decline yet. And that's characteristic actually   00:01:43.040 --> 00:01:48.640 of all coronaviruses that they tend to decline-  start declining around 90 days and that's already   00:01:48.640 --> 00:01:52.400 three months in advance and a short epidemic  like this there's just not enough information   00:01:52.400 --> 00:01:58.720 to figure out what the decline of antibodies  is and what its correlation with immunity is. 00:01:58.720 --> 00:02:05.920 So that news that in fact you know  that in fact antibodies may decline   00:02:06.480 --> 00:02:12.960 sparked some fears that immunity to COVID-19  wanes fast. This is just a news article saying   00:02:12.960 --> 00:02:19.760 studies show coronavirus antibodies may  fade fast raising questions about vaccines. 00:02:19.760 --> 00:02:24.800 And then but you can find answers sort of both  ways. So many have deemed the question impossible   00:02:24.800 --> 00:02:28.400 to address. This epidemic is so recent. There  have been few well-monitored re-infections.   00:02:29.200 --> 00:02:34.320 So here on the left you see one saying my patient  caught COVID-19 twice so long to herd immunity you   00:02:34.320 --> 00:02:39.360 know. Is there no immunity that you get from this  disease? And then another article simultaneously   00:02:39.360 --> 00:02:44.800 or very nearly simultaneously can you get  COVID again? It's very unlikely experts say. 00:02:44.800 --> 00:02:48.880 So it would be really great to answer this  question and I'm here to say we can answer   00:02:48.880 --> 00:02:52.640 this question. On the contrary there's not  nothing known but we do know something about   00:02:52.640 --> 00:02:57.760 the durability of immunity to SARS-CoV-2 and the  reason we know it is because of the historical   00:02:57.760 --> 00:03:04.320 contingencies of evolutionary biology. SARS-CoV-2  is a coronavirus like multiple other coronaviruses   00:03:04.320 --> 00:03:10.560 that are listed here SARS-CoV-1 the three- the  human coronaviruses you may be familiar with   00:03:10.560 --> 00:03:14.080 that are- cause the common cold regularly MERS  [Middle East Respiratory Syndrome] is another   00:03:14.080 --> 00:03:20.880 example and those coronaviruses all have genetic  differences that tell us how closely related they   00:03:20.880 --> 00:03:25.840 are to each other. So we can learn something  from the other coronaviruses about SARS-CoV-2   00:03:25.840 --> 00:03:29.440 and there's a very rigorous way of doing  that and that is by phylogenetic analysis. 00:03:29.440 --> 00:03:34.400 So if we look at the different viruses, we can  see how closely related they are, and the fact   00:03:34.400 --> 00:03:39.840 is that viruses can't evolve super quickly. They  have limits the rate at which they can evolve and   00:03:39.840 --> 00:03:44.320 how fast they can change, and we have methods in  evolutionary biology to understand how fast they   00:03:44.320 --> 00:03:49.280 change across a phylogenetic tree such as this one  which we reconstructed from the genome sequences. 00:03:49.840 --> 00:03:55.600 So the kind of data that we want to take into  account to do this is to look at continuous issue-   00:03:55.600 --> 00:04:00.160 we want to do continuous ancestral and descendant  state inference under a Brownian-motion model of   00:04:00.160 --> 00:04:06.640 trade evolution and the kind of data we're looking  at is this Anti-N which is one of the genes of the   00:04:06.640 --> 00:04:13.920 coronavirus IgG which is just an antibody type  across time. So this paper by Edridge et al. very   00:04:13.920 --> 00:04:19.200 fortunately looked at the three- I'm going to talk  about three of the seasonal coronaviruses here   00:04:19.200 --> 00:04:24.640 over time and examined when they had peaks:  these starry points which indicate that there   00:04:24.640 --> 00:04:29.920 was an infection in an individual. These are  in the seasonal coronaviruses and then that   00:04:29.920 --> 00:04:35.200 allows us both to understand what levels of  antibody allow an individual to get infected   00:04:35.200 --> 00:04:40.960 but also how long it takes for them to  decline between times of being infected. 00:04:40.960 --> 00:04:46.640 So analyzing both of those things across the  coronaviruses, and I'm sorry this appears   00:04:47.280 --> 00:04:52.720 much more pixelated on the screen than it did when  I made it, but we're able to actually characterize   00:04:52.720 --> 00:04:58.320 the peak normalized antibody levels over time  based on that kind of data for these human   00:04:58.320 --> 00:05:04.080 coronaviruses. We're also able to characterize the  daily probability of infection or how long over   00:05:04.080 --> 00:05:10.240 over time it takes- how likely are you be infected  as your antibody level declines. This is for these   00:05:10.240 --> 00:05:15.280 seasonal coronaviruses for which that longitudinal  data over many many years that was over decades   00:05:15.280 --> 00:05:20.320 was collected. Now in addition to understanding  this about the seasonal coronaviruses, they're   00:05:20.320 --> 00:05:24.560 embedded in this phylogenetic tree that  it allows us- enables us to also study   00:05:24.560 --> 00:05:29.120 the seasonal coronaviruses for which we don't  have daily probability of infection data, but   00:05:29.120 --> 00:05:37.600 for which we do have some data on the IgG, IgA,  IgM on this information about the antibody level. 00:05:37.600 --> 00:05:43.200 So combining that data based on what we  already know about SARS-CoV-2, SARS-CoV-1,   00:05:43.200 --> 00:05:48.960 MERS-CoV, and these three seasonal coronaviruses  combined with the information on daily probability   00:05:48.960 --> 00:05:52.800 of infection, we're able to use phylogenetic  methods to impute what the daily probability   00:05:52.800 --> 00:05:57.440 is an infection and what the rest of the  antibody decline is probably like for   00:05:57.440 --> 00:06:04.000 each of these zoonotic coronaviruses enabling  us to estimate the time of waning immunity,   00:06:04.000 --> 00:06:08.160 the problem of infection over time, and  the probability density of re-infection. 00:06:08.160 --> 00:06:12.640 So what this gives us is this probability of  density of reinfection over time. This is an   00:06:12.640 --> 00:06:16.480 axis of days on the axis here, and you can  see that although there's some differences   00:06:16.480 --> 00:06:20.640 between the antibody decline and the daily  probabilities of infection among these different   00:06:20.640 --> 00:06:27.040 diseases, the overall distribution of  when you get in of a time of reinfection   00:06:27.040 --> 00:06:30.400 does not appear to be all that different  between these different diseases. 00:06:30.400 --> 00:06:36.720 Consequently what we can conclude in our main  analysis is the following. That the median time   00:06:36.720 --> 00:06:43.840 to reinfection by SARS-CoV-2 appears to be about  1 year, 7 months. SARS-CoV-1 is quite longer.   00:06:43.840 --> 00:06:49.840 SARS-CoV-2 by our best estimate SARS-CoV-2  again 1 year, 7 months. MERS 1 year, 4 months   00:06:49.840 --> 00:06:54.000 although it's never been you know you don't  get reinfections because the zoonotic disease   00:06:54.000 --> 00:06:59.120 does not spread from human to human. And  for the different seasonal coronaviruses,   00:06:59.120 --> 00:07:03.200 we get somewhere between four to  six years for the duration of that. 00:07:03.200 --> 00:07:07.200 So what are my conclusions? They are  that this ancestral and descendant states   00:07:07.200 --> 00:07:11.680 estimate of the timing of the waning of immunity  can facilitate a quantitative analysis of all   00:07:11.680 --> 00:07:16.400 policy decision making with regard to individuals  who've recovered from COVID-19 and who may be   00:07:16.400 --> 00:07:22.480 viewed as immune to reinfection but may not  be after some time. Secondly, the durability   00:07:23.840 --> 00:07:27.600 of immunity has implications for the  deployment of recovered health care workers,   00:07:27.600 --> 00:07:32.400 of travel restrictions, decisions on how  students retain their education, prospective   00:07:32.400 --> 00:07:36.800 vaccination protocols for clinical trials as  well as the opening and closing of economic   00:07:36.800 --> 00:07:43.120 sectors in response to predictive models of the  epidemic. Our estimate argues strongly against   00:07:43.120 --> 00:07:47.360 the claim that a long-standing resolution of  the epidemic could arise due to any kind of herd   00:07:47.360 --> 00:07:52.720 immunity from natural infection. Such a strategy  jeopardizes millions of lives entailing high rates   00:07:52.720 --> 00:07:58.560 of infection morbidity and death every 1.5 years.  It provides some guidance as to the likely time   00:07:58.560 --> 00:08:02.880 scale of immunity conferred by a typical vaccine.  I'll have a caveat about that in a moment.   00:08:03.440 --> 00:08:08.400 This approach has general applicability to rapid  prediction of parameters for any novel pathogens   00:08:08.400 --> 00:08:12.560 provided they're embedded in a plate containing  three or more previously studied human pathogens. 00:08:13.440 --> 00:08:17.680 I just want to give you a few caveats to make sure  it's clear what we can and can't say from this.   00:08:17.680 --> 00:08:21.200 The research addressed durability  of immunity in response to typical   00:08:21.200 --> 00:08:27.600 natural infections under endemic conditions. The  durability in response to vaccination requires a   00:08:27.600 --> 00:08:33.040 little further analysis because vaccination  doesn't give you the same antibody level   00:08:33.040 --> 00:08:38.880 response that natural infection does necessarily.  And also we're under pandemic conditions until the   00:08:38.880 --> 00:08:44.240 world's population has sort of been exposed to the  disease or vaccination and so that means that some   00:08:44.240 --> 00:08:49.360 of the timings are going to be slightly different  for some complicated epidemiological reasons. Our   00:08:49.360 --> 00:08:53.280 estimate and some certainty should be understood  as a prediction of the average durability not   00:08:53.280 --> 00:08:57.760 universal to everybody. We know that different  antibody levels are sparked by different   00:08:57.760 --> 00:09:03.280 infection levels and by different vaccines so each  individual is slightly different this is what's   00:09:03.280 --> 00:09:09.040 typical. And because SARS-CoV-2 is a novel virus  the human immune system, the reinfection may not   00:09:09.040 --> 00:09:14.640 exhibit the same severity as first infections.  We will have to see as time goes on. Thanks very   00:09:14.640 --> 00:09:20.000 much for the time and from the support from  NSF to do this very interesting research.