COVID Transmissions for 7-21-2020
Better living through vaccination!
Greetings from an undisclosed location in my apartment.
It has been 247 days since the first documented human case of COVID-19.
Housekeeping: Shout out to our paid subscribers for supporting the new graphics; the new header image today was paid for by their support.
In-depth today will focus on major vaccine papers published yesterday. Papers are linked in the Headlines section.
As has become usual, glossary terms are bolded words with links to the running newsletter glossary.
If you like what you see—or what you might see in the future—tell others about it so the newsletter continues to grow:
Now, let’s talk COVID.
Vaccine trial results:
The Lancet published two major vaccine trial results in the last day, from early trials of a Chinese adenovirus-based vaccine and also the Oxford ChAdOx1 vaccine.
Chinese “Ad5” vaccine: https://www.thelancet.com/lancet/article/s0140-6736(20)31605-6
ChAdOx1 nCoV-19: https://www.thelancet.com/lancet/article/s0140-6736(20)31604-4
We’ll look more closely at these in the in-depth section.
Viral load:
One of the big questions in COVID-19 has been the amount of virus that grows in patients—the “viral load.”
This recent report shows number of genome copies in respiratory fluids from a large number of patients: https://www.sciencedirect.com/science/article/pii/S1386653220301815
The most interesting bit is in figure 1B:
Image is a bar chart of the number of positive samples (y-axis) that had a specific log10 of SARS-CoV-2 virus copies. Adapted from Journal of Clinical Virology. This demands some explanation. Each step along the x-axis here is half an order of magnitude increase in number of genome copies found in the patient sample; the graph counts how many patients fell into each category for the count. The range in this graph goes from ~10 copies to ~10 billion copies, though the number of patients with the extreme counts are very small. What’s interesting is that there is a fairly even distribution of patients with counts ranging from 10,000 genome copies per mL to 100 million genome copies per mL. This implies that there really is substantial variation across people in terms of the amount of virus they produce, and may bring us closer to understanding why some people are capable of triggering “superspreader” events and others are not.
What am I doing to cope with the pandemic? This:
Reading
This time around I want to share one of the most delightful short stories in the history of science fiction—Naomi Kritzer’s “Cat Pictures Please”: http://clarkesworldmagazine.com/kritzer_01_15/
Listening
Met Opera on Demand had The Barber of Seville free last night, so of course that was something we had to watch. They stream a different opera each night: https://www.metopera.org/season/on-demand/
Cooking
At some point this week I’m going to make sweet potato and beet fritters based on this recipe: https://www.blissfulbasil.com/beet-and-cumin-fritters-from-peace-parsnips/
Almost certainly going to make a lot of substitutions, though, so I’ll update with details when I’ve made them.
We have new human vaccine data, as mentioned in the “Headlines” section.
All in all, ChAdOx1 remains the vaccine candidate about which I am most excited. The paper looked at both “prime” (single-dose) and “prime-boost” dosing, which is a first dose followed by a second, booster dose. If we compare back to the Moderna vaccine, their vaccine appeared to require a booster dose. The ChAdOx1 vaccine…well, let’s take a look. The virus neutralization looks good:
Box-and-whisker graph shows increase in virus neutralization by serum from vaccinated patients vs. plasma from recovering patients infected with virus (convalescent plasma). The boxes by day 35 and day 42 are in the same neutralization range as the box for the convalescent plasma. Adapted from The Lancet.What we are looking at here is similar to what we looked at for the Moderna vaccine a few issues back. But it's important to note you can't cross-compare data from different clinical trials except under very specific circumstances. The patients and experiments are different. What we see here is that the median (the line across each box) gets close to what the convalescent plasma group had by day 35, and the interquartile range (the 75th and 25th percentiles for response) are also within the range for the convalescent plasma. Then we see the response is stable at least through day 42 around that range. Notably (and not shown here), the vaccine produced comparable responses in both the single-dose and the boosted condition.
Problem is, the numbers of patients are very small, so it is quite hard to be sure that this would bear out in a bigger population.
The paper also has data on T-cell responses. We talked about T-cells in the last issue; they are another part of the immune infrastructure and are very important to both regulating antibody responses as well as directly fighting infections. They were measured by an assay that concentrates these cells out of blood samples, and then looks for their specific activity against SARS-CoV-2:
Image is a dot plot comparing two vaccination conditions with the control group (an irrelevant vaccination). Details of image are described in next paragraph. Adapted from The Lancet.This shows that there was a T-cell response. We have no idea from this what type of T-cell response, and we cannot determine what this means for protection. What we know is there was a T-cell response in lots of the patients in the trial, and that T-cell response was a lot better than the control (the panel on the far left in blue).
So we know that the Oxford candidate elicits both antibody and T-cell responses. The safety data make up the beginning of the paper, and there were no surprises there. The vaccine will likely be safe in larger trials.
But like we have discussed before, we don’t know if any of this represents a correlate of protection. So while this is all very encouraging, the real test of this vaccine—and any other—will be in a field trial where we see the ability to protect people against catching the virus.
The Chinese Ad5 vaccine results also show antibody responses and also show T-cell responses. However, the Chinese Ad5 vaccine doesn’t show a very robust antibody response. Not all patients in their trial “seroconverted,” which is a term for showing a robust antibody response. The authors of the study put forward some reasons for why this may have happened, but among those reasons the one I consider most likely is inherent to the vaccine’s design. This vaccine uses a human adenovirus as its vector. Adenoviruses cause common colds (among other viruses that do this), and so there is preexisting immunity to specific adenoviruses in the human population. In this trial, about half of the participants had preexisting neutralizing antibodies against the vector. That is likely to have interfered with the ability of the vaccine to function properly.
I don’t think this is or will be a problem for the Oxford vaccine, because the vector for that vaccine is a chimpanzee virus and does not circulate among humans, but this is my optimistic guess. Dr. Heather Lander, a fellow virologist who I follow on Twitter, pointed out that data about immune response to the vector were not provided in the Oxford vaccine paper. There is a possibility that the ChAdOx1 vector itself produces a strong immune response that blunts the reaction to the SARS-CoV-2 antigens that are included in the vaccine. The immune system has limited resources, so a strong response to something else in the vaccine could be a problem. While I think this is a bigger issue when there is preexisting immunity like for the Ad5 vaccine, she’s right to point out that any vectored vaccine can have this problem:
I’m sure everyone hopes this is not an issue for this vaccine, but I’d like to see the data either way.
Again, though, a lot of this is academic. We don’t know what a protective response really looks like, or how to look for one, except to get these vaccines out in the field and see if vaccinated people are less likely to get sick with COVID-19 during such a field trial.
These papers show us that the vaccine candidates do something, but we don’t know what the something is. And there are still additional questions. I think this tweet thread from Dr. Peter Hotez, an absolute legend in emerging and neglected diseases, tells us a lot:
Dr. Hotez’s thinking here is very similar to my own. This is no accident, because he was one of the first experts who I looked to when I saw these papers. Dr. Hotez was already working on an emerging SARS-like coronavirus vaccine before the pandemic, and his funding to develop it further was cut. If it hadn’t been, we might be in much better shape right now. So I put a lot of stock in his opinion.
Having read the papers, I still agree with him. These are interesting results that tell me we should do Phase 3 trials at least with the Oxford candidate. I could see doing a Phase 3 trial with the Ad5 vaccine also, but I’m more on the fence about that one because of the design of the vaccine as well as the results. I am encouraged by the safety profile of both vaccines and don’t expect big surprises in the Phase 3 trials.
But, none of the data so far show me a vaccine candidate that is ready for widespread use. I think it is very important that we get it right the first time with these vaccines. If we were to authorize any of these before Phase 3 data were available looking at the actual protective effect, it would be a mistake. The immune system tends to react most strongly to its first experience with a given antigen. It’s not impossible to get a protective response again later, but the first exposure can be the best opportunity to get durable protection. For this reason, I think it’s important that we have truly good vaccine candidates that we are confident elicit a protective response before we start vaccinating the general public.
Another important consideration is safety. In the trials so far, none of the safety signals observed have been deal breakers. We have seen things like injection site irritation, tenderness, chills, low grade fever, etc. However, the patient numbers have been small. We will need to be practical about these vaccines. In trials with a few hundred patients, we’re not very likely to detect safety issues that occur in about 1 in 1000 patients or fewer. However, in the larger Phase 3 trials with tens of thousands of patients, we have a good chance at detecting these issues. If a vaccine against COVID-19 has a serious adverse event that kills 1 in 1000 people who receive it, then it’s not a vaccine that I want to see approved. The reasoning is quite simple: the virus kills between 1 in 100 and 1 in 1000 people who get it, and we don’t know exactly what percentage of people who are exposed end up getting the virus. We do know that a lot of people will get the vaccine. Let’s imagine a hypothetical group of 100,000 vaccinees. Consider a New York City-like outbreak scenario where 20% of these people ultimately get COVID-19. That’s 20,000 cases. If 1% of them die, that’s 200 deaths. If 0.1% of vaccinees die, that’s 100 deaths. In a realistic scenario, the vaccine would not be perfectly protective—perhaps it would be 70% or 80% protective. So with this hypothetical vaccine, 150 people might die instead of 200. This is not a good benefit-risk balance.
This oversimplified thought experiment doesn’t consider the long-term impacts of virus infection even among those who do not die—but it also doesn’t consider potential long-term impacts that a bad vaccine could have. The point of this isn’t to put anyone off the idea of a vaccine, nor is it to suggest that I expect this situation. It’s to underscore that we need to do the large trials with tens of thousands of patients so that we can make an informed decision about the potential benefit-risk tradeoff that we might expect.
I do have some concerns that we will not have a lot of opportunity to follow vaccinees to see if there are longer-term impacts before we roll out the vaccine. Thankfully, most of these candidates are made of things that either come from the virus itself or have been tested in humans before with long follow-up. If the components that originate with SARS-CoV-2 have long-term impacts on people, we will hopefully learn about these impacts from recovered patients before a vaccine is deployed, and if the components that we have previously studied have issues, then we will learn about that from ongoing followup of the previous work. I don’t think it’s very likely for us to discover some big surprise safety issue 6 months after the vaccine rollout, but never say never.
I’m encouraged by the possibilities for a safe, effective vaccine here. My vaccine forecast remains the same as it was when I wrote about timing last week. Early 2021 is when I expect to see a vaccine realistically deployed, assuming that at least one Phase 3 trial goes well. So much depends on those trials.
I welcome any feedback on structure and content. I want this to be as useful as possible, and I can only make that happen with constructive comments.
This newsletter will contain mistakes. When you find them, tell me about them so that I can fix them. I would rather this newsletter be correct than protect my ego.
No corrections since last issue.
See you all next time.
Always,
JS