Vaccines must undergo a series of tests in order to establish that they are safe and effective before they are available to the public. In the US, the FDA must grant approval for a drug or vaccine to be sold. A company outside of the US can apply for FDA approval, but of course they still must meet the FDA standards. Other countries have their own regulatory systems for approval.
The FDA requires three phases, although two phases are often combined. In order to even start phase 1, though, a vaccine will first be tested in animals. Animal trials test for safety, and in the case of a vaccine, it's much easier to test for efficacy, because you can expose the vaccinated animal to the pathogen to see directly if the immune system fights off the pathogen. It is possible for a vaccine to produce a good immune response, but once the pathogen is introduced the immune response from the vaccine ends up being harmful. There was some real worry that this would be the case for SARS CoV-2, as some animal trials with a different coronavirus found that the vaccine actually made the pathogen worse. But happily, none of the animal trials for SARS2 vaccines have found any hint of that sort of outcome.
Normally this sort of direct challenge trial with a potentially fatal pathogen would not be done in humans. But in this case, the AstraZeneca team has said that they intend to do human challenge trials by the end of the year, in parallel with regular phase 3 trials.
Phase 1 is technically just to test for safety, although if you're going to give an experimental vaccine to some people, you might as well do the tests to see if it is likely to be effective, in which case it becomes a combined phase 1/2 trial. Phase 2, if you are doing it separately, tests for likely efficacy. Both phase 1 and 2 are small studies, with dozens to maybe a few hundred volunteers, who must be young and healthy.
Phase 3 is similar to phase 2, but with more people (thousands) and a wider range of people in terms of age and health status. We know that in general, older people have weaker immune systems, and tend to have weaker responses to vaccines. At this point, only Pfizer has published data on older people, but all of the large phase 3 studies ongoing now are testing older people. Unfortunately, the only country that is testing in children is China. So this is going to leave parents in other countries with the decision of whether or not to give their kids a vaccine that has not been tested in kids.
At this time, two vaccines have been approved for conditional use, although neither has been approved for use in the US or EU. One is the CanSino Biologics vaccine approved for limited use by the Chinese military. It has finished phase 2 testing, but not phase 3. The other is the vaccine developed by the Gamaleya Research Institute in Russia. It has finished phase 1 testing.
Eight vaccines are currently in phase 3 trials. These are the ones that most of the world is waiting for. The three that are most relevant to the western world are Moderna's mRNA-1273, the Pfizer / BioNTech collaboration's BNT162b2, and the U. of Oxford / AstraZeneca collaboration's ChAdOx1 nCoV-19. I'm going to call these by the names of the largest company involved: Moderna, Pfizer, and AstraZeneca. I will outline what we know about these three below.
The other five include four developed in China: CanSinoBio (the one approved for military use in China), Wuhan Institute of Biological Products, Beijing Institute of Biological Products, and Sinovac each developed a vaccine based on inactivated virus. The last one is an Australian trial of the BCG vaccine, normally used to prevent TB, but in this case it's being tested against covid-19.
AstraZeneca / Oxford -- ChAdOx1 nCoV-19
This one was initially the front-runner in the competition to be the first through the approval process, but Moderna and Pfizer are catching up. Although a common concern is that these vaccines are being developed too quickly, the people at Oxford have been working on the technology for this vaccine for a very long time, which gave them a big head start. They were trying to make a MERS vaccine, and so had inserted the MERS spike protein gene into their fancy vector, ChAdOx1. This is a chimp adenovirus that is missing a gene needed to replicate. In order to grow it for the vaccine, they inserted that one missing gene into a cell line, HEK293, which is used to grow the viruses. These adenoviruses containing the gene for the MERS spike protein are injected as a vaccine. Although they can't replicate to make new viruses, they can enter cells and do what viruses do -- hijack the cell's protein-making machinery to read the code of the foreign genes that it has brought into the cell. So the cell makes the MERS spike protein, which the immune system recognizes as foreign, which triggers the development of immunological memory. The reason that they are using a chimp adenovirus is that many humans already have immunity against human adenoviruses. They screened many different adenoviruses to find one that would infect human cells and not be attacked by the immune system before it can deliver its payload. If you search Google Scholar for ChAdOx1, and restrict the date to 2019 or earlier, there are 303 academic papers available. So the vector is not new. And importantly, the vector has been tested in older people, who had about the same immune response as younger people. When covid-19 happened, the researchers could quickly put the gene for the SARS2 spike protein into the vector and start animal trials immediately.
Animal trials
The results of the animal (mice and monkeys) trials are published here. I'm looking primarily at the monkey data, since they are more similar to humans than are mice. They used three groups of six rhesus macques -- one got a single shot of the vaccine, one two shots 28 days apart, and one got a placebo vaccine of the vector containing the gene for GFP.
All of the monkeys that got the real vaccine had an immune response. In the graph below, red symbols are the moneys with the single shot, blue are those with the shot plus booster, and green are the controls. The left graph shows antibodies, the middle neutralizing antibodies (which can block the virus from infecting cells) and the right graph shows T-cell responses. Both antibody and T-cell responses are better for the monkeys with two shots rather than just one, which is not surprising. The booster shot is designed to mimic a natural infection, where the immune system sees foreign material for an extended time, rather than just once. All of the monkeys that got the vaccine had an antibody response. Unfortunately, the T-cell response was above baseline for only 4 of the 6 monkeys.
The monkeys were then given a high dose of virus in both the upper and lower respiratory tract. The result was good but not as good as we'd like. The best result is shown in the graph below. The green symbols show the control monkeys are testing positive for virus growing in fluid from their lungs, whereas the immunized monkeys have at most a low level of virus that goes away quickly. This is why the title of the paper is "ChAdOx1 nCoV-19 vaccine prevents SARS-CoV-2 pneumonia in rhesus macaques." The best news is that the vaccine protects the monkeys lungs.
The less-good news is that the nose swabs are showing virus growing in the upper respiratory tract. This is virus detected by PCR, so not the same as infectious virus. They did test for infectious virus as well. That data in "extended data table 1" below shows that the monkeys with two shots (prime-boost) had less infectious virus for a shorter period of time. The monkeys with one shot (prime) were about the same as the controls with respect to live virus in their nasal swabs. Less infectious is good, but we'd rather see not infectious at all.
Something I missed the first time I read this paper in preprint form is that the vaccinated monkeys were not free of symptoms when they were exposed to the virus. This data is presented differently in the preprint than in the final paper that is now available. Below, the left graph is from the preprint, and the right one from the final paper. The left graph looks like the vaccine led to milder symptoms for a shorter period of time, although the error bars are larger than I'd like to see. The right graph, which separates out the monkeys that got one shot vs two, looks like a rather messy result, with the two-shot monkeys actually doing worse than the one-shot monkeys at later time points. If you lump all the vaccinated monkeys together, the difference is statistically significant, but still, this is a bit concerning.
They then killed the monkeys (sorry animal lovers!) and looked at their lung tissue. The left image is from a one-shot monkey, the middle a two-shot monkey, and the right a control monkey. The left and middle look totally normal. The right looks like a really nasty case of pneumonia. This is good news, and consistent with the finding of low or no virus in the vaccinated monkeys' lungs
They also looked at cytokine levels for the monkeys. The little paired asterisks show statistically significant differences. Interferon gamma, is higher for the vaccinated monkeys, probably due to T-cells formed during exposure to the vaccine. IL-10 and IL-13 are both lower for vaccinated monkeys. These are both cytokines associated with a Th2 response, which makes sense. Fighting a virus requires a Th1 response, so the exposure to the viral vaccine pushed the immune system toward Th1 and away from Th1. The bad thing here is that there isn't a reduction in IL-6 with the vaccine. This is an inflammatory cytokine that is highly elevated in people who have the most severe cases of covid-19. On the other hand, none of the control monkeys are showing the crazy high IL-6 levels that are happening in some humans, so maybe monkeys are not a great model for that particular part of covid-19.
Human trials
The combined phase 1 and 2 results for testing this vaccine in humans are available here.
1077 healthy people aged 18-55 were randomized to receive the ChAdOx1 vaccine or a different vaccine as a control. The nice thing about using a different vaccine is that subjects don't know if their side effects are due to the experimental vaccine or the control vaccine, so they can't guess which group they are in. This is a fairly large group for phase 2, so we have the potential to get more data about rare side effects. The dose was twice that given to the monkeys, which makes sense given that humans are larger than the monkeys. Ten people who received the ChAdOx1 vaccine were given a second shot as a booster. Those ten people were told that they were getting the real shot and not the control.
The side effects from the real vaccine were clearly worse than for the control (meningitis) vaccine. The symptoms experienced were the sort of thing you'd expect from activation of the immune system -- fatigue, chills, headache. Think flu-like symptoms. A few people described these as severe in the days following the injection, but none had severe or even moderate symptoms a week out from the injection. I suspect that those individuals are the ones who would have been more likely to have a nasty reaction to the real virus. Remember, severe covid is severe at least partly because of immune system overreaction.
Looking at antibody responses, those that got the real ChAdOx1 vaccine had a good response. There does not appear to be an advantage here for the booster shot. Compared to people who actually got covid, they had roughly the same level of antibody response.
A subset of people were tested for neutralizing antibodies. These antibodies block the virus from entering cells. Here there is a clear advantage to the booster shot. Some people had a weak neutralizing antibody response with one shot, but everyone who got he second shot had a solid response.
A subset were also tested for T-cell responses. Happily, the humans had *much* better T-cell responses to this vaccine than did the monkeys. This suggests that humans will have better immunity when faced with wild virus than did the monkeys. The best monkey had about 175 activated T-cells per million, and the average human is peaking around 800. There are a few humans in the one-shot group that didn't have a T-cell response. But the worst-responding humans in the two-shot group are roughly the same as the best-responding monkey. 
The official phase 3 description is here. They've changed the name of the vaccine to AZD1222. They are in the process of enrolling up to 30,000 participants. This time it is double-blind (the phase 1/2 was single-blind), and uses a saline placebo instead of a control vaccine. Instead of splitting people 50-50, the experimental group is twice the size of the placebo group. The dose is the same as the phase 1/2 trial, but this time everyone is getting two shots, four weeks apart. They are not accepting kids, but there is no upper limit for old people. They are not testing immunodeficient or seriously ill people.
They started injecting people in late July, and claim on the official forms that they expect to have their primary results by 12/2/2020. I think it's possible that they might have results sooner, since immunity is pretty clearly kicking in within a month. To see if the immunity is actually protecting people, we need to see if the placebo group is getting sick and the vaccine group isn't. Right now about 0.7 per thousand Americans is officially testing positive every day. If you have a placebo group of 10,000, that's 7 people per day on average. That's going to add up fast.
They are also planning on doing human challenge experiments, which are much faster, as the subjects will be exposed to live virus as soon as they have had time to develop immunity. They haven't given a hard date on that, although the general "by the end of the year" may mean that they are holding the option as a last resort if they don't get enough data from people catching the virus in the wild.
In April, they said that they would release results in June, and they didn't. In May, they said that they would be approved by September, and that seems like a long shot now. They have also said that they expect the vaccine to last a year, although without giving any hint of why they think that.
Moderna -- mRNA-1273
Animal trials
The most notable thing about their mice study is that even though the mRNA-containing lipid vesicles were injected into a muscle, most of the expression of a reporter gene showed up in the liver. Upon reflection, this makes sense. One of the functions of the liver is to scavenge lipids, and so it's probably taking up the lipid vesicles. There are lots of immune cells in the liver, so this might actually help the vaccine provoke a strong immune response.
The monkey study is here. They used three groups of eight monkeys each. One group got a saline control, one got 10 ug of the vaccine, and one got 100 ug. All monkeys were given a second shot a month later of the same thing they got the first time. The units here are micrograms, commonly written as ug to avoid having to look up the keyboard code for the greek letter mu. To give you an idea of how much this is, find a ruler with millimeters. A cubic millimeter of water weighs one milligram. The larger 100 ug dose is 0.1 milligrams, so a tenth of the weight of that cubic millimeter of water. The vaccine is made synthetically, not grown in cells like the AstraZeneca vaccine. Because RNA is unstable and won't just go into cells by itself, it is encapsulated in tiny lipid vesicles. Unfortunately the papers describing these lipid vesicles in detail are behind paywalls.
The monkeys all developed general antibodies. All but one (in the lower dose group) had neutralizing antibodies. The purple dots show same test run on humans who have had covid. So far so good, especially with the higher dose.
They also looked at T-cell responses. Unfortunately, they are not doing exactly the same test as was done with the AstraZeneca vaccine, so we can't do a direct comparison between the two. But we can say that with the lower dose, four of the eight monkeys had T-cell activation, and with the higher dose, all eight did. The responses were all Th1 rather than Th2, which is what you want to fight a virus.
Human trials
Moderna has said that they've done phase 1 and 2 trails, but they have only published phase 1, here. There is a supplement with additional data here. Being a phase 1 trial, it is small, only 45 people. They were divided into three groups of 15 that received 25 ug, 50 ug, or 250 ug, then the same dose 28 days later. There was no placebo group, since this is phase 1 only.
"A participant in the 250 mcg dose group had severe fever, onset the evening of the second vaccination, along with severe chills and mild fatigue, myalgia, and headache. In the early morning of the day after vaccination the participant developed recurrent severe fever, chills, fatigue, and headache, moderate myalgia and nausea, and mild arthralgia. The participant was evaluated in an urgent care center and received symptomatic treatment prior to discharge. A nasal swab specimen was negative for SARS-CoV-2 by polymerase chain reaction and positive for adenovirus by a fluorescent antibody assay. After sleeping for several hours at home, upon standing the participant was lightheaded and nauseous, vomited, and then fainted. Lightheadedness persisted for several hours. Other systemic symptoms improved over the course of the day. Mild headache was present the next day and mild fatigue was reported through post vaccination day 6."
Pfizer -- BNT162b2
General antibodies were lower in the older group, but still
higher on average than people who have had covid (far right bar)
The official clinical trial webpage lists both the b1 and b2 versions, but various media reports claim that they are moving forward with the b2, 30 ug dose, two-shot version.
Back in July, Pfizer said that they were on track to have 100 million doses of the vaccine available by the end of the year, and 1.2 billion doses in 2021.
