Again, a briefing for journalists, this time on vaccine trials, types of vaccine, efficacy and safety.
(Context: I am a two-steps expert away from vaccine development; I am a one-step away from clinical trials; I am an expert on genetics + computational data science. I am, finally, on one of these clinical trials as a participant)
Standard context: SARS_CoV_2 is an infectious virus that causes a nasty disease, often leading to death, in a subset of humans. It will continue to be a massive issue to manage until we either have good enough vaccines or good enough treatments for the disease.
A reminder of vaccine fundamentals. Vaccines are ways of "priming" our immune system to basically say "this thing is not yourself and not nice; if you see even a little bit of this in your body, execute a fast and quick immune response".
The "thing" is either part of the virus (often the bit that sticks out - Spike protein) or even the entire virus in some weakened form. Our immune system is super clever, so even getting a hint of "this is not good news" really changes how our immune system can react
To test a vaccine is like testing a drug - you need to give it or a "sham" or "placebo" alternative at random to lots of people. You then wait, carefully tracking these individuals (I do a PCR test and survey each week) and ultimately note who has developed the disease.
If the disease is higher in the people who were randomly given the placebo or sham treatment compared to the vaccine treatment you know it works. You can also monitor if people get ill from the vaccine; severe illness is called "adverse events" in the lingo.
First off, the rigour of well run clinical RCT trials is amazing. There is complete separation of who knows about whether it is vaccine or placebo (sometimes the other choice is another vaccine) from all the other people - doctors, nurses, patients, data collectors, testers.
(This blinding to everyone, and the rigour to make the randomisation information "blind" was hard won last century. People discovered if patients or doctors knew which thing they got they started biasing their reports - sometimes just a bit, sometimes quite a lot.)
Furthermore you can only "unblind" at pre-determined points, and you must follow through on your registered points to unblind and analysis. Everything has to be stated up front (there is a web site listing this. This is the trial I am on clinicaltrials.gov/ct2/show/NCT04…)
The second thing is that trials do two things - firstly they work out how well the vaccine works (thats key!). The second is that they get a good handle on safety. Both of these, but in particular the safety one, needs large numbers
You need large numbers because often severe events are rare. For a vaccine which you will be giving to often healthy people. As it happens, in this sort of infectious disease which we are trying to keep under control you also need large numbers to just have enough people infected
In the lingo of trials, the infections are called "events" and the unblinding process is triggered by a total number of events in the trial, without knowing whether they were vaccine or placebo. You hit the event level in your analysis, you can unblind, and look.
To keep everything squeaky clean, each trial has an *independent* data and safety monitoring board (DSMB) who basically both control this unblinding process *and* flag when severe events happen, potentially closing down the trial early if too many severe events happen
(a couple of trials paused due to single severe events, which is actually quite reassuring that the who system can spot them, monitor and work out)
The final unblinding is a number of events (often around 150 ish), but you can have an pre-registered "peek" earlier to see how things are going. Usually if you do a "peek" is you are not allowed to stop the trial because until it comes to the end
(there are trial designs where one can stop the trial if the results are so good that you don't need more events - adaptive trials - these make more sense for acute disease and far less sense for vaccine trials)
The Pfizer and Moderna vaccine results are these pre-registered "peeks" - and the Pfzier peek was *great*. Honestly, I'm not sure we can ask for better interim results.
Slightly frustratingly for me, the AZ/Oxford trial (which I am on) did not pre-register some peeks but one does need to be a bit philosophical about this - the full results will present soon they say and similar time to the full results from Pfzier and probably Moderna.
The vaccine/trial community seems pretty buzzed by this. One excitement about this is that the Pfzier vaccine, created by a German Biotech company, BioNTech is the more risky one, using mRNA. So the more standard vaccines are likely (but not guarenteed) to work
Lifting the lid on the vaccine types. The Pfzier/BioNTech and Moderna vaccine are mRNA vaccines. RNA is the chemical cousin of DNA, and in all life it is used a bit like the "working memory" of life's instructions, which is stored in the hard-drive like scheme of DNA.
"m" in "mRNA" means messenger, which is the main way RNA is used (I will wander through the veritable jungle of RNA biology here). This is basically "standard issue working memory of life".
Slightly amazingly, if you design a bit of mRNA (working memory) to make a bit of virus (in the SARS_CoV_2 case, the Spike protein), not only do our cells faithfully do this, but then the immune system sees fragments of this protein, and say "ooh - that doesn't look like us" >>
<< and our immune systems work up their memory of these things. Honestly, if you told me this idea 10 years ago I would have said "that's a bit farfetched". But it works. In mice - where we can do lots and lots of trials and experiments it works regularly.
And seemingly it works humans in this situation. We need to see all the results but this is a great interim look.
The AstraZeneca/Oxford vaccine is slightly more established, but also clever. Here one takes a pretty harmless virus, engineer it to be *definitely* harmless, and add in a bit of the SARS_CoV_2 virus (namely the spike program).
You then have to make plenty of this virus in biofacilities (basically - massive fermentation looking things), purify these chimeric viruses and then use them as a vaccine.
One good thing about this is that you know the actual vaccine is going in (we don't need to rely on our cells making the fragment) *and* our immune system gets a more clear cut "this is definitely not part of us" signals from the viral backbone.
One twist which seems a bit weird until you get your head around it is that Oxford vaccine uses a harmless *chimpanzee* virus which they then engineer to make it definitely harmless, and add the Spike protein to the right bit.
Why do this? Well the problem about harmless *human* viruses is that some of us have an good immune response to these viruses, and so for those people our immune system could easily say "yup, we know about this one, lets mop it up and move on" without noticing the new bit.
Using a harmless, and further deactivated (so definitely "dead") chimpanzee virus as the backbone means our immune system gets (hopefully) the full emergency "invader on the premises" response.
Other vaccines include the Russian vaccine - which uses the same technique as the Oxford vaccine but a human harmless virus (I think a mixture). One massive frustration is that the Russians have not gone through the same trial registration process
The chinese vaccine which has been partially licensed in China is more old school - you take the real virus and deactivate it chemically, and in effect inject this "dead", chemically inactivated virus. Again, the trials aren't as deep yet for this vaccine.
What about safety? As a trial participant I had to read a pretty long and scary list of **potential** side effects, but even these are considered rare or even very very rare. Most importantly, (lots of) people like me have taken slightly more risk to explore this
On safety, you can never say never about rare or very rare adverse events; rather what we can do is talk about how rare these events have to be consistent with the large number of people, and then time (in my case, since late June).
Basically people can be very confident if the trial result says the vaccine is safe (to this level of what will be "very rare" events) that is safe. Many trials have deliberately recruited some older and diverse patients to check this from all angles.
I appreciate there are some people who just don't like needles and injection (I am pretty needle phobic, and need to look the other way). But ... the rigour, the design, and the transparency of these trials is the way to know and be confident about safety.

• • •

Missing some Tweet in this thread? You can try to force a refresh
 

Keep Current with Ewan Birney

Ewan Birney Profile picture

Stay in touch and get notified when new unrolls are available from this author!

Read all threads

This Thread may be Removed Anytime!

PDF

Twitter may remove this content at anytime! Save it as PDF for later use!

Try unrolling a thread yourself!

how to unroll video
  1. Follow @ThreadReaderApp to mention us!

  2. From a Twitter thread mention us with a keyword "unroll"
@threadreaderapp unroll

Practice here first or read more on our help page!

More from @ewanbirney

14 Nov
Right. Deep Breath. RT-PCR "false positives" and Ct numbers (again). tl;dr it is complex, but the RT-PCR testing systems deployed across the world are sound and the people who run them report positives are positives and little can be improved obviously.
Context: I am a genomics/genetics + computational biology expert. I know a large number of infectious disease testing experts. I have a COI in that I am a long established consultant for a company (ONT) that makes a new test here; this gives me additional insight
There a number of classes of false positives which don't concern the current debate (eg, sample swaps, lab contamination). To repeat an early point all the people I know in this are paranoid about this, test and check in a rather detailed way and these are looooow.
Read 22 tweets
13 Nov
Small moan about models and parameters in COVID. R - the number of infections each person makes on average is a parameter but it is also something that one can measure. Each infected person has their own "R" - it is a count - and one could in theory measure all these little Rs
This R is both something one can measure and the average R, or the distribution of R is often components of model. It is a "real measurable thing" and it is "part of our COVID models".
You might want to model other things; one is the distribution of how R varies between people/events. A sensible choice is a negative binomial. Often people use k as a parameter (in fact, in other uses, they often use the letter r as a parameter, but this would be confusing!)
Read 5 tweets
12 Nov
It's great to see this paper on scaled up Cactus graphs (Progressive Cactus) - on 600 vertebrate genomes - from the great team lead by @BenedictPaten nature.com/articles/s4158…
This paper is a very much a methods paper, but I hope Benedict and colleagues will also dive into the data - I don't think we have use the realised ancestor reconstructions (reminds me of the older Enredo days - also with Benedict!). There is a treasure trove in there
Just repeat evolution I think is fascinating here, but also niche-loss pseudogenes for example.
Read 6 tweets
10 Nov
A riff on data, models and intervention in a COVID world.
When you don't have data, you absolutely need models - it helps you understand what could happen, good and bad scenarios and what you need to measure to understand more thing. With no - or little - data, models are king.
When you have data, the role of models change. In fast moving epidemic even working out what is happening *right now* is complex; your data is coming in different streams, it has biases (which change over time), technical issues (also time dependent) >>
Read 18 tweets
1 Nov
There are debates - important debates - to have in these COVID times, but there are some either stupid debates or misguided in my view debates. Here's my list with brief rejoinders.
1. False positives on tests are grossly inflating the number of cases. Straightforwardly they are not; the system understands false positives, goes to a lot of length to prevent them, and, acknowledging that they can never be 0, carefully models them in analysis.
2. "Hard" Stratify and Shield (or segmentation) is a solution. By "Hard" I mean placing all the at risk people in entirely COVID "safe" environments (extremely low risk of infection) and then having the remaining people at low risk live normal lives, and get the infection.
Read 24 tweets
1 Nov
It is somewhat hard to know which strand on England's November lockdown to pick apart - a large number of people in the press (and twitter) are commenting ("hot takes" in the US parlance) - most heat and not so much light.
Yesterday, before the announcement, I tweeted on this here:
Read 17 tweets

Did Thread Reader help you today?

Support us! We are indie developers!


This site is made by just two indie developers on a laptop doing marketing, support and development! Read more about the story.

Become a Premium Member ($3/month or $30/year) and get exclusive features!

Become Premium

Too expensive? Make a small donation by buying us coffee ($5) or help with server cost ($10)

Donate via Paypal Become our Patreon

Thank you for your support!

Follow Us on Twitter!