COVID Transmissions for 12-28-2020

The AstraZeneca vaccine mistake

Dec 28, 2020

Good morning! It has been 407 days since the first documented human case of COVID-19.

Welcome back from a brief holiday break. This week, again, will be a short week for this newsletter. I’ll be doing issues today (12/28), 12/29, and 12/30 only. I will skip 12/31 and 1/1 for New Year’s, although these would normally be days when an issue would come out.

Today I have just one headline, about the AstraZeneca vaccine trial. I wrote a lot about this, though, because I think it’s an important issue and I want to make sure everyone understands what happened. You’ll see what I’m talking about.

As usual, bolded terms are linked to the running newsletter glossary.

Keep the newsletter growing by sharing it! I love talking about science and explaining important concepts in human health, but I rely on all of you to grow the audience for this:

Share COVID Transmissions

Now, let’s talk COVID.

More information on the dosing error in the AstraZeneca vaccine trial

In previous issues, I have covered the situation with the AstraZeneca-Oxford COVID-19 vaccine trial. In this trial, a dosing regimen that provides apparent 90% efficacy for the vaccine was discovered due to a dosing accident. Some participants in the UK were given a half-strength initial dose and then boosted with a full-strength dose. Originally this group was reported by AZ as a separate trial group of interest. In the main group, the vaccine was only 62% efficacious.

After these results were reported, it was later revealed that this dosing group emerged due to an accident. The story then reported was that a contractor had misreported the strength of the vaccine batch that they had manufactured for the trial, which led to a halving of the dose inappropriately. I found this highly objectionable as something that might occur in a vaccine trial led by a major multinational Big Pharma corporation.

Now, according to Reuters, it appears that this story is also misleading. A major story from them adds complexity to this tale: https://www.reuters.com/article/us-health-coronavirus-britain-vaccine-sp-idUSKBN28Y0XU

To summarize, it is not the contractor who made an error, according to this report. The contractor accurately described the dosing strength of the batch they created, using qRT-PCR as a technique to quantify the amount of mRNA in the batch. Researchers at Oxford, who received the batch, used a different technique, spectrophotometry, to measure the amount of mRNA present. Their measurement said the batch was much more concentrated than the Italian contractors told them. So they diluted it, because they trusted their results and not the Italian results. The Oxford researchers were wrong.

Let me explain the error here, because it really is egregious. First, we need to understand both techniques. If you already know how these techniques work, I’ve added section headings so you can skim or skip right to the end. Everyone else, strap in.

qRT-PCR

For what it’s worth, the “q” in qRT-PCR stands for “quantitative.” qRT-PCR is a very accurate technique for the measurement of the amount of RNA present in a sample. It is the technique currently used in the “PCR” tests for COVID-19. It relies on simple mathematics of a chemical reaction that doubles the amount of signal at every step. The number of steps it takes to be able to detect a signal tells you how much RNA you started with. It’s elegant and it works.

I know that’s a little technical, so I’ll describe it by analogy. Eyes are not perfect tools. There are certain things that are too small for us to see. Imagine that I put a line of text on a screen in front of you, and it was far too small for you to read. Then, I ask you the question: what is the size of the font in this text? Obviously you would not be able to answer since you cannot even see the text properly. However, what if you knew that the smallest font you could read is 8 points in size? What if you could ask me to double the size of the font, at will? Imagine that you did, and I had to double the size of the font three times before you were able to read it. If it took three doublings (2 x 2 x 2 = 8) of the font size to get to 8 point font, that means the original font size was 1 point.

This is how qRT-PCR works to quantify RNA. There is a limit of detection, below which we cannot get a reliable signal from the reaction. We know how much nucleic acid is present when that limit of detection is passed. If we count how many doubling steps it takes to get to that limit of detection, we can divide by two the appropriate number of times and figure out how much was in the original sample.

PCR reliably doubles the amount of nucleic acids in a sample. It doesn’t multiply them by 1.4 or 1.7—it doubles them. As a result, this technique is very accurate.

Spectrophotometry

Spectrophotometry can also measure the amount of RNA in a sample. It relies on the fact that different substances absorb ultraviolet light to different degrees. This is true for all light, really—so let’s use an analogy again.

When you add ink to a cup of water, the water gets darker. Shine a light through that water, and less light will come out the other side the more ink has been added to the water. It is a simple concept that is mathematically described in science by something called “Beer’s Law.” My chemistry professor in undergrad described Beer’s Law as this: “The deeper the glass, the darker the brew, the less the amount of the light that gets through.” This was a professor who truly understood what undergrads are interested in.

Anyway, imagine that we prepare an example of our ink-water mixture carefully. We have a machine that can measure the amount of light added to a sample, and the amount of light that gets through. We have a fixed volume of water. If we add drops of ink, one by one, and measure each time, we can understand exactly how many drops it takes to reduce the light that gets through by 10%, or by 50%, or by 95%. Imagine that we have done this, and we have a big book of values—or better yet, a computer database. Now I give you a glass of water that has an unknown amount of ink added to it, and I ask you how many drops of ink are in it. You can figure this out—you have your machine, you have experimental values showing you how much light will get through a glass with X or Y or Z drops added to it. All you have to do is put the glass into the machine and check the value against your table.

This is how spectrophotometry works. Nucleic acids absorb ultraviolet light; this increases at a fixed and known value based on their concentration in the sample. The more ultraviolet light is absorbed by a sample, the more nucleic acids we know are present.

The problem is, this technique is extremely error prone. Sure, your ink-water experiment may work perfectly to measure the ink concentration in a glass that has exactly the same volume of the exact same type of water as the one you used to make your reference values, but what if I gave you a glass that had less water to start? You’d get the wrong answer. What if the water evaporated slightly before I gave it to you? Your answer would be off a little. What if I added a few drops of milk, too, and you didn’t know about it? What if your machine had fallen a little out of calibration from when you first developed your list of reference values?

These types of variations are extremely common with spectrophotometry. Usually, very small volumes of liquid are used to measure concentrations. So errors in measuring out these volumes can have big effects. Also, water tends to stick together in small volumes, so it’s hard to consistently get the same volume to measure. Additionally, evaporation can wipe away a percentage of your sample pretty fast, and small impurities that you don’t know about can really throw off your measurement. I know this firsthand because I used to regularly measure DNA and RNA in the lab using a spectrophotometer. I had a lot of inconsistent measurements and problems that arose from this method. I took to doing every measurement two or three times, and occasionally re-measuring my samples every few weeks to be sure I had accurate values. Spectrophotometers are not really that reliable. They approximate to a degree that works in lab research, but that I’d never rely upon for dosing something into a human unless very great care was taken.

Back to what happened

As mentioned, two techniques were used. qRT-PCR gave a result that the Italian contractor reported, following their rigorous manufacturing process according to what’s called “Good Manufacturing Processes,” or GMP. Quality Assurance and Quality Control processes were in place, and the batch was logged and sent to Oxford.

There, spectrophotometry, a less reliable technique, showed a different value. For whatever reason, the quality assurance and control processes at that location did not raise the fact that spectrophotometry is less precise than qRT-PCR. Instead, the spectrophotometry results were trusted above the qRT-PCR results. An amendment to the rigorous trial protocol was requested, and a diluted dose was given. The Italian contractor’s numbers were not trusted; not by the researchers nor by anyone who oversaw the researchers.

As it turns out, the Italian contractor absolutely was right. Diluting the sample led to a half dose being given. The Oxford researchers, for reasons I cannot understand at all, trusted an imprecise technique over a precise one. I won’t try to speculate what their reasons were. They were wrong.

In this case, the bottom line is that an interesting dosing regimen for the vaccine was discovered, and this may be the reason that the AZ-Oxford vaccine gets approved at all. This accident has turned out for the best.

However, that is sheer dumb luck. Imagine instead that the wrong dosing caused some safety effect that hurt people. Imagine that it caused the vaccine to be less effective. These situations could just as easily have happened as this helpful accident, and both would have been terrible.

This is why the pharmaceutical industry has GMP. Nobody wants to be the reason that a batch of a drug comes to market at the wrong strength, or contaminated, or otherwise adulterated, and thus someone is hospitalized or dies. Nobody wants to be audited by FDA regulators and discovered to be playing fast and loose with these rules—because if that happens, you can go to jail. We take manufacturing really seriously, because a mistake can make the difference between a medicine or a poison. We take it so seriously that I am annually trained in GMP—even though I am not now, and never have been, involved directly in manufacturing.

As a result, I am not happy at all to hear that this is what happened in the AZ trial. It makes me wonder about the overall validity of the entire trial, because something like this was allowed to happen. It makes me doubt the quality of the work and the quality of both AstraZeneca’s organization as well as the research conducted at Oxford University.

These types of accidents are what make people lose faith in biomedical science. These are the stories that turn people into anti-vax conspiracy theorists. This should never have happened, and it is truly an embarrassment.

But at the end of the day, I can only be thankful that this mistake worked to the benefit of the vaccine trial participants, and may ultimately have helped us to identify a way to make this vaccine candidate effective.

What am I doing to cope with the pandemic? This:

Watching: The Goes Wrong Show

This weekend my wife and I were kind of low on new things to watch together and wanted to settle in for a relaxing night of something diverting. We found a British series called The Goes Wrong Show, which is a pretty classic British farce in its style. The premise is that a play is being put on, live on TV, and pretty much everything that can go wrong in a play, goes wrong. Props fail. Sets collapse. Designs are incorrect. Actors don’t show up. Scenes are set up wrong by the crew. Sound cues are missed. Lines are forgotten.

It’s full of amazing visual gags and deconstructs a lot of what makes theatre interesting by making everything go wrong. The wrongness is the performance, and it’s hilarious!

I recommend it. It’s apparently based on a British theater production called The Play that Goes Wrong, unsurprisingly, and I’m glad we decided to give it a try. You should too; what could go wrong?

I think it’s hard to predict how the story will evolve with the world of vaccination as well as COVID-19 specifically. Thankfully, I don’t intend to shut down this newsletter anytime soon, so you should hear about it here.

You might have some questions or comments! Send them in. As several folks have figured out, you can also email me if you have a comment that you don’t want to share with the whole group.

I’ve been contemplating changing the schedule of this newsletter a little bit; perhaps reducing it from daily to 3 times a week. I’m interested in your thoughts; feel free to comment publicly or send them my way privately.

Thank you to those who answered me about this already; I’m taking all of the feedback under consideration.

Join the conversation, and what you say will impact what I talk about in the next issue.

Also, let me know any other thoughts you might have about the newsletter. I’d like to make sure you’re getting what you want out of this.

Viral Transmissions

Discussion about this post