The bradykinin hypothesis

September 7, 2020 5:49 AM

"Earlier this summer, the Summit supercomputer at Oak Ridge National Lab in Tennessee set about crunching data on more than 40,000 genes from 17,000 genetic samples in an effort to better understand Covid-19... When Summit was done, researchers analyzed the results. It was, in the words of Dr. Daniel Jacobson, lead researcher and chief scientist for computational systems biology at Oak Ridge, a “eureka moment.” The computer had revealed a new theory about how Covid-19 impacts the body: the bradykinin hypothesis. The hypothesis provides a model that explains many aspects of Covid-19, including some of its most bizarre symptoms."
posted by thoughtful_ravioli (32 comments total)

Fascinating.
posted by glonous keming at 6:18 AM on September 7, 2020 [1 favorite]

Wow. That really is fascinating. Thank you.
posted by davidmsc at 6:22 AM on September 7, 2020 [1 favorite]

One thing I did not see in the article: What is the relationship of this mechanism to the overall infection rate? There are some noted correlations to varying mortality rates among specific demographics, but what accounts for how many people get it and show no or minimal symptoms? I'd like to understand whether or not this theory addresses that aspect of the disease. I suspect my question has more to do with the reporting, rather than the actual research.
But otherwise, wow! I like being reminded how much we depend on extremely smart people to solve our problems.
posted by Carmody'sPrize at 6:45 AM on September 7, 2020 [1 favorite]

Huh. This is super interesting. Thanks for posting.
posted by lazaruslong at 6:47 AM on September 7, 2020 [1 favorite]

Oh good! More big data crunching without experimental work! This work may or may not be good......

But straight away, I can see why it was published in eLife...... in a more-examined journal, if the first line of the abstract was the sentence

"Neither the disease mechanism nor treatments for COVID-19 are currently known."

you would be slapped down.

Even in July when this was published, PLENTY was known about the mechanism of covid-19 infection and the disease, and plenty of somewhat-effective treatments were utilized.

At least the medium article, throwing out a grab-bag of compounds to try , suggests "All of these potential treatments are speculative, of course, and would need to be studied in a rigorous, controlled environment before their effectiveness could be determined and they could be used more broadly."
posted by lalochezia at 6:55 AM on September 7, 2020 [13 favorites]

If you want to see what's happening in the fast-moving field with this hypothesis (and cousins of it), look at citations to a more clinically oriented paper (also published in eLife).
posted by lalochezia at 7:03 AM on September 7, 2020 [8 favorites]

Re. demographics/age, some researchers think that ACE2 receptors increase with age, which is why it affects older people more severely. Link. I'm 100% not expert in this so the link will have to the explaining for me (if it indeed does that).
posted by carter at 7:05 AM on September 7, 2020 [2 favorites]

One thing that confused me, as a layperson, when I first read this was why this hypothesis hadn't already been confirmed or disproved. A bit of searching appears to indicate that bradykinin is norrnaly very short-lived: "[t]he half-life in blood is about 17 s but less than 4 s in various vascular beds."

Perhaps this short duration is a source of testing difficulty. In any case, I wonder how it might be possible to confirm or disprove this thesis.
posted by washburn at 7:06 AM on September 7, 2020 [3 favorites]

Please, can someone contextualize this information for readers?

This is a Medium blog post by an author who is "Co-Founder & CEO of Gado Images. I write, speak and consult about AI, privacy, photography, tech and the Bay Area." (Medium does not fact-check nor in any way review articles on their platform.)

I don't claim that the posted information is incorrect or misleading; I do claim that COVID is an intensely hot topic right now, this development is a bold one, and the burden rests with people posting information to provide more context than a single link to a blog post - such as more information on the author's COVID credentials, or even better a well-established journalism source with fact-checkers and editors.
posted by splitpeasoup at 7:22 AM on September 7, 2020 [43 favorites]

Here's some smart people with perhaps more closely relevant credentials who seem to think the hypothesis has merit, and is worth further study:

Is a Bradykinin Storm Brewing in COVID-19?

...Frank van de Veerdonk, an infectious disease specialist at the Radboud University Medical Center in the Netherlands, was heading down the same molecular pathway in mid-March. He had noticed two features in COVID-19 patients in his clinic—fluid in the lungs and inflammation. Because other labs had pegged angiotensin-converting enzyme-2 (ACE2), a key enzyme in the RAS, as being the SARS-CoV-2 receptor, and because he knew that ACE2 regulates the kinin system, van de Veerdonk began connecting the dots. In April, he and his group hypothesized that a dysregulated bradykinin system was leading to leaky blood vessels in the lungs and perhaps causing excess fluid to build up.

...

Josef Penninger, director of the Life Sciences Institute at the University of British Columbia in Vancouver who discovered that ACE2 is the essential in vivo receptor for SARS, tells The Scientist that he is convinced that bradykinin plays a role in COVID-19 pathogenesis. “It does make a lot of sense.”

He adds that Jacobson’s study lends support to the hypothesis, but further confirmation is needed. “Gene expression signatures don’t tell us the whole story. I think it is very important to actually measure the proteins.”

posted by lazaruslong at 7:38 AM on September 7, 2020 [9 favorites]

To be fair,

"Neither the disease mechanism nor treatments for COVID-19 are currently known."

is math/computer science speak, where "known" is used kind of like a big, blunt hammer to mean a more or less complete and certain understanding, while in the experimental sciences, you would have a more nuanced "we know this with x% confidence, this other detail with y% confidence, and don't have sufficient data yet to draw conclusions about this third detail." So this may be a cultural difference. Although it's also entirely possible that it's ignorance of the details of another sub-field, or over-confidence in their own approach on the part of the authors.

Modeling studies like this have a particular role in science: they can suggest avenues for research that haven't been thought of yet, or potential connections between currently known research results that haven't been noticed by people yet. I shared another link about this study last week with a small select group of friends who I knew would understand what the study was saying and what it was not saying, but did not share it more generally due to the potential for confusion by folks who aren't as familiar with the processes or methodologies of scientific research.
posted by eviemath at 8:09 AM on September 7, 2020 [13 favorites]

this development is a bold one, and the burden rests with people posting information to provide more context than a single link to a blog post

Skepticism of Medium posts on science is warranted, but there have been positive results using icatibant, a kinin pathway inhibitor (cite).
posted by They sucked his brains out! at 8:22 AM on September 7, 2020 [5 favorites]

I'm really uncomfortable with this sort of science influence through popular media. We saw it abused early on in Covid by the scandalous Stanford / Santa Clara antibody study. (IMHO John Ionannidis and Jay Bhattacharya should never be allowed to do science again and everyone associated should have a giant asterisk next to their name in anything ever published in the future.) Every week there's some breathless new article, often well meaning, and it's really confusing for laypeople to sort out.

The bradykinin hypothesis sounds like an interesting hypothesis and the way it was formed via data analysis is sort of novel and interesting. But I'm 100% unequipped to evaluate it, much less put their theory into action. And I have a Master's degree in science and many years experience working cross-disciplinary between computer science and other fields. Maybe it'd be better to let the science cook for awhile? I'm not saying it's bad for citizens to have access to science in its early raw form, but when it comes with a PR department or an agenda (even self promotion) it's dangerous.
posted by Nelson at 8:27 AM on September 7, 2020 [18 favorites]

IMHO John [Ioannidis] and Jay Bhattacharya should never be allowed to do science again and everyone associated should have a giant asterisk next to their name in anything ever published in the future

Wow, I had completely missed Ioannidis's role in this fuckup. His continuing to push poorly run, non-replicated studies is pretty rich coming from Mr. Replicability Crisis.
posted by jedicus at 8:39 AM on September 7, 2020 [4 favorites]

(Some more info on icatibant.)
posted by They sucked his brains out! at 8:54 AM on September 7, 2020

Yes, we don't want bad science. But what's happening right now is proposing a hypothesis that can guide current and future work. That's a pretty reasonable thing for a popular science article, even by a non-expert.
posted by medusa at 8:55 AM on September 7, 2020 [11 favorites]

I suggest to no trust any of these Dunning-Kruger bro's any further than you can throw them. However this one is doing better than many by at least liberally linking to more professional journalism and the original study.

I've been casually reading about ACE2 and vitamin D since early March, and have been a bit surprised to not see it picked up more. Anyway, nice to see that bit getting more support and ink, and unlike many new or exotic drugs, vit D cannot hurt you in sane doses.
posted by SaltySalticid at 9:01 AM on September 7, 2020 [1 favorite]

Extremely endorse splitpeasoup's comment above, which was my exact reaction. The information in this article may be totally true, who knows! But the state of science reporting in the Anglophone media is so utterly, utterly abysmal (and I've been burned by it so many times) that at this point I assume that if it's written by a journalist (or in this case, barf, a CEO) and not someone with actual scientific experience in a relevant area, it's almost certainly 90% made up. Again, this may be unfair, but given the serious crimes against public understanding of science committed by even reputable publications recently, it's where I'm at.
posted by demonic winged headgear at 10:47 AM on September 7, 2020 [2 favorites]

I think a lot of people outside of the biomed research world don't have a good sense of how many promising and sensible hypotheses don't pan out. It's not the exception, it's the rule. Molecular biology requires something like an inverse Occam's razor: all other things being equal, the simplest explanation is usually the wrong one. You need a certain degree of humility to appreciate the complexity of the problem you're investigating and the limitations of your ability to understand it. That doesn't mean this hypothesis is wrong, it just warrants a lot of skepticism.

There's a worrisome trend of laypeople on the internet latching onto certain hypotheses because they're simple, explicable, and satisfying. You will still find people arguing about the effectiveness of hydroxychloroquine, and not just MAGA wingnuts. These are people with some degree of scientific literacy but not enough depth or experience to understand why they're wrong. They make semi-scientific arguments about why it works, or why studies haven't shown a significant benefit, and trying to debate them is like playing tennis with a wall. I don't know how to solve a problem like that, but I feel like popsci articles like this one aren't helping. For hot-button topics like COVID, you need a measured, conservative approach in terms of how you engage the laypublic, otherwise you risk creating cult science.
posted by dephlogisticated at 10:58 AM on September 7, 2020 [14 favorites]

You will still find people arguing about the effectiveness of hydroxychloroquine, and not just MAGA wingnuts.

Yep. I sincerely wish that I did not know this, but I do, way more personally than I would like. God forbid the close relative responsible get sick, because they’ll chug that shit. Also, the same person is a total Ionnadis fanboy. Just ugh.
posted by dragstroke at 11:08 AM on September 7, 2020 [1 favorite]

HAE therapies may find a new use.
posted by metasunday at 11:15 AM on September 7, 2020 [1 favorite]

TIL that the fastest super computer in the world is way, way faster than the second-fastest one -- and it's ARM-based. So that's something anyway.
Who knew?
posted by The Bellman at 11:20 AM on September 7, 2020 [1 favorite]

It's a bit weird that the article frames this as a new hypothesis that this tech guy discovered. One click on one of the supporting links in the article takes you too an actual scientists preprint from April 1, before this guy had his results, and also suggests the link with bradykinins.
posted by lazaruslong at 11:33 AM on September 7, 2020 [7 favorites]

I would like to know more about the supercomputer, like were they using all 4608 * 6 Nvidia Tesla GPUs, or just a rack or two so that IBM could write the press release? Looks like some R and a software package called H-MAGMA was involved.
posted by RobotVoodooPower at 11:40 AM on September 7, 2020 [1 favorite]

Slight derail: did anyone else find his analogy of a burglar who opens all your doors and windows so other burglars can come in almost nothing like the mechanisms he described? If he's going with COVID as a burglar, based on the rest of the article it would be one that breaks all your stuff, or somehow rewires everything so it'll break itself... not a good analogy, is what I'm sayin'
posted by Saxon Kane at 11:43 AM on September 7, 2020 [4 favorites]

Dude's lack of burgling chops clearly noted
posted by some loser at 12:29 PM on September 7, 2020 [3 favorites]

This is definitely interesting, though, with all the caveats noted above. It feels like a useful follow-up to another Medium article from the end of May writing up findings that COVID-19 was likely more of a vascular disease. Sure, it might not be conclusive by any means, but it's interesting. I don't actually love that I'm getting so much of my info about the virus from articles on Medium, but it's absolutely the case that reading things on there helped me prepare at times when those around me downplayed the virus's potential severity. I remember reading the article I just linked on there back in mid-March, at a point when so many still weren't taking it seriously in NYC. The fact that people on Medium felt the need to write articles like this at the beginning of March speaks to the way the virus was downplayed by so many around us. I guess what I'm saying is that I'm open to receiving information from intelligent, informed people wherever they might publish, even though of course I will take them with due skepticism if they're not published in a journal or by a journalism outlet.

I guess where I come out on the state of science reporting is that yes, absolutely, often the media does a terrible job of reporting on actual science in a coherent way. Having been a journalist and knowing many journalists and J-school grads, I understand on a deep level exactly how science writing, editing, and fact-checking can and often do fail. But that actually makes me more willing to consider someone's writing on Medium as a potentially useful source of information, knowing most journalists aren't operating with any more resources or deep understanding of science than someone on Medium might be, and in many cases, actual subject-matter experts go directly to places like Medium to air ideas that might not be journal-ready or easily parsed by journalists but that deserve discussion. Why go to a journalist, who might mangle whatever you're trying to explain and not even necessarily garner it a wider audience than you could find yourself, by writing it up on Medium and posting it on social media? I have sympathy for anyone who decides to just write about it themselves.

Is the free press a crucial part of our society, a critical check on power? Yeah, totally. Is the free press particularly great at science writing and comprehension? Not really, excepting publications whose expertise specifically lies in that area.
posted by limeonaire at 2:49 PM on September 7, 2020 [3 favorites]

Part of the problem of journalistic coverage of science is the tendency to report a study out of context of the overall literature. For example, correlational studies often provide useful guides to further, more expensive studies that can examine causation, but on their own they aren't enough to say A leads to B. Nevertheless, the reporting often elides the difference with words like "linked", which don't convey that meaning to the reader.

Anyway, here is a direct link to the paper the article in the post describes describes: Oak Ridge Laboratory

Here is a medcram video that describes the study and places it into context with other COVID research. Medcram link

The part about the Oak Ridge paper starts at 11:45, after a section on cause of death reporting and excess deaths over yearly average, which is interesting but not relevant to the article in the post.
posted by eagles123 at 9:24 PM on September 7, 2020 [1 favorite]

I'm not sure why using a supercomputer to come up with a previously published hypothesis makes the news. I guess better PR connections and the OMG SO MANY COMPUTERS angle?

I think the bradykinin hypothesis is plausible but the eLife paper didn't add anything that hadn't already been postulated.
posted by benzenedream at 1:02 AM on September 8, 2020

SARS-CoV-2 was Unexpectedly Deadlier than Push-scooters: Could Hydroxychloroquine be the Unique Solution?

1. INTRODUCTION

As the number of push-scooters has been rising in France, so has the number of push-scooters accidents. Some of these accidents have proven to be deadly and previous YouTubeTM and Dropbox© studies have warned against the deadly potential of push-scooters [1]. For a comparison, only three Chinese people had died from the novel coronavirus SARS-CoV-2 at the end of 2019 [2]. It is therefore important to reflect on the use of push-scooters through an accurate and ethical cost-benefit analysis...

Given the obvious similarities between COVID-19 and push-scooters accidents (i.e. both can have deadly outcomes in which the patient might even die), it seemed natural to expand the use of HCQ + AZT to push-scooters accidents (henceforth: PSA), even if no in-vitro study ever found an effect of HCQ on PSA.
posted by They sucked his brains out! at 7:04 AM on September 8, 2020 [1 favorite]

A little more backstory on that journal article.
posted by They sucked his brains out! at 7:08 AM on September 8, 2020

Derek Lowe's In the Pipeline blog has a new post on the bradykinin hypothesis. His overall take is that it seems like a strong hypothesis but needs clinical evaluation:

It’s a very appealing theory, but it’s also important to remember that a lot of very appealing theories in this business turn out not to be true. Or not completely true. Or not true in the ways that were originally thought. The way to find out is through clinical testing. To be honest, I’m worried that this proposal is almost too neat and form-fitting; rarely do you get something that falls together this well. It’s also quite possible that you could come up with a reasonable set of literature references and previous reports that cast many of these connections into doubt – the medical literature is large, and you can find support for a lot of things if you’re putting together a brief for the prosecution. But overall, I find this work pretty plausible.

Luckily there are several existing FDA-approved drugs that could be used to directly test the hypothesis.
posted by jedicus at 7:48 AM on September 8, 2020 [5 favorites]