(8/8) I know this issue has become extremely politicized. We all need take a deep breath and try to look at the possible science behind what is going on, so we do not make an enormous mistake.

(7/8) Another possible explanation of the Lancet harm result is observational bias despite extensive attempts to control for it. Controlling for LDH levels and FiO2 levels (not O2 sats) could have helped us understand this bias if it happened.

(6/8) I am also a bit surprised that while cardiac arrhythmias were mentioned in the @Lancet article, QT measurements were not. Does that concern you?

(5/8) We still don't know if #HCQ works from randomized trials. We give it extensively to pregnant women ( https://t.co/XyTVXzRvDq  & https://t.co/Pgc0h9aJs8) and it's been used in lupus with few complications for 70 years, so my guess is that any true harm is limited at best.

(4/8) HCQ and CQ are the only FDA-approved autophagy inhibitors. Were subjects in the Lancet study stratified by LDH level? It seems to predict for severity.

(3/8) Everything in the study reported in the Lancet can likely modulate autophagy in the pneumocyte (ACE inhibition, statins), and famotidine can do so as well. Were these controlled for?

(2/8) The genes expressed in COVID-19 lung are in a large part related to autophagy ( https://t.co/Ztiovf14j5), and increased neovascularization is a sign of autophagy.

Thread on the #hydroxychloroquine study for Dr. Topol: (1/8) Assume for a minute that what is going on is that the virus is hijacking autophagy pathways in cells to live and to thrive. We can discuss the support for this hypothesis at length but look at the @NEJM yesterday.

Sorry, not a twitter expert. Someone told me you have to read this from the bottom up!

I'm not a virologist or an immunologist. I'm a medical oncologist with a molecular biology background and training. Mike is a world class dendritic cell immunologist.

Now I have put this out there not as the absolute truth. But a possible framework for what is going on. To be tested for truth or not. That is what theories are supposed to do. But it can be tested.

Most of the mutations in SARS-CoV-2 now seen appear to predict for attenuation. So this is kind of supportive of the hypothesis. Even in D614G the Q57H is appearing more in that class over time. Suggesting that D614G is also attenuating.

Of all the mutations in SARS-CoV-2, only D614G appears so far to predict for virulence. We never know, there could be more. But viral evolutionary theory predicts that viruses when introduced into a new host trend towards attenuation.

40% of the protein-protein interactions in the Nature paper affect endosomal processing and transport. Thus a rich area for attenuation mutations of the viral synapse.

If virus is spit out at the wrong time, there will be lots of virus and then the immune system (now trying to recover) hits back big. Hence cytokine storm and the rest. Mutations that affect endosomal processing should affect virulence of the virus

By infection of the dendritic cell (DC), the virus just sits there (like SARS-CoV). The immune system to gets messed up a bit (lose T cells, etc) since the DC is the master controller. Then spits it out in something called the "viral synapse." HIV does this as well.

In a paper before Journal of Translational Medicine right now, we try to make the case that Q57H of ORF 3a is right in the middle of an ion channel that affects viral expulsion from the phagolysosome of the dendritic cell. ORF3a deletions of SARS-CoV do this in culture.

When you look at COVID Glue, D614G is the top mutation found right now. But others are catching up. Like Q57H.

Sorry about the typo there. I'm kind of new to twitter. The Nature paper, if looked at carefully, suggests lots of targets for stable SARS-CoV-2 attenuation through mutation.

There are likely more. As well as lots of other mutations in other proteins of SARS-CoV-2 A SARS-CoV-2 protein interaction map reveals targets for drug repurposingthat predict for attenuation. This Nature paper last week was really interesting. https://t.co/dayXGqmioJ