Comment by epmaybe | Hacker News Reader

Comment by epmaybe | Hacker News Reader

I'll take a stab at explaining this. Neuroscientists and biomedical researchers, feel free to chime in with corrections.

In biomedical researchers, the strength of evidence of your data correlates with how close it is to modeling the human body. So, in general, strength goes as follows:

Animal cell cultures < animal tissue < in vivo animal models <= human cell culture < human tissue < in vivo human trials (phases 1-3)

In this study, the researchers studied the ApoE4 subtype of the ApoE protein (involved in fat metabolism). The gene coding for this protein has been associated with greater risk of getting Alzheimer's disease then other subtypes, so it has been a target of study for a while now.

These researchers cultured brain cells that were differentiated from stem cell lines. Here's how they did it:

Skin from Alzheimer's patients and healthy patients (so apoe4 vs apoe3 containing cells) -> induced back to stem cell precursors -> differentiated into neurons

What they found was the following:

-neurons with apoe4 had production of Alzheimer's hallmarks (Tau phosphorylation, ABeta, neuron degeneration)

-neurons with apoe3 had less to none of these changes.

-editing the apoe4 gene to apoe3 prevented the changes from occurring.

-gettig rid of the apoe4 gene completely prevented the changes from occurring.

-giving a drug that changed the protein structure from apoe4 to apoe3 prevented the changes from occurring.

I'm shamelessly pulling all of this from the paper abstract, which you should find here: https://www.nature.com/articles/s41591-018-0004-z

As for what this means: it's exciting, and if these researchers can prove that the drug is safe in animals and humans they can probably proceed with trials on Alzheimer's patients with the apoe4 gene.

But that's the thing: this discovery, while cool, is relevant to only some of the patients dealing with Alzheimer's that have the ApoE4 gene. This isn't a unifying mechanism to how Alzheimer's occurs in all patients, as it doesn't tackle the problem of patients with apoe3 getting Alzheimer's. Nonetheless it is still relevant especially for early onset, which is more strongly associated with this gene.

Hopefully the results can be generalized to other experiments? Possibly the techniques they used?

That's a great summary, but I'll add one point. You could argue (and many people do) that strong genetic correlation (of one gene / mutation with disease) paired with in vitro / animal elucidation of causation is the second highest "level of evidence", second only to clinical trials. This is the core thesis of "precision medicine" and has radically improved success rates of science, drugs and startups

One of the most widely used methods of "target validation" (determining a molecules / genes causal role in disease) is the knockout mouse model. Basically delete the gene of interest in the mouse and see what happens. This is great for determining causality if a gene. However, mice are different than humans.

Genomics allows us to observe "human knockouts / knockins ". This provides very strong evidence that a gene is implicated in disease. Apoe4 is an example, but the canonical example is pcsk9

That's a decent summary of the paper, though it's important to note that the "structure corrector" is, at this point, more of a "compound"--there's a lot more work to be done before you'll be able to pick some up at CVS.

I also want to quibble a bit with your "strength ranking": it dramatically understates the value of in vivo experiments in animal models. Animals do differ from humans in some known--and presumably other unknown--ways, and this has certainly caused issues in the past. For example, the subjects in the CB28-SuperMAB trial all became seriously ill despite successful tests in rabbits and monkeys.

However, a major advantage of in vivo data is that it can give you a "whole organism" perspective on a possible treatment. This ensures that the therapy is targeting the actual condition (cognitive deficits) and not just a biomarker/side effect/epiphenomena. As an analogy, you want to actually fix a car's engine, instead of just turning the radio up until you can no longer hear the misfires. The Alzheimer's field has had issues with this before. ABeta is associated with Alzheimer's disease. While it's possible to reduce Abeta levels (e.g., with an antibody), this does not seem to have much effect on the mental symptoms.

I mentioned in a later comment that the researchers stated in the article that the small-molecule drug needs more work before human trials. I think you and another commenter are correct in that in vivo animal models should be higher up in that ranking, especially since I have seen human cell culture evidence not line up with in vivo experiments.

Good overview. Does this mean that the next steps to confirming/expanding these findings would be to reproduce this result in neuron cells, rather than skin cells which were differentiated to neuron cells? What are the next steps required to move further up the evidence chain?

I'm not entirely sure. It seems, from the paper, that mouse models do not share the same phenotype with humans, so animal tests may not be all that helpful, except for some safety data. I'm not sure they will be able to culture actual human neurons from tissue, as that was only recently accomplished back in 2017.

From the paper itself, this is what the researchers say: "These findings warrant further development of ApoE4 structure correctors and, ultimately, testing in clinical trials." So it looks like they want to move forward clinically.

Sounds like a great time to start a biotech company. Though there will be some competition. Am aware of at least two companies in this space ;)

I'm not so sure it's a great time at all. Most alzheimer's drugs have failed dramatically. vTv Therapeutics is the most recent I think. They had a Phase 3 drug trial end early due to futility, and their stock tanked hard. In one day I think it lost 70% of its value.

As far as I'm aware the majority of failing biotech companies have very broadly targeted Amyloid beta (apparently because they thought it was a good risk tradeoff). A narrow focus on APOE4 could be a wiser strategy. I could even see gene therapy playing a role.

There are a few well funded startups targeting apoe4 and other genetic markers of neurodegen disease. Exciting field

Alzheimer's drugs have not been successful, but in general it is the best time ever to start a biotech company. Returns in biotech VC the last 5 years are better than tech, more IPO and big M&A (three $8-12B startup acq in last 4-6 months) even with 1/5 the funding of software, tons of great science out there, massive amounts of capital ($4B venture funding to startups in Q1)

Reply on news.ycombinator.com