What Cassava Sciences should do now

Apparently someone important didn’t like the way Cassava Sciences analyzed their data and their stock tanked again today..  Unfortunately all of this seems to be behind a paywall, and the someone important isn’t named.  I’d love a link if any reader knows of one — just put it in as a  comment below.

I’m not important, but I thought Cassava’s results were quite impressive.  They had enough cases and enough time for the results to be statistically significant

For one thing,  Cassava dealt with severely impaired people (see below) who would be expected to show greater neuronal dropout, senile plaques and neurofibrillary tangles, than recently diagnosed patients.   Neuronal loss is not reversible in man, despite hoards of papers showing the opposite in animals.

Since everything turns on ADAS-CoG, here is a link to a complete description along with some discussion — https://www.ncbi.nlm.nih.gov/pmc/articles/PMC5929311/

On a slide from Cassava’s presentation yesterday the ADAS-CoG average of the 50 patients on entry 9 months ago was 16.6.  With a perfect score of 70, it’s clear that these people were significantly impaired (please look at the test items to see how simple the tasks in ADAS-CoG actually are).    So an improvement of 3 points at 9 months  is significant, particularly since a drop of 5 points is expected each year — yes I’ve seen plenty of Alzheimer patients with ADAS-CoG scores of zero or close to it.

So an increase of 3 points in this group is about a16% improvement.

Here’s what Cassava should do now.  Their data should be re-examined as follows.  Split the ADAS-CoG scores into 3 groups: highest middle and lowest. Quartiles are usually used, but I don’t think 50 patients is enough to do this.  Then examine the median improvement in each of the three.  I’d use median rather than average as with small numbers in each group, a single outlier can seriously distort things — think of the survival of Stephen Hawking in a group of 12 ALS patients.

If the patients with the highest ADAS-CoG scores have the highest median improvement, there is no reason mildly impaired individuals should have a less than 16% improvement in their scores.  This means that a person with ADAS-CoG of 60 should achieve a perfect score of 70,  e.g. return to normal.

This would be incredibly useful for early Alzheimer’s disease.

There is a precedent for this.  Again it’s Parkinson’s disease.

As I mentioned in an earlier post, I was one of the first neurologists in the USA to use L-DOPA for Parkinsonism.  All patients improved, and I actually saw one or two wheelchair bound Parkinsonians walk again (without going to Lourdes).  They were far from normal, but ever so much better.

However,  treated mildly impaired Parkinsonians became indistinguishable from normal, to the extent that I wondered if I’d misdiagnosed them. These results were typical.   For a time, in the early 70s neurologists thought that we’d actually cured the disease.  It was a very heady time.  We were masters of the neurologic universe — schizophrenia was too much dopamine, Parkinsonism not enough. Bring on the next neurotransmitter, bring on the next disease.

We hadn’t cured anything of course, and the underlying loss of dopamine neurons in the substantia nigra continued.  Reality intruded for me with one such extremely normal appearing individual I’d diagnosed with Parkinsonism a few years earlier. He needed surgery, meaning that he couldn’t take anything by mouth for a while.  L-DOPA could only be given orally, and he looked quite Parkinsonian in a day or two.

If reanalysis of the existing data shows what I hope, Cassava Sciences should start another study in Alzheimer patients with ADAS-CoG scores of over 50.  If I’m right the results should be spectacular (and lead to immediate approval of the drug).

A little blue sky.  Sumafilam will then come to be known as intellectual Viagra, as all sorts of oldsters (such as yrs trly) will try to get it Alzheimer’s or no Alzheimer’s.

Amyloid

Amyloid goes way back, and scientific writing about has had various zigs and zags starting with Virchow (1821 – 1902) who named it because he thought it was made out of sugar.  For a long time it was defined by the way it looks under the microscope being birefringent when stained with Congo red (which came out 100 years ago,  long before we knew much about protein structure (Pauling didn’t propose the alpha helix until 1951).

Birefringence itself is interesting.  Light moves at different speeds as it moves through materials — which is why your legs look funny when you stand in shallow water.  This is called the refractive index.   Birefringent materials have two different refractive indexes depending on the orientation (polarization) of the light looking at it.  So when amyloid present in fixed tissue on a slide, you see beautiful colors — for pictures and much more please see — https://onlinelibrary.wiley.com/doi/full/10.1111/iep.12330

So there has been a lot of confusion about what amyloid is and isn’t and even the exemplary Derek Lowe got it wrong in a recent post of his

“It needs to be noted that tau is not amyloid, and the TauRx’s drug has failed in the clinic in an Alzheimer’s trial.”

But Tau fibrils are amyloid, and prions are amyloid and the Lewy body is made of amyloid too, if you subscribe to the current definition of amyloid as something that shows a cross-beta pattern on Xray diffraction — https://www.researchgate.net/figure/Schematic-representation-of-the-cross-b-X-ray-diffraction-pattern-typically-produced-by_fig3_293484229.

Take about 500 dishes and stack them on top of each other and that’s the rough dimension of an amyloid fibril.  Each dish is made of a beta sheet.  Xray diffraction was used to characterize amyloid because no one could dissolve it, and study it by Xray crystallography.

Now that we have cryoEM, we’re learning much more.  I have , gone on and on about how miraculous it is that proteins have one or a few shapes — https://luysii.wordpress.com/2010/08/04/why-should-a-protein-have-just-one-shape-or-any-shape-for-that-matter/

So prion strains and the fact that alpha-synuclein amyloid aggregates produce different clinical disease despite having the same amino acid sequence was no surprise to me.

But it gets better.  The prion strains etc. etc may not be due to different structure but different decorations of the same structure by protein modifications.

The same is true for the different diseases that tau amyloid fibrils produce — never mind that they’ve been called neurofibrillary tangles and not amyloid, they have the same cross-beta structure.

A great paper [ Cell vol. 180 pp. 633 – 644 ’20 ] shows how different the tau protofilament from one disease (corticobasal degeneration) is from another (Alzheimer’s disease).  Figure three shows the side chain as it meanders around forming one ‘dish’ in the model above.  The meander is quite different in corticobasal degeneration (CBD) and Alzheimers.

It’s all the stuff tacked on. Tau is modified on its lysines (some 15% of all amino acids in the beta sheet forming part) by ubiquitination, acetylation and trimethylation, and by phosphorylation on serine.

Figure 3 is worth more of a look because it shows how different the post-translational modifications are of the same amino acid stretch of the tau protein in the Alzheimer’s and CBD.  Why has this not been seen before — because the amyloid was treated with pronase and other enzymes to get better pictures on cryoEM.  Isn’t that amazing.  Someone is probably looking to see if this explains prion strains.

The question arises — is the chain structure in space different because of the modifications, or are the modifications there because the chain structure in space is different.  This could go either way we have 500+ enzymes (protein kinases) putting phosphate on serine and/or threonine, each looking at a particular protein conformation around the two so they don’t phosphorylate everything — ditto for the enzymes that put ubiquitin on proteins.

Fascinating times.  Imagine something as simple as pronase hiding all this beautiful structure.

 

 

Kuru continues to inform

Neurologists of my generation were fascinated with Kuru, a disease of the (formerly) obscure Fore tribe of New Guinea. Who would have thought they would tell us a good deal about protein structure and dynamics?

It is a fascinating story including a Nobelist pedophile (Carleton Gajdusek) https://en.wikipedia.org/wiki/Daniel_Carleton_Gajdusek and another (future) Nobelist who I probably ate lunch with when we were both medical students in the same Medical Fraternity but don’t remember –https://en.wikipedia.org/wiki/Stanley_B._Prusiner

Kuru is a horrible neurodegeneration starting with incoordination, followed by dementia and death in a vegetative state in 4 months to 2 years. For the cognoscenti — the pathology is neuronal loss, astrocytosis, microglial proliferation, loss of myelinated fibers and the kuru plaque.

It is estimated that it killed 3,000 members of the 30,000 member tribe. The mode of transmission turned out to be ritual cannibalism (flesh of the dead was eaten by the living before burial). Once that stopped the disease disappeared.

It is a prion disease, e.g. a disease due to a protein (called PrP) we all have but in an abnormal conformation (called PrpSc). Like Vonnegut’s Ice-9 (https://en.wikipedia.org/wiki/Ice-nine) PrPSc causes normal PrP to assume its conformation, causing it to aggregate and form an insoluble mess. We still don’t know the structure of PrPSc (because it’s an insoluble mess). Even now, “the detailed structure of PrPSc remains unresolved” but ‘it seems to be’ very similar to amyloid [ Nature vol. 512 pp. 32 – 34 ’14]. Not only that, but we don’t know what PrP actually does, and mice with no PrP at all are normal [ Nature vol. 365 p. 386 ’93 ]. For much more on prions please see https://luysii.wordpress.com/2014/03/30/a-primer-on-prions/

Prusiner’s idea that prion diseases were due to a protein, with no DNA or RNA involved met with incredible resistance for several reasons. This was the era of DNA makes RNA makes protein, and Prisoner was asking us to believe that a protein could essentially reproduce without any DNA or RNA. This was also the era in which X-ray crystallography was showing us ‘the’ structure of proteins, and it was hard to accept that there could be more than one.

There are several other prion diseases of humans (all horrible) — mad cow disease, Jakob Creutzfeldt disease, Familial fatal insomnia, etc. etc. and others in animals. All involve the same protein PrP.

One can take brain homogenates for an infected animal, inoculate it into a normal animal and watch progressive formation of PrPSc insoluble aggregates and neurodegeneration. A huge research effort has gone into purifying these homogenates so the possibility of any DNA or RNA causing the problem is very low. There still is one hold out — Laura Manuelidis who would have been a classmate had I gone to Yale Med instead of Penn. n

Enter [ Nature vol. 522 pp. 423 – 424, 478 – 481 ’15 ] which continued to study the genetic makeup of the Fore tribe. In an excellent example of natural selection in action, a new variant of PrP appeared in the tribe. At amino acid #127, valine is substituted for glycine (G127V is how this sort of thing is notated). Don’t be confused if you’re somewhat conversant with the literature — we all have a polymorphism at amino acid #129 of the protein, which can be either methionine or valine. It is thought that people with one methionine and one valine on each gene at 129 were somewhat protected against prion disease (presumably it affects the binding between identical prion proteins required for conformational change to PrPSc.

What’s the big deal? Well, this work shows that mice with one copy of V127 are protected against kuru prions. The really impressive point is that the mice are also protected against variant Creutzfedlt disease prions. Mice with two copies of V127 are completely protected against all forms of human prion disease . So something about V/V at #127 prevents the conformation change to PrPSc. We don’t know what it is as the normal structure of the variant hasn’t been determined as yet.

This is quite exciting, and work is certain to go on to find short peptide sequences mimicking the conformation around #127 to see if they’ll also work against prion diseases.

This won’t be a huge advance for the population at large, as prion diseases, as classically known, are quite rare. Creutzfeldt disease hits 1 person out of a million each year.

There are far bigger fish to fry however. There is some evidence that the neurofibrillary tangles (tau protein) of Alzheimer’s disease and the Lewy bodies (alpha-Synuclein) of Parkinsonism, spread cell to cell by a ‘prionlike’ mechanism [ Nature vol.485 pp. 651 – 655 ’12, Neuron vol. 73 pp. 1204 – 1215 ’12 ]. Could this sort of thing be blocked by a small amino acid change in one of them (or better a small drug like peptide?).

Stay tuned.