These are my notes from week 9 of Harvard’s Genetics 228: Genetics in Medicine from Bench to Bedside course, held on April 3, 2015. Lectures by Ali Atri and Rudy Tanzi.

Ali Atri: clinical aspects of early onset Alzheimer disease (EOAD)

The first patient who Alois Alzheimer saw in 1902, named August D, was actually 51 years old - she had early onset Alzheimer disease. Historically, individuals under 50 have often been specifically excluded from clinical trials in AD, though this is starting to change now. Clinically speaking, even though Alzheimer disease is incurable at present, early diagnosis is critical, most importantly for education. Getting patients educated about how to manage their AD makes a big difference. For instance, they may need to make financial plans or changes to their career. And making changes to reduce stress, get better sleep, and so on can also improve cognition in some patients.

Risk factors for AD include:

  • Age
  • Sex (females have higher risk)
  • Severe head injury
  • Altered cerebral perfusion
  • Environmental stressors
  • Cerebral amyloidosis
  • Family history
  • Genetics - highly penetrant variants in APP, PSEN1 and PSEN2; strong risk factor in APOE, smaller risk factors in CLU, CR1, PICALM, SORL1, TOMM40.

Vascular disease and Alzheimer disease often co-exist.

Two classes of medications are specifically indicated in Alzheimer disease:

  • Cholinesterase inhibitors (ChEls; examples include donepezil, rivastigimine, and galantamine) which inhibit breakdown of acetylcholine in synapses. (I am not sure why this mechanism is relevant to AD).
  • Memantine, which is an NMDA antagonist. It is still controversial how exactly this helps with Alzheimer disease - the prevailing theory is by supressing excitotoxicity.

In addition, AD patients are often prescribed:

  • Antidepressants
  • Antipsychotics
  • Vitamin E, at very high doses, for which there are actually good data that it slows functional decline (by about 20% for year) [Dysken 2014]

Collectively, these therapies make a real difference in the rate of decline in AD patients. Large meta-analyses have shown that combination therapy is better than monotherapy is better than placebo at slowing decline on various functional metrics and delaying the time to entering a nursing home [Atri 2008].

Still, education on management of AD and having good caregivers around may make an even larger difference in outcomes [Atri 2011]. Meanwhile, one needs to look after the caregivers themselves, as a large fraction will suffer from depression.

Rudy Tanzi

The “three pillars” of AD pathology are:

  • Aβ pathology
  • Neurofibrillary tangles
  • Inflammation

For three decades, the big question in the AD field has been in what directly the causality flows between these three things.

The amyloid hypothesis of Alzheimer disease, first formulated by George Glenner [Glenner 1983], holds that Aβ is causal in Alzheimer disease.

The first pathogenic genetic mutation identified in APP was actually linked to vascular amyloidosis rather than AD [[Frangione 1990]]. About 50% of cases of AD that strike under age 60, and nearly 100% under age 50, have a mutation in APP, PSEN1 or PSEN2. Of these, PSEN1 mutations are the most prevalent. The gamma secretase complex, which contains PSEN1 and PSEN2, actually “bites” APP at the residue that would correspond to Aβ50, and then “chews” it back to residue 43 or all the way to 39. The PSEN1 and PSEN2 mutations reduce the “chewing” efficiency, thus causing longer Aβ peptides to be produced (higher Aβ42/Aβ39 ratio) - though the number of Aβ peptides produced actually goes down, in contrast to what the APP Swedish mutation does.

The “amyloid cascade hypothesis” holds that Aβ cerebral amyloidosis causes synaptic and neuronal injury and inflammation, and that tau neurofibrillary tangles arise secondary to all of this damage. However, Dr. Tanzi now believes that the amyloid cascade hypothesis is wrong, in that his recent work with 3D human neuronal culture models of AD [Choi & Kim 2014] indicates that the presence of amyloid, without extensive neuronal damage, is sufficient to induce formation of tau neurofibrillary tangles. I have previously blogged about that paper here. Dr. Tanzi says the important points are that Aβ can directly cause tau tangles without causing neuronal damage first, and the enzyme GSK3β is required for forming tau tangles.

Dr. Tanzi remains optimistic about the possibility of gamma secretase inhibitors as therapies for Alzheimer disease. The animal data [Kounnas 2010] indicate that these inhibitors really do reduce Aβ42 production and Aβ plaque deposition, and Lilly’s drug candidate failed on safety rather than efficacy. Now that more selective gamma secretase “modulators” are available, these may prove to be safe and effective.

That said, the failure of anti-Aβ antibodies such as Lilly’s solanezumab is probably just due to treating patients too late - Aβ plaques form 15 years before dementia onset, but all of the unsuccessful trials sought to treat symptomatic patients. And the same is also probably true of gamma secretase modulators and beta secretase inhibitors - they’ll need to be administered much earlier to be effective. Appropriately, Lilly is now running an A4 trial to see if solanezumab can prevent [phenoconversion of individuals who are asymptomatic but have amyloid in their brains. The remaining question about that trial is whether 2 years will be long enough to see an effect. There is also a phase III coming up for Biogen’s BIIB037 antibody. BIIB037 is different from other mAbs tested for AD in that it was discovered as a cloned naturally occurring human autoantibody against oligomeric Aβ, whereas many other mAbs are humanized mouse antibodies.

For information on preclinical stages of AD, a useful reference is [Sperling 2011].

Dr. Tanzi has seen brains of elderly people who had extensive Aβ and tau pathology but no inflammation and who were asymptomatic at time of death. Therefore, inflammation must be critically important in AD. Dr. Tanzi has also reported that Aβ has antimicrobial properties [Soscia 2010]. He is now promoting the “immunogenetic hypothesis of AD” involving compromised adaptive immunity.


Q. Why are there no good mouse models of AD?

A. Aβ production in mice does not lead to tau tangles unless you co-express mutant tau, at which point you no longer know if the Aβ is what is causing tau tangles. Based on our 3D culture model, we now think tangle formation requires 4-repeat tau, whereas mice predominantly express 3-repeat tau, so that may be why.

Q. On what scale and timeline are you now able to produce your 3D neural cell culture models?

A. We have scaled it to a 96-well format and we see tangles within 9-12 weeks.

Q. For clinical trials in asymptomatic individuals who may be 15 years away from disease onset, what would you use as an endpoint in clinical trials?

A. One possibility is to get people who are in the very early stage of starting to have mild cognitive impairment, and show a benefit there, in the hopes of later expanding the indication to include entirely asymptomatic individuals who have amyloid in their brains.