Here's the theory. Hypertension, atherosclerosis, smoking, and the gene for ApoE4 increase the risk of cardiovascular disease, vascular dementia (obviously - vascular dementia is generally thought to be caused by multiple and increasing stepwise vascular insults to the brain, like little strokes or clots) and Alzheimer's dementia (which is associated with excess tau and amyloid protein build-up in the brain, so the brain starts operating rather like a fish tank that never gets cleaned). Folks with type 2 diabetes have twice the risk of developing Alzheimer's, and diabetics on insulin therapy have four times the risk.

There's a protein called insulin-degrading enzyme that does just what you might expect. It clears out insulin in the brain. It also clears out excess amyloid (at least in test tubes), so one can imagine if it were super-busy with the insulin, amyloid might get left cluttering up the joint. Unlucky mice with no insulin-degrading enzyme get dementia, and elderly people get increased amyloid in their cerebral spinal fluid when insulin is injected into their veins.

The obvious conclusion is that one wants appropriate, physiologic insulin levels so that your insulin-degrading enzyme can keep itself busy with the pesky amyloid, leaving none to form plaques. One way to achieve that is a low carbohydrate diet. Even the high-carbohydrate eating Kitavans, though, had exceedingly low fasting insulin levels (1), so a paleolithic-style diet will seem to do the trick if you don't have metabolic syndrome to begin with.

Is Alzheimer's disease increasing? Yes, absolutely. Wouldn't it be nice if we could do something about it? The National Institutes of Health convened a panel in early 2010 and determined there is no reliable way to prevent Alzheimer's.  One interestng fact must be considered - Staffan Lindeberg found no evidence of dementia or anyone who knew anyone with symptoms of dementia on the island of Kitava, who eat a paleolithic-style diet with starchy tubers, coconuts, fish, and no modern industrial foods, wheat, vegetable oil, or dairy.

So we know what Gary Taubes thinks.  Since Good Calories, Bad Calories came out, there has been quite a bit more research into the topic of metabolic syndrome and the risk for Alzheimer's dementia. And if you look at the primary sources, the results (not surprisingly) paint a less certain picture than in Gary's book.  For example, there are groups who seem to be protected from dementia by metabolic syndrome, which can be associated with hyperglycemia (2). Then there are diabetics whose brains don't seem to have an increase in beta-amyloid plaques despite insulin-degrading enzyme issues, directly disproving Gary Taubes' theory (3).  Other researchers who found that diabetes is assoicated with Alzheimer's, proving Gary Taubes' theory (4).  There's a brand new study that came out this week, showing more links between diabetes and Alzheimers. (5) And another brand new pilot study showing that intranasal insulin treatment (which would increase the ability of cells to soak up glucose) improved symptoms of dementia (6). This same study notes that folks with Alzheimer's disease tend to have lower than normal insulin in the brain than average - exactly the opposite of what you would expect in a type II diabetic, and opposite of what you would expect from Gary Taubes' theory. Finally some autopsy studies from Japan show no link between diabetes and Alzheimer's dementia (7).

How do we pull all that contradictory information together?  It's complicated.

When you try to connect dementia and hyperglycemia, you do come to some simple conclusions - hyperglycemia speeds up aging.  It's like our metabolisms in fast-forward.  Anatomic brain differences have been shown in patients with diabetes (type I and II) consistent with non-diabetic patients > 80 years old.  Also shrinkage of the hippocampus and the amygdala (these are also found to be shrunk in type II diabetes).  Patient with uncontrolled type II diabetes have worse cognitive function and memory.  Patients with more diabetic complications (suggesting poorer glycemic control) also have more cognitive difficulty.  Studies of the "oldest old" (>85 years) don't seem to show a difference between diabetic and non-diabetic populations, though.  Though at that point almost everyone starts losing weight (possibly improving diabetic control).  Alzheimer's disease is the cause of dementia in 82.5-91% of type II diabetics - which is greater than the general population.  Insulin degrading enzyme (IDE) is still important - patients with the genetic predisposition for Alzheimer's have decreased expression of IDE in the hippocampus.  There's enough interesting epidemiologic evidence and common sense plausible biologic mechanism that you do have to wonder. And it certainly doesn't hurt to keep one's blood sugar and insulin under good control for overall health and wellness.

The best hypothesis I can come up with is that a combination of the genetic predisposition to Alzheimer's (carrying the ApoE4 allele) and diabetes could put one at higher risk.  People with ApoE4 seem to have lower amounts of enzyme to clear away amyloid in the first place, so add lots of insulin, and you've got plaque city.  But hyperglycemia alone wasn't going to explain it - though everyone agrees that hyperglycemia is rotten for the brain.

So what does cause Alzheimer's?  Truth be told, a whole cascade of things played out over decades (8).  Beta-amyloid peptide is definitely a key player, but it certainly isn't the only one.  It all begins with amyloid aggregating in vulnerable areas of the brain (called plaques - and this takes years and years), followed by accumulations of tau protein (called tangles).  The plaques and tangles (when I was in residency, there was a holy war going on between different sects of brain researchers about which was more important - plaques or tangles) seem to interact with inflammatory cells in a way that the accumulated plaques and tangles finally trigger diffuse brain toxicity and neuronal death.  At the beginning, measuring amyloid can predict problems even before someone experiences the first clinical stage of Alzheimer's called "mild cognitive impairment" (MCI).  The cognitive decline seems to be triggered when tau protein increases.  So, to recap - long symptomless amyloid buildup, tau takeover, inflammation and neuron destruction.  Boom.  Alzheimer's dementia.