t's often said that placebo-controlled randomized clinical trials are the
“gold standard“ in medicine. Why is it, then, that they so often
contradict each other? We've all seen the studies on hydroxychloroquine,
some of which showed a benefit and some of which didn't. We can ascribe some
of this to politics. But the real problem is that running a clinical trial is
very difficult.
The purpose of a clinical trial is to test a theory, and one theory that is being tested is that Alzheimer's disease, or AD, is a form of diabetes. It is associated with Suzanne de la Monte and colleagues in Rhode Island, who coined a catchy term for it: “type 3 diabetes.”
It's based on two observations: first, that patients with AD, like those with type 2 diabetes, have insulin resistance, which means that there's something wrong with their insulin receptors—most likely they're downregulated, meaning there are fewer of them that are working—so a dose of sugar causes their blood sugar to rise to dangerous levels, known as hyperglycemia.
The second observation is that diabetes is considered a risk factor for AD. For type 1 diabetes, where there is a loss of beta cells in the pancreas that make insulin, there are defects in information processing, cognitive flexibility, and attention. In type 2 diabetes, most but not all clinical studies show a small but statistically significant increased risk of dementia.
Sugar and the brain
The reason sugar is harmful isn't clearly understood. One theory is that high concentrations of sugar can react non-enzymatically with proteins, glycosylating them and changing their biochemical properties. This reaction is well known: it's called the Maillard reaction, and it's what causes cookies and marshmallows to turn brown. What we don't know is how these glycosylated proteins cause the vascular and kidney complications that are seen in patients with diabetes.
Sugar also increases reactive oxygen species, or ROS, in the brain. A high fat diet is said to do something similar. A high-fat diet can cause systemic inflammation, caused by oxidation of double bonds in unsaturated fatty acids. This is known as lipid peroxidation, and it does many other things besides increasing ROS, such as releasing cytokines, which can trigger nasty cellular effects. If diabetes were simply an excess of peroxide, superoxide, and hydroxyl radicals, we could treat it with N-acetylcysteine and free radical quenchers, but we can't.
Some studies also showed that hyperglycemia raises beta-amyloid levels, which were until recently thought to cause AD. However, this theory is rapidly losing adherents due to repeated failures of antibodies against beta-amyloid to slow AD in clinical trials. The latest theory at the moment is that beta-amyloid acts by binding to tau protein (which causes neurofibrillary tangles), and so we must eliminate both. But this is a long shot, as beta-amyloid and tau start in different brain regions.
Another theory is that ApoE4, a protein that's a risk factor for AD, is unable to bind tau protein and so cannot protect it from being hyperphosphorylated, suggesting it could worsen the neurofibrillary tangles seen in AD. There are claims that insulin inhibits the enzyme that phosphorylates tau. But oddly enough, insulin completely blocks the ability of another protein called IDE to break down beta-amyloid. This means insulin would cause beta-amyloid to accumulate and form plaques. Insulin and beta-amyloid compete for IDE, so beta-amyloid would also protect insulin from degradation, and indeed one clinical study found more insulin in blood plasma of AD patients. This is all very confusing, and the message is clear: interactions among all these proteins are really, really complicated.
Neurosteroids
What's happening is that the brain takes ordinary molecules like estradiol, glutamate, and allopregnanolone and uses them for purposes unrelated to their function elsewhere. For instance, glutamate, an amino acid used to make proteins, is the most important neurotransmitter. Estrogens are found in the brain in both males and females; they're important for learning and memory. And lower levels of allopregnanolone, a metabolite of progesterone, are associated with the perception of social isolation, leading some to call allopregnanolone the ‘loneliness hormone.’ That's consistent with claims that loneliness causes AD, and indeed allopregnanolone does cure AD in transgenic mice; but this doesn't mean much, because almost everything we give these mice cures them. One group in Chicago even suggested giving this molecule, a member of a class of cholesterol derivatives known as neurosteroids, to humans to alleviate loneliness. I'm not convinced that's a good idea.
A friend of mine found evidence a few years ago that small amounts of insulin may be expressed in the brain, but most insulin is made in the pancreas and imported through the blood-brain barrier.
Part of what makes the diabetes theory appealing is the amyloidosis connection. Diabetic patients produce human islet amyloid polypeptide, also known as amylin, which leads to pancreatic beta-cell dysfunction. Serum amyloid P, not to be confused with beta-amyloid, is a small acute phase protein (meaning it's induced by interleukin-6). It is a pentraxin similar to C-reactive protein that binds dead cells and promotes the killing of cells by the complement system. (Serum amyloid A is another acute phase protein. It is upregulated by up to 1000-fold during inflammation, and it associates with high-density lipoproteins or HDLs.) The similarity to amyloidosis is compelling and it may explain the staying power of the beta-amyloid theory. The insulin theory has led to the use of streptozotocin, which induces diabetes, by some researchers as an animal model for AD.
Type 2 diabetes shares much in common with AD: both involve insulin and insulin growth factor, inflammation, oxidative stress, and beta-amyloid formation. But this does not mean that it makes sense to call AD a form of diabetes. Inflammation and oxidative stress have many causes, as does beta-amyloid (the source of amyloid plaques), which shoots through the roof after a head injury.
Anti-diabetes drugs bombed out in clinical trials
Clinical trials of anti-diabetes drugs have been disappointing. Metformin bombed out in phase 2, though a phase 4 trial is recruiting to test metformin in conjunction with lifestyle interventions. Rosiglitazone bombed out in phase 3; a similar drug, pioglitazone, is still being tested. But the final nail in the coffin is a study just published in June 22 2020 JAMA showing no cognitive or functional benefits of intranasal insulin in patients with mild cognitive impairment or moderate AD (MMSE > 20).
So, is AD a form of diabetes or not? At the moment, it doesn't appear so. We have a lot of clinical correlations, but what we really needed was a coherent theory, and that can only come from basic science. The relationship of basic science to clinical research is the same as that of theoretical physics to experimental physics: basic science comes up with theories, often shown as causality diagrams (which tend to be more complicated but also much more colorful than those used by philosophers) and the clinicians test them.
The idea that AD is a form of diabetes was really more of a catchy slogan than a bona fide theory. Theories are judged by what they can predict, in this case whether insulin can treat AD. So far, it does not. The term type 3 diabetes never really caught on, and the theory now looks like its time is past.
aug 16 2020, 8:17 am
