Has nothing important happened in psychiatric research in the last 25 years?
Some believe so. Here’s a quote from the science writers Carl Zimmer and Benedict Carey in the July 22, 2014 New York Times “Despite decades of costly research, experts have learned virtually nothing about the causes of psychiatric disorders and have developed no truly novel drug treatments in more than a quarter century.”
The writers were profiling a gift of $650 million to the Broad Institute of Harvard and MIT to research psychiatric illness. The announcement was timed to accompany new research reported in Nature. Studies on the genomes of more than 140,000 people – including 37,000 diagnosed as schizophrenic, added 83 gene loci to the 25 previously known “related” to schizophrenia. As Eric Lander, head of the Broad Institute declared, “for the first time there’s a clear path forward.”
We have heard this story before. The Human Genome Project was sold as the way to “find the genes” for schizophrenia, and bipolar illness, and cancer. What was found? Hundreds of genes that “appeared” to increase incidence by one or two or four percent – just like the “new” loci do.
What have we missed? The importance of serendipity.
And that’s where Big Data – crunching all those vast data sets again and again – might help us out.
The Big Breakthroughs
Consider the origins of many breakthroughs of the last seventy years:
1.Where did cognitive therapy come from? A young psychiatrist from Philadelphia named Aaron Beck who found that the “gold standard” of psychoanalysis was not helping his depressed patients, but seemed to make them worse.
2. How did lithium become used for mood disorders? From an Australian opthamologist who supposedly believed there was something “schizophrenogenic” in patients’ urine. He wanted to inject it into animals - then people - and see what it did. But his solubilizing agent, lithium urate, was calming his rabbits down. So he used lithium instead.
3. What about arguably the most statistically effective antidepressants ever developed, the Monoamine Oxidase Inhibitors (MAOIs)? Trials of an “antitubercular” drug in mental patients produced awful side effects. But the patients were less depressed. Modifying the anti-tubercular drug produced the MAOIs - and much antidepressant research.
A large part of successful treatments for many illnesses – from antibiotics to depression – exist because something “really strange” happened – results good and unexpected. We call that serendipity.
It’s time for serendipity to make a comeback. Because Big Data – all that potentially awesome processing power – should make it easier to find needles in haystacks.
But to succeed you have to recognize the power of chance – which means looking at data in non-traditional ways. Here are a few possible guidelines:
1. Assume much less is known than known. You don’t want to be the researcher who looks for his lost keys under the streetlamp because that’s the only light nearby.
Researchers are very excited characterizing the approximately 22,000 human genes. They don’t talk nearly as much about the 2-20 million control genes residing in “junk DNA” that determine how those genes actually are produced and used - and where some of the new "gene loci" of schizophrenia appear to be.
Yes, the brain is very, very complicated – on levels most have yet to recognize and appreciate.
2. Utilize an information approach to biology. One simple credo: life is a regenerating information system powered by chance.
Much of that chance gets expressed through evolution – a continuing process in every human body.
Take for example the forced evolution – “somatic hypermutation” – that provides us with new antibodies by the billions to help control the continued mutation of the viruses, prions, bacteria, rickettsia and other stuff that might kill us. Immunity is just one biological information system that may rival the brain for complexity. Is it any surprise that a lot of the new gene loci associated with “mental illness” reside near immune function regions?
Fortunately we don’t have to figure everything out to save many lives. Epidemiology can discover problems and potential solutions – like the relationship of cigarettes and lung cancer – without explaining the mechanism. That’s one thing that's nice about serendipity.
Consider this – if you inject lactobacilli into the guts of gene identical mice they are much, much harder to make depressed or to stress. Yogurt may not treat human depression, but what mechanisms are working here?
4. Don’t worry if you can’t see the real connections at the beginning. The human brain is very good at discovering salience – finding outliers that don’t fit our expert systems/forms of logic. So if a group of Faroe Islanders all get multiple sclerosis – and start to suddenly gamble uncontrollably – that is something to look at.
Most of the relations Big Data will throw up will happen by chance. They’re still worth studying – particularly if they get us thinking.
5. Fat cells don’t know they’re just supposed to store fat. That your abdominal fat is a giant endocrine gland should put paid to the overall utility of our many disciplinary silos.
Information goes all over the place. It likes to be used. Biological information molecules – from neurotransmitters to nitrous oxide – are multi-valent. Evolution conveniently uses the same stuff for many purposes. Like human words. They may have several meanings in one language – and completely different meanings in another.
Think of when a “neurotransmitter” like serotonin is found in the gut.
6. Use stuff you know that works now.
In June I attended a lecture by a prominent chief of a well known Neurology Department.
It was mesmerizing – Alzheimer’s disease was just waiting to be conquered.
There were beautiful slides of amyloid proteins folded into bizarre contortions. Just a bit more research and we can “dissolve” the complexes and save tens of millions from dementia.
It was another drug fueled fantasy – in this case, the drugs supplied not by Latin drug cartels but by Big Pharma. Just use the new drugs early enough and everything should be fine.
The lecture made little mention of tau proteins. No mention of RISK proteins – which deeply complicate the Alzheimer’s picture.
Not even a hint that the treatments that “revolutionized” Alzheimer’s treatment in the 1980s and 1990s don’t work that well. Nor lip service to the new analyses declaring one third of Alzheimer’s could be prevented by lifestyle – particularly almost any kind of physical exercise.
That’s 33% versus 0% success – and you know which part gets the money.
Yes, we need much basic research to figure out dementia. But it’s time to stop repeating beautiful lies when there’s stuff we can do right now we know works – especially for problems that are big.
Big Data Redux
Most every electronically filed text is getting compiled somewhere. It’s not just the NSA that’s reading your email.
For there is a real revolution in information. There’s a lot more of it – and more ways to look at it. Just as in electronic data, so in biological data.
But how you look at a problem changes the problem.
Regarding the human body as an information system – or series of interlocked information systems working together on a dizzying number of levels – is not really sexy. It doesn’t fit the nice linear models we have of the world.
It is nice when X directly causes Y – and that eliminating X wipes out the disease.
Yet the brain is far too complicated for such models. So is life itself.
But we can appreciate and use all that complexity if we’re smart. And there’s very good evidence that we are smart.
Much of the time.