Cancer treatment has been transformed, at its root, by a transformational change in how it is classified.
These days, which organ a tumor arises in is often less important than its molecular drivers, which can be sensitive either to specific targeted treatments, or increase the chance that a tumor will respond to immunotherapy.
Those successes have not escaped the notice of researchers in other areas of biomedicine, and diseases including heart failure, asthma and polycystic ovarian syndrome are being looked at with an eye to subdividing them in ways that brings diagnostics into the molecular era.
Nowhere do those changes have greater potential than in disorders of the brain – in part because there is nowhere much to go but up as far as classifying neurological diseases goes.
Even the most basic categorization into neurodegenerative, neurodevelopmental and psychiatric disorders does not hold up to scrutiny.
Huntington’s disease (HD), for example, is “technically a neurodegenerative disease and a prominent movement disorder, but some of the cognitive and psychiatric symptoms are among the most debilitating and among the earliest” to manifest themselves, Steven Finkbeiner told BioWorld. Parkinson’s (PD) is another movement disorder that can come with both dementia and Parkinson’s psychosis, and Alzheimer’s disease (AD) can also have psychiatric components.
Some neurodevelopmental disorders, too, appear to be less developmental than they once seemed. The disabilities of Rett syndrome, once thought to be a developmental disorder, are actively maintained by ongoing processes, and in animal models, repair of the underlying genetic defect can reverse many symptoms even if such repair is delayed until the animals are near death.
“Disorders got divided arbitrarily many years ago in medicine,” as Finkbeiner, director of the Gladstone Institutes’ Center for Systems and Therapeutics & Taube/Koret Center for Neurodegenerative Disease, summarized the state of affairs.
And those arbitrary divisions are not doing patients or drug developers any favors.
Modeling traditional diseases whose diagnosis is based on symptoms is part of the problem. “For many years, the approach that has been taken has been to use mice or other model systems… [and] do whatever it took to get them to show symptoms,” Finkbeiner said.
And frequently, what it took was “wild overexpression” of proteins that had been implicated in humans to model disease symptoms, or “phenocopy” a disease.
That approach, by and large, “has not led to anything that has worked in a patient,” Finkbeiner said.
An alternate approach, sometimes called endotyping, is to split umbrella categories up by their underlying mechanisms.
“If there are patient strata, then that could be a major explanation for why things have failed,” Finkbeiner said, because “we just haven’t really figured out the subtypes that would respond to certain medicines and not to others.”
There are multiple sources, such as electronic health records and samples collected by the UK Biobank and the U.S. NIH’s All of Us, for patient stratification, which is in a sense the diagnostic arm of precision medicine.
Finkbeiner and his team look at both patients and stem cell models, ultimately allowing the team to link “things we can measure in a dish and things we can measure in a patient,” he said.
Microstate of play
At the virtual annual meeting of the International Society for Stem Cell Research (ISSCR) in June, Finkbeiner gave an overview of how he and his colleagues are using their approach to stratify autism into subtypes.
They first identified de novo mutations by sequencing trios of neurotypical parents and their autistic children, leading to the discovery of roughly 130 genes which, when mutated, appear to contribute to disorders on the autism spectrum disorder (ASD).
The team is now “trying to knock those genes down… and look at what happens in cells,” he said.
The goal is to identify different forms of disorders such as ASD, which is very heterogenous in its presentation, and the different genetic constellations that are presumed to underlie them.
In the day and age of single-cell analysis, EEG can seem like a dinosaur. An EEG signal integrates the activity of thousands of neurons, and the technique was first used to record brain signals nearly 100 years ago.
But the dinosaur is still yielding new insights into brain disorders.
In June 2020, Swiss scientists published a study on using EEG “microstates” to identify an activity endotype that was present in schizophrenic patients as well as their unaffected siblings.
EEGs measure the collective activity of many neurons, giving insights into what the brain as a whole is up to. Microstates are recurrent global activity patterns that last for somewhere between 60 and 120 milliseconds. Four main microstates, prosaically named A through D, have been identified.
In their paper, which appeared online in Nature Communications, the team showed that both schizophrenic patients and their unaffected siblings had brains that were more frequently in microstate class C, but less frequently in microstate class D, than healthy individuals. Siblings also showed an increase in microstate B, which the researchers interpreted as a potential compensatory mechanism.
The authors did not look for any genetic alterations that correlated with microstate frequencies. But in their paper, they noted that previous work has shown a similar effect for the 22q11.2 deletion syndrome, which also raises the risk of psychosis.
What the microstates and their alterations mean in terms of differences in brain function remains unknown for now. "But for diagnostic purposes,” senior author Michael Herzog said, “you don't need to know that.”