Wanted: Better Brain-Process Biomarkers for Drug Trials


by Jim Schnabel

January 3, 2012

Eight years ago, Peter Lansbury decided to become a drug developer. He was a neurology professor at Harvard Medical School with a background in chemistry, and had been doing well-regarded research on the origins of neurodegenerative diseases. Now he had an idea for a blockbuster treatment: a drug that would boost the brain’s ability to dispose of the harmful protein aggregates that seem to cause many of these diseases.

Lansbury and two investors founded a company, Link Medicine, and began screening libraries of chemical compounds for those that would increase autophagy, a natural process that clears away unwanted protein aggregates. Eventually they had a lead autophagy-boosting compound, LNK-754, that worked well in mouse models and appeared safe in human volunteers. In principle, LNK-754 could be useful in treating a number of illnesses including Parkinson’s, Alzheimer’s, and Huntington’s diseases. All that remained, it seemed, was for a large pharmaceutical company to buy the marketing rights and fund the large-scale clinical trials needed to prove LNK-754’s efficacy in humans.

But that hasn’t happened. “It used to be that if you had a drug whose mechanism makes sense, a drug that works in animals and is safe in people, you could immediately go and do a large clinical trial. But for drugs against neurodegenerative diseases, those days are gone,” says Lansbury.

The problem is that for these diseases, the standard measures needed in a conclusive clinical trial—such as cognitive test scores, for Alzheimer’s patients—are expected to show clear evidence of a treatment effect only very gradually, and only in a population of at least hundreds of patients, even for a drug that works. For a would-be disease-modifying drug, that typically means a trial of at least 18 months, costing on the order of $100 million. At the same time, pharmaceutical companies are all too aware that potential neurodegenerative disease treatments have almost always failed in clinical trials, in recent years. “The big pharma companies have become very risk averse,” Lansbury says. “The big issue now is not really the lack of drug strategies, it’s the inability to assess those strategies in short, economical trials.”

Brain-process biomarkers

More companies would justify undertaking such lengthy, expensive trials, says Lansbury, if startups such as his could provide evidence, up front, that their drugs at least hit their targeted brain processes—in humans, not just in animals. Such evidence could come from short, inexpensive trials that use relatively fast-changing “biomarker” measures. For example, Lansbury’s company would like to be able to show, either with brain-imaging techniques or by sampling cerebrospinal fluid (CSF), that a few days or weeks of treatment with LNK-754 truly does boost autophagy in the human brain. “The idea is that the drug should have a measurable effect in the brain much sooner than one sees any effect on disease progression,” says Lansbury.

Scientists can gauge the level of autophagy in the brains of lab mice by sacrificing the animals and directly measuring the levels of autophagy-related proteins in their brain tissue. However, they currently have no feasible way to make such measurements in live humans, for example by sampling CSF with a spinal tap. “If you look for connections between what’s measurable in CSF, and what’s known to be important in autophagy, you get really no answers, unfortunately,” says Lansbury.

The problem applies to many—perhaps most—brain processes that are relevant in neurodegenerative diseases. Boston-area startup Proteostasis Therapeutics, for example, is developing a drug to boost a different kind of protein-clearance process in the brain, known as proteosomal clearance. “At Proteostasis we’ve spent quite a bit of time and money identifying markers that give you an early [treatment] response,” says Jeffery Kelly, a Proteostasis co-founder who is also chairman of Molecular and Experimental Medicine at The Scripps Research Institute in La Jolla, CA. “These things take a lot of work to develop and validate, though.”

Large drug companies also are trying to develop biomarkers that can give them early evidence of drug effects in their studies, says Lansbury. In fact, a distinct industry seems to be developing around this concept. Companies including Emeryville, CA-based Kinemed and St. Louis-based C2N Diagnostics now specialize in developing biomarker tests that can be used to speed drug development.

Kinemed, for example, has more than a dozen pharma company clients, and has developed clinical techniques to measure, among other things, the efficiency with which neurons transport key proteins down their output stalks—a basic brain process (“axonal transport”) known to be impaired in Parkinson’s and other neurodegenerative diseases. To measure axonal transport of a target protein, researchers give subjects water that contains a high concentration of deuterium, a stable, non-radioactive form of hydrogen; the deuterium is soon drawn into neurons, and incorporated into newly synthesized target proteins; these are eventually transported out of neurons into cerebrospinal fluid, which can be sampled with a spinal tap. As neurodegeneration progresses, axonal transport becomes less efficient and deuterium-labeled proteins take longer to appear in CSF. “We can have a readout of this pathway within weeks,” says Patrizia Fanara, Kinemed’s vice president for neuroscience.

Such a fast readout could help a pharma company decide quickly whether or not a drug is working, and which patients respond best to it. In a recent study of the technique in Parkinson’s disease patients for the Michael J. Fox Foundation, Fanara and her colleagues confirmed that patients’ axonal transport measures varied with the length of time they had had the disease. “We have to confirm it in future studies, but this marker may be useful even for early disease diagnosis,” she says.

Biomarkers that work during early stages of neurodegeneration are especially needed, argues Reisa Sperling, an Alzheimer’s researcher at Harvard Medical School and Boston’s Brigham and Women’s Hospital. “The real hurdle is having a marker that says we’re altering the course of the disease and is also sensitive to very subtle changes early in the disease process,” she says.

Biomarkers used at late stages of disease may be less informative for drug developers because, by then, too much damage may have been done for drugs to have a meaningful effect. Researchers from Hoffmann-La Roche and collaborating institutions reported in October in the Archives of Neurology that monthly infusions of their anti-amyloid-beta antibody product strongly reduced brain amyloid deposits, as measured by positron emission tomography—and they were able to show this clearly with data from only sixteen patients, who were treated for only two to seven months. Yet previous studies hint that even a powerful reduction of brain amyloid deposits doesn’t come soon enough to alter the disease course in already-demented Alzheimer’s patients.

In any case, says Lansbury, there simply aren’t enough measurable human biomarkers for key brain processes. Finding them may require, in some cases, studies of rare patient groups with conditions that specifically affect these processes—such as Lafora disease, a rare and fatal genetic condition in which neuronal protein clearance is impaired. Some incentive exists for specialist companies like Kinemed to do such studies, but Lansbury argues that this is an area that also deserves support from the National Institutes of Health and philanthropic research organizations. “It’s a perfect area for the government and/or philanthropy to get involved in, because it would have implications for everybody,” he says. “Once you can measure these basic processes quickly and cheaply and non-invasively in human brains, I think the drug development floodgates will open.”