After Asheville and a too-short visit with friends in Memphis, we made it to Austin to plant our sabbatical roots for the remainder of 2015. The week we arrived, I was happy to discover UT’s Business of Science seminar series—which I helped launch with the SBS (now CNS) Postdoc Association in 2012—is still going strong, thanks to ongoing support from area entrepreneur Grant Schoenhard.
Speaking this fall was Spiros Liras, also a former UT postdoc, who served until recently as Senior Director of Neuroscience for pharmaceutical giant Pfizer; he now heads their unit on cardiovascular, metabolic, and endocrine diseases. Liras warned that the two drug development stories he would share did not have happy endings; but, they shone a rare spotlight on current challenges in modeling and validation, and on recommendations for the future.
Liras told the engaging tales of two neuroscience drugs developed under his watch. On the coattails of sildenafil (Viagra), Pfizer focused substantial attention in the early 2000s on other drugs to inhibit phosphodiesterases (PDEs), a large family of proteins that control chemical signaling in the brain, heart, and other tissues. The pharmaceutical importance of PDE inhibitors—and Liras’ prominence in the field—are evidenced, among other things, in a recent textbook he edited with Viagra co-inventor Andy Bell. One of Liras’ targets, PDE10, has been implicated in schizophrenia; another, PDE9, plays a role in cognition that could be leveraged for dementia and Alzheimer’s Disease.
Credit: Yikrazuul: PDE5 with Viagra-like inhibitor
A few themes arose in the development of both PDE drugs. For one thing, drug modification continues to surprise computational chemists with unpredictable changes in binding modes, supporting a need for better docking or other modeling tools. For another thing, chemical knowledge remains key: even selective, efficacious binding can be trumped by a decrease in half-life or bioavailability, a drug’s ability to reach its target through the circulatory system. And some neurological conditions constrain treatment options: for example, individuals with schizophrenia may not tolerate bitter, injectable, or frequent dosing. Paraphrasing Stewart Lyman, Liras quipped It’s not rocket science—it’s harder: it’s drug discovery. (Dean Burnett recently ranked neuroscience and rocket science a tie on 8 top science measures).
Credit: xkcd.com: DNA
Long story short, the iterative integration of computation, chemistry, structural biology, bioassays, behavioral pharmacology, and toxicology pursued by Liras’ team was impressive—and ultimately successful. The PDE9 and PDE10 drugs brought to human trials were safe and capable of reaching the brain; they were also selective and effective at inhibiting their targets in vitro, and in modifying animal behavior proxies for human cognition and schizophrenia, respectively. Yet frustratingly, in both cases, behavioral efficacy in humans was marginal.
Credit: Nature Reviews Drug Discovery: Reasons for drug attrition
According to Liras, these stories are becoming the rule in neuroscience, not the exception; and the data may back him up. In the early 1990s, the most common reason for a promising drug to fail in the clinic was poor bioavailability. But by 2000, improved in vitro, in silico, and animal models had reduced attrition in this category to ~10 %. Early assumptions that only very oily drugs—with accompanying high toxicity—could pass the blood-brain barrier were updated with a more nuanced understanding of molecular size and charge factors, enabling better targeting to the brain. In contrast, problems of efficacy—how well a drug actually modifies its human target—continued to motivate nearly 30 % of drug failures. And for drugs targeting the central nervous system (CNS), success rates were well below the discouraging ~11 % average, and less than half as successful as in cardiovascular or infectious disease.
Credit: Ibid.: Drug success rates from first-in-man to registration
Liras attributed much of this attrition to the fundamental limitations of animal models for measuring neurological efficacy: Biology may be conserved; behavior is not. He pushed for greater development and reliance on ex vivo methods, such as computational physiology, tissue culture, and noninvasive biomarkers: for example, gamma wave scalp recordings can indicate disrupted GABA signaling in patients with schizophrenia. He also emphasized the importance of basic biology research in identifying appropriate pharmaceutical targets, and understanding the molecular pathways in which they interact. In retrospect, it’s clear that PDEs were a focus not just for their clinically relevance, but also for their accessibility to high-resolution structural studies. Alternative methods such as cryoEM are already expanding the library of such targets; and molecular simulations promise to open new windows on the dynamic motions of proteins, including potentially key mechanisms of allosteric modulation.
Credit: Rachel Davidowitz: Allostery
Comfortingly, some failed drugs may yet prove useful: Liras’ new group at Pfizer is now testing the same PDE9 inhibitor, originally intended for cognitive enhancement, against heart disease. Indeed, Viagra was first developed as a cardiovascular drug (and still serves that purpose), though its side effect gained more attention. At least for this former Longhorn, and his inquisitive Postdoc Association audience, the search for new and better model systems in drug development evidently continues to present new challenges—but also to motivate growth in both the basic research and therapeutic communities.
Top credit: drugdevelopment-technology.com: Biotech pipeline