After a long, challenging period for CNS R&D – marked by late-stage clinical failures, mass layoffs, and limited early-stage investment interest – it has recently become popular to suggest that neuroscience drug discovery is poised for a renaissance.
I certainly hope so. From my perch – co-founding, leading or advising multiple CNS discovery-focused small companies over the last 10+ years – I am indeed encouraged by the progress and potential in the field. My recent discussions with both researchers and BD professionals, many with decades of experience in this space, are confirming my optimism.
There have been a number of positive recent developments, such as Biogen’s enhanced focus on neurology, Novartis rebuilding its neuroscience discovery research, and the successful progress and financings of companies like Sage Therapeutics, Voyager Therapeutics, and Denali Therapeutics. Other large companies are also increasing CNS research budgets and announcing new commitments to neuroscience research, and a slew of acquisitions and partnering deals suggest renewed interest in the space (here).
So, why is this happening?
The clinical unmet need in neurodegenerative and neuropsychiatric diseases remains daunting…but this has always been true.
What has changed in the last few years? Encouraging recent clinical data (or at least fewer high profile failures), along with translational science progress, advances in basic research and increasing deal activity, are starting to produce more early-stage Pharma and VC investment in CNS and, hopefully, a virtuous cycle of more breakthroughs in the clinic.
I’m not quite ready to declare a renaissance, however…at least not until we can better answer some fundamental questions:
More precisely defining and diagnosing diseases and selecting patients: Can we improve upon antiquated, “syndromic” disease categorizations?
At a reception at the JPMorgan conference, I suggested to a startup CEO that “in 50 years there will be no Alzheimer’s disease.” He looked at me skeptically, and kindly suggested that if I really thought we could cure AD in 50 years, I might want to try the cognition-directed agents we are developing at Rodin.
My point was not that we would cure Alzheimer’s in 50 years. Rather, we may no longer refer to the diverse set of pathophysiologically and clinically distinct cognitive dysfunctions as a single disease.
Reinforcing this perspective, a senior head of R&D recently said at a conference, “it’s time we stop referring to neurological disorders by the names of long-dead neuropathologists.” With apologies to the legacies and profound contributions of both Dr. Alois Alzheimer and Dr. James Parkinson, patients deserve a more nuanced approach than this syndromic characterization.
Neurologists and psychiatrists are starting to migrate towards new diagnostic approaches. Some predict that diagnostic and treatment paradigms in neurology will soon more resemble the targeted, personalized medical management prevalent in oncology. It still feels like early days here. But while CNS disorders are exceptionally multi-factorial, more precisely targeted diagnosis will help drive discovery of novel CNS drug candidates over time.
Target validation: What does it mean, and how do we get there?
Countless targets have been proposed in CNS drug discovery, with varying extents of functional, genetic, and pharmacologic validation. Yet perhaps nowhere more so than in CNS discovery are so few targets deemed adequately validated to be tested in well-powered clinical trials.
But what constitutes a validated target in CNS disease? I co-founded Link Medicine over 10 years ago with Peter Lansbury and Craig Justman, leaders in understanding the mechanistic role of alpha-synuclein in Parkinson’s disease. Synuclein has long been considered as a very promising target in Parkinson’s disease, yet at Link we ultimately pivoted towards other targets. Our challenge: even if we could demonstrate that our compounds reduce synuclein aggregation, or provide a motor or cognitive benefit in mice, would this translate to a clinically meaningful benefit for Parkinson’s patients? And do we even know what trial to run to answer that question?
In the last few years, there has been a notable uptick in activity for synuclein modulators (e.g., clinical programs from Biogen, Roche/Prothena and UCB /Neuropore).
This is encouraging, but still … have we answered whether synuclein is a validated target? Not if the standard is an approved, synuclein-targeting therapeutic, or clinical proof of concept that effective synuclein modulation leads to patient benefit. Targets should be thoroughly vetted, to be sure. Yet it should not take 25-40 years to get from initial target identification to clinical proof of concept.
For this reason, much of the rest of this post will focus on new opportunities and remaining challenges in how we can intelligently but more effectively discover and translate in the CNS area.
Evaluating, prioritizing, and translating innovation: How can we prioritize or reconcile different/competing mechanistic hypotheses to mobilize investors, researchers, and clinicians behind a new approach?
This is a corollary of target validation. Rigorous inquiry and debate are healthy and necessary components of scientific progress. But too often in complex fields like neurology, individuals’ strongly-held beliefs can be roadblocks. Neuroscience, perhaps more than any other biological/medical field that I’ve been exposed to, seems as polarized as voters in the American primary elections. Because so little has resoundingly “worked,” it is harder to point to what works, and easier to predict that a new approach is likely to fail. All of us in neuroscience are prone to this.
Biomarkers: Can we accelerate biomarker development and streamline biomarker adoption, for patient selection, target engagement, and preclinical/clinical pharmacology?
Serendipity once ruled CNS drug discovery. Many neuroscience drugs discovered in the 1950s and 60s (arguably the first golden age of CNS drug development) were discovered in large part via animal pharmacology, opportunistic clinical observations, or by chance. This led to progress and patient benefit, but also ultimately yielded a heavy reliance on modulation of serotonergic, dopaminergic, and cholinergic neurotransmitter pathways with preclinical and clinical readouts largely focused on subjective behavioral measures.
In today’s world, focused on target-driven discovery and disease-modification, there is less room for serendipity. Biomarkers are crucial to streamlining target validation and clinical evaluation of new candidates, while minimizing subjective endpoints that have historically been troublesome in the CNS space, leading to false-positives, strong placebo effects, and extensive interpatient/intrapatient variability.
As such, biochemical and imaging biomarkers are increasingly incorporated as early as possible for patient selection, measuring target engagement, dose selection, and, ultimately, clinical activity readouts. We need to continue to validate alternative approaches, such as EEG (subject of an upcoming blog post by Vanessa King, CEO of Luc Therapeutics), which can be more easily bridged from preclinical to clinical measures, so we can rely less on the animal-to-human conservation of subjectively evaluated behavioral measures.
Asking startups to develop a suite of target-relevant companion biomarkers in addition to their therapeutic adds risks, costs and complexity. Biomarker development therefore is a great leverage point for “pre-competitive” activity. Indeed, disease-focused foundations and NIH have taken great strides to fund and support biomarker validation that will help the translational potential of specific therapeutic strategies and benefit the field as a whole.
This is a significant focus at Rodin, and in our collaboration with Biogen. Of course our primary effort is driving towards a safe and efficacious clinical compound, but biomarker strategy needs to be incorporated as early as possible into discovery timelines.
Better enabling tools: Can we leverage vastly improved technology and analytics to streamline discovery and development?
A broad array of new tools is emerging in the CNS field to collect and analyze data with increased precision, breadth, and accuracy. Like with EEG, enabling technologies (such as imaging and cloud computing power) are improving rapidly, making the new generation of tools to measure desired neurological modulations almost unrecognizable from prior generations.
This has improved a variety of screening and biology assay modalities. For example, take in vivo phenotypic screening, which until quite recently mostly comprised of simple observational assessments, which are variable both intrinsically and in their measurement.
Big data and machine learning techniques improve our ability to assess behavioral phenotypes in other models – such as flies, worms and zebrafish – to dramatically increase throughput and lower cost relative to rodent models. I co-founded a company, Teleos Therapeutics, that measures over 14,000 features over a 20-minute assay in zebrafish, producing a robust, reproducible data set that can assess neuroactive compounds in vivo, with throughput comparable to in vitro systems. Zebrafish larvae develop functional blood-brain-barriers at 3-4 days post-fertilization, providing an early and deep analytical read on neuroactivity (certainly more so than in a cellular system), and neurotoxicity.
“Target-driven phenotypic screening” may seem like an oxymoron, but increasingly we can use gene editing and other approaches to research systems capable of merging target-driven screening with behavioral outcome measures earlier in the process.
Rallying VC/Pharma investment and powering the virtuous cycle: Will VCs and industry fund innovative early neuroscience research with long-term conviction?
After the financial markets swooned in 2008, a number of venture firms abandoned preclinical investments. The pain of this approach for entrepreneurs was perhaps most acutely felt in CNS. With few novel CNS approaches in the clinic, and several high-profile clinical failures, funding dried up across the board. This just compounded the flight of several Pharmas from discovery-stage CNS research.
This is indeed one of the main reasons I joined up with Atlas to lead Rodin. Atlas is one of the few firms that has continued to fund early, high-risk/high-reward CNS research, even in the dark days when it seemed that both investors and Pharma were collectively heading to the exits (here).
More recent trends seem to support renewed venture funding in neuroscience, which should help drive a virtuous cycle of innovation, deal-making and pipeline-building. Hopefully this will continue to erode the stigma that CNS still carries in some parts of the venture and Pharma worlds.
Epilogue: Where have we been, and where are we going?
At times, CNS drug discovery has felt like a vast, lonely wilderness. And if some of the current generation of high-profile, expensive clinical trials fail to reach positive endpoints, we may have to work through some retrenching in this area.
But there is fundamental cause for optimism. We seem to be moving away from over-reliance on non-conserved behavioral readouts from contrived animal models with questionable predictive value, towards a more nuanced view of translational priorities. We are witnessing rapid evolution in how we diagnose and characterize CNS-involved diseases. And we are starting to apply new technologies to validate targets, discover neuroactive compounds and use more objective and translatable biomarkers and clinical endpoints to improve the speed and precision by which we can make more rapid and robust assessments of drug candidates.
Here at Rodin we are trying to apply these philosophies in real-time. And within the next few years, with successful readouts from some of those high-profile clinical trials – along with productive utilization of modern patient selection, biomarkers, and tools – we may well be in the position to truly declare a research renaissance in CNS.