“No one is going to care about your project as much as you do.”
Steven Weissman is Vice President, Head of CMC at Cognition Therapeutics, a small pharma company outside New York City developing CT1812, a neurodegeneration candidate now through three Phase 2 trials in Alzheimer’s disease, Lewy body dementia, and dry AMD. He spent 18 years in the process chemistry groups at Merck before moving into the small pharma space, where he has managed CMC functions at Concert Pharmaceuticals, Lexicon, and now Cognition Therapeutics.
In the latest PharmaSource podcast, Steven explains how small biotechs can build a manufacturing supply chain from scratch, why cost of goods is a lever most teams pull too late, and what working on drugs that made it to market, and some that didn’t, taught him about when to invest in process development.
Get Comfortable With Uncertainty
The defining difference between big pharma and small pharma, Steven explains, is urgency. At a small company, the money to answer every question before a decision simply isn’t there. Steven argues that learning to act with incomplete information is a core CMC skill, not a compromise.
Decisions are never truly blind, he says. They are calculated. “You always want more information. The more information you have, the better decision you can make. Sometimes you have to go on your intuition, on your previous experiences.” The discipline is to de-risk cheaply along the way while accepting that some risk is unavoidable.
“You have to get comfortable with the uncertainty and the risk associated with making those decisions,” Steven says. “It’s an uncomfortable place to be, but it can be very exciting as well. It can cut both ways.”
How Cognition Cut the Cost of Goods by 85%
Cognition reduced the cost of goods for its lead compound, CT1812, by roughly 85%. Steven breaks the result into distinct levers rather than a single move.
The first is economies of scale. Early clinical material is often made at the one-kilogram scale for tox screening, so fixed costs are spread over a small base. Simply scaling to 40, 50, or 80 kilograms distributes those costs across a far wider base. The bigger gains came from redesigning the chemistry itself. Also, investment in process optimization of the two starting materials contributed to the cost reduction.
“When we moved from our first-generation process to our second-generation process, there were four fewer bond-forming steps, and we were able to eliminate three unit operations,” Steven says. “Then we also relocated our manufacturing from the United States to China. Those were the drivers that allowed us to reduce the late-stage API cost vs. that of the first kilo by about 85%.” The knock-on effect was a materially longer runway for the company.
Manufacturing in China and Managing Geopolitical Risk
The move to China was not a leap of faith. Steven had spent roughly a decade working with Asian providers in earlier roles and had travelled to China around once a year before 2020. That familiarity made the switch a measured decision when Cognition needed capacity that its legacy US vendor could not provide.
His guidance is to know exactly who you are working with. “There are some highly respectable and reliable vendors in China. A lot of big pharma, including Merck, have been working in China for over 25 years. It’s important to know who to work with, and to do your due diligence, auditing remotely when necessary but on-site when possible.”
He is watching the geopolitical backdrop closely, from the BIOSECURE Act to tariffs. His response is pragmatism over reaction. Many companies are qualifying a US-based backup source so they can switch if they need to, typically around the NDA filing stage when process validation begins. On the BIOSECURE Act specifically, Steven notes that scrutiny has narrowed to a small number of named companies and has quieted over the past 18 months, leaving Cognition comfortable with its current supply chain strategy while it monitors the news daily.
Where Tech Transfers Fall Down
Tech transfer is one of the challenges Steven hears about most, and in his view, most failures trace back to one root cause: a weak transfer package that ignores scale-relevant variables.
A process developed at the kilogram scale may be transferred to a vendor running it at 100 kilograms, a hundredfold jump. A good development partner designs the process with that next scale already in mind. “Where it goes wrong is the failure to understand the scale-relevant variables,” Steven says.
He offers an example. In the lab, a solution might be added to a reaction in five minutes to hold temperature. At 100-kilogram scale, adding gallons of that solution could take hours, producing a completely different reaction profile. The fix is to anticipate it early, deliberately slowing that addition on small scale to mirror what will happen later. The same foresight applies to mixing, heating, raw material quality, and setting specifications for raw materials, intermediates, and impurities early. “It can be very painstaking to get it right, but the sooner you get it right, it’ll pay off in spades down the road.”
How to Vet a CDMO
When assessing a new CMO, Steven starts with the regulatory record. Has the FDA or a foreign regulator inspected the site? Have they received any Form 483s? He treats an on-site visit as non-negotiable. “It may cost you five or ten thousand dollars to travel overseas, but that is such a small amount considering the implications of getting that decision wrong.”
Beyond the paper trail, he assesses the technical skill of the process chemists who will actually run the project, and something simpler: responsiveness. How quickly do they respond? Do they understand the nuances of the chemistry you are bringing them? He also leans heavily on his network. Many former Merck colleagues now run or lead CDMOs, bringing with them the same shared training and rigor, which gives him built-in confidence from the outset.
That last point is a real disadvantage for early-stage biotechs. As PharmaSource Editor, and podcast host, Dora notes from PharmaSource’s own research that a company just starting its first molecule often has no such network to draw on, making partner selection far harder than it is for established players.
Process First, Regulatory Second
Steven pushes back gently on the common advice to design everything around regulatory strategy from day one. His framing is to fix the process, then solve the filing.
“You don’t want the tail wagging the dog,” he says. “It’s process first, and then the rest can be easily figured out.” Early in development, regulatory scrutiny is lower, so doing what is right for the program tends to take care of the filing on its own. Later, the burden of a post-approval change is far higher in time, cost, and risk, so the payoff has to justify it.
The principle he carried from Merck is that the company was always willing to change a process for the better, no matter the stage. “In either case, early or late stage, there should be a way to figure it out,” he says. The cost is effort, not feasibility.
Why Design of Experiments Should be Standard
Steven is a vocal advocate for Design of Experiments (DOE), a form of multivariate analysis that varies many process parameters at once rather than the traditional one-factor-at-a-time approach. Its biggest advantage is detecting interactions between variables, where the best setting for one factor depends on another, information the old method simply cannot unearth.
The math behind DOE dates to the 1930s, yet uptake in the industry has been slow, gaining real ground only in the last five to ten years. Steven brought the capability into Merck’s process research group in 2007 after flying to London for a three-day course co-taught by a process chemist and a statistician.
The payoff came almost immediately. In his first week back, a colleague brought him a stubborn post-approval problem: a reaction running at just 77% selectivity, with 23% going to a byproduct that was hard to separate and forced the team to discard around 30% of output. “I gave him a set of reactions to run, having created a design of experiments. Two weeks later he came back with the data, I put it into the software, and we had solved the problem almost immediately.” Selectivity climbed from 77% to over 99.9%, delivering the targeted 20% cost reduction on an already-approved product.
Invest in the Commercial Process Early
Because roughly 80% of molecules fail, the old logic was to defer serious process development until a candidate had proven itself. Steven believes that calculus has changed. High-throughput screening, AI, machine learning, and robotics have lowered the barrier to getting the process right the first time, and, crucially, CDMOs and CROs have invested in the same technologies.
The result is that small pharma now has access to tools that were the domain of big pharma in 2005. “As the technology advances, one can invest in the commercial-type process early on, and that will save you time and regulatory burden down the road if you have the right process early,” Steven says.
The Value of a Good Project Manager
Asked what he wishes someone had told him earlier about external partners, Steven doesn’t hesitate: the value of a strong project manager, especially at the CDMO.
“They really are the touch point,” he says. He has requested changes in PMs when the fit was wrong, and advises assessing the bench strength of a vendor’s project managers during due diligence. That means visiting vendors once or twice a year to build camaraderie and trust, which pays off whenever process development hits a snag.
His parting lesson is the conviction that opened the conversation. Too many companies throw the tech package over the fence and treat it as the vendor’s problem. “No one is going to care about your project as much as you. You need to be constantly involved and always driving the research from the sponsor side as much as possible.”
