Why Clinical Development Programs Fail Before FDA Ever Reviews the DataA Strategic Perspective on Evidence Generation, Regulatory Alignment, and Development Success
Denis Katz, MD
Founder, Salience Clinical LLC
Executive Summary
Over the course of my career, I've participated in and advised development programs across pharmaceuticals, biologics, regenerative medicine, diagnostics, neuroscience and many other therapeutic areas. One pattern keeps showing up regardless of the therapeutic area.
The programs that struggled were rarely the ones with the weakest science. More often, they generated evidence that didn't quite match what regulators, clinicians, and payers needed to make a decision. The programs that succeeded had usually done something harder: they'd stayed aligned with those stakeholders the whole way through. Approval, in my experience, isn't the result of one breakthrough decision. It's the accumulation of thousands of smaller ones and every single one of them can quietly open a gap between what a sponsor is building and what a decision-maker needs to see.
I call this the Development Alignment Gap: the disconnect between the evidence sponsors intend to generate and the evidence decision-makers require. It's one of the most important, and most underappreciated, drivers of development risk. It shows up as protocol amendments. Regulatory delays. Trials that fail not because the drug didn't work, but because the trial couldn't prove it did. Costs that balloon past what the science alone would justify.
The numbers back this up. Across the most recent decade of phase-transition data, only around 28–29% of programs that enter Phase II make it through. The overall odds of a drug reaching approval starting from Phase I now sit near 7–8%. Efficacy issues not toxicity drive a large share of that attrition. Median capitalized R&D cost per approved drug, once you account for failures and cost of capital, now runs north of $700 million, and mean estimates are considerably higher still. Closing the Alignment Gap is one of the few competitive advantages left that doesn't require a bigger budget just better discipline.
The Development Alignment Gap
Successful development isn't really about generating data. It's about generating the right data, for the right audience, at the right time.
Most organizations start with something genuinely promising a strong hypothesis, a compelling mechanism, an innovative platform. The misalignment creeps in later, as that science moves toward regulatory review. An endpoint hits statistical significance without showing meaningful clinical benefit. A study population is broad enough to dilute a real treatment effect into noise. A biomarker that could have guided an early go/no-go decision shows up two years too late to matter. None of this requires anyone to make an obvious mistake. It happens through a series of individually defensible choices that drift, cumulatively, away from what regulators and payers are going to ask.
Sponsors move along their own track scientific innovation, development strategy, evidence generation while regulators, payers, and HTA bodies sit on a parallel track governed by review standards, reimbursement criteria, and adoption decisions. The gap opens wherever the first track gets ahead of, or simply diverges from, the second. Closing it isn't something you do once, near the end. It takes ongoing, iterative contact between the two tracks for the life of the program.
Case Studies: Two Trajectories
The two scenarios below are composites, built from patterns I've seen recur across multiple programs rather than any single client's story.
Scenario 1: The Program That Drifted
The first involved a mid-sized sponsor developing a treatment for a fibrotic disease. Their Phase IIb trial was well powered and hit its primary endpoint a continuous biomarker measuring tissue scarring with strong statistical significance. Walking into the End-of-Phase 2 meeting, the team was confident.
FDA's response wasn't what they expected: the agency had no established basis for believing that biomarker actually tracked with how patients felt or functioned day to day, and without that link, a clean p-value wasn't going to carry the program. The sponsor had generated rigorous evidence. It just wasn't evidence anyone downstream could act on. What followed was roughly 18 months retrofitting a patient-reported outcome measure into the confirmatory study a fix that would have cost a fraction of the time and money if someone had raised the question before the trial was designed around that endpoint in the first place.
Scenario 2: The Program That Aligned
The second sponsor, smaller and working in a rare neuromuscular indication, did something different. Before locking their pivotal trial design, they brought the proposed endpoint, population, and statistical plan to FDA in a Type B meeting and asked directly whether a functional outcome measure under development would be considered adequately validated in that population.
The agency's answer included something the sponsor hadn't fully considered on its own: a baseline severity threshold that would meaningfully sharpen the trial's ability to detect a real effect. The study that resulted was smaller and faster, and it landed on a pivotal readout the agency had effectively already signaled it would accept. Whether that single meeting alone explains the program's eventual approval is hard to say with certainty plenty else had to go right too but the team was unambiguous that it changed the trajectory of everything that came after.
The Problem is Often Not the Molecule
When a program fails, attention turns first to the product. Was the mechanism wrong? Was the target misunderstood? Was the therapy just not effective enough?
Those are fair questions, but they rarely tell the whole story. Plenty of programs spend years on protocol refinement and operational excellence only to discover, late, that stakeholders never agreed on what "meaningful clinical benefit" meant in the first place. Elsewhere, a real signal gets buried in a population that was too heterogeneous to find it, or a biomarker that could have guided enrollment arrives after the enrollment decisions are already made.
Call these what they are: failures of alignment, not failures of science.
Population Alignment
Genetics, imaging, molecular biology, and biomarker science have made one thing clear: most diseases are far more heterogeneous than we used to assume. Enroll too broad a population and a real treatment effect can get diluted past the point of detection not because the drug doesn't work, but because the trial wasn't built to find where it works.
Enrichment strategies help close that gap without engineering an artificially favorable cohort:
? Biomarker-guided enrollment
? Genetic stratification
? Imaging-based characterization
? Disease severity thresholds
? Prior treatment-response profiles
The Goal: The intent isn't to stack the deck toward a positive result. It's to enroll the exact population that can actually answer the clinical question the trial is asking.
Endpoint Alignment
Endpoint selection is one of the highest-consequence decisions in a development program, and it's easy to get subtly wrong. An endpoint can clear every statistical bar in the protocol and still fail to show meaningful clinical benefit. The reverse happens too, more quietly: a real benefit exists, but the endpoint chosen wasn't sensitive enough to catch it.
Strong programs anchor on one question before anything else gets decided: what evidence would convince a regulator, a clinician, and a patient that this therapy meaningfully improves outcomes? That question not precedent from a competitor, not statistical convenience should drive endpoint selection, balanced against clinical relevance, operational feasibility, and direct regulatory feedback.
Regulatory Alignment
Regulatory misalignment is one of the more preventable risks in development, which is exactly why it's so frustrating when it occurs. Sponsors who get this right build regulatory strategy into the program from day one rather than treating it as something to clean up once the science is settled. Every meaningful decision population, endpoints, statistical plan, biomarker strategy, safety monitoring, even manufacturing carries regulatory weight that compounds as the program advances.
The sponsors who do this well tend to show up to regulatory meetings with a specific kind of preparation:
? INTERACT or Pre-IND Meetings: Work best when the team arrives with a sharp, isolated question. Instead of asking for a generic program review, ask: "Does the agency agree this surrogate endpoint, used this way, in this population, could support approval?"
? Type B Meetings (Post-Phase II): Follow the same philosophy. The briefing package should isolate the one or two core assumptions the entire pivotal design rests on and ask the agency directly whether those assumptions hold up under scrutiny.
Treated this way, the meeting becomes a genuine pressure test before the trial design is locked in not a bureaucratic formality on the way to a plan that was already decided.
Alignment Isn't Always Achievable All at Once
It would be convenient to say that enough foresight lets a sponsor satisfy every stakeholder simultaneously. That isn't always true, and a strategic framework does sponsors no favors pretending otherwise. FDA's evidentiary bar and a payer's or HTA body's bar for "meaningful benefit" overlap, but they are not identical.
A regulator might accept a well-validated surrogate that a payer won't reimburse against. A payer's preferred comparator might differ from the one that makes the most sense for a lean registration trial. In situations like these, alignment stops being about finding one design that pleases everyone and becomes about making an informed trade-off.
Sponsors must know ahead of time where regulatory and payer expectations will diverge, deliberately deciding which gap to leave open now and close later usually through a post-approval study or a separate health-economics analysis. This foresight doesn't make the tension disappear, but it ensures it is managed intentionally rather than discovered as a crisis after the pivotal trial reads out.
Evidence Generation as a Strategic Discipline
Rising costs and growing pressure for efficiency have turned evidence generation into a core strategic capability rather than an operational hand-off. The goal isn't just more data; it's data that is decision-relevant addressi
Comments