Discovery Decisions Are Already Clinical Decisions
The boundary between discovery and clinical development has always been artificial, and in today’s drug pipeline it can be expensive. Decisions made at the earliest stage of research shape whether a program ever reaches patients with a credible chance of success, yet many teams still optimize first for preclinical performance and only later ask whether the candidate can actually translate to human benefit.
Julie Frearson, global executive lead for discovery sciences at IQVIA Laboratories, and David Fischer, the company’s chief translational officer for discovery sciences, framed that disconnect as a translational gap. When discovery runs on a separate track from clinical thinking, programs can drift toward targets with weak evidence in human disease, models that do not reflect real biology, and endpoints that are chosen too late to guide development.
The practical consequence is simple: a molecule can look persuasive in the lab and still fail to map onto the patients who need treatment most.
Patient Benefit Has to Shape the Target, the Modality and the Assays
Frearson’s argument is not that discovery science should become less ambitious. It is that ambition has to be aimed at a clearer clinical destination. Her framing centers on a more disciplined target product profile, defined earlier, with the unmet need, patient segment and likely route to benefit identified before the program becomes too invested in a single experimental path.
Fischer extended that logic to modality selection. Drug developers no longer work in a world where small molecules are the default answer. Antibodies, bi-specifics, ADCs, oligonucleotides, gene and cell therapies, lipid nanoparticles and mRNA have widened the set of possible tools, but they have also raised the stakes for choosing the right mechanism for the right patient population. A modality that is elegant in theory can become impractical if it cannot reach the relevant tissue, cannot be dosed often enough, or carries liabilities that become unacceptable in the clinic.
The underlying pharmacologic question is not just whether a molecule works. It is whether it works in the body that will receive it.
Biomarkers, Safety Readouts and Human-Relevant Models Belong Upstream
A recurring theme in the discussion was the use of biomarkers earlier in discovery. Fischer divided those markers into three broad categories: target engagement, predictive efficacy and safety. In his view, those readouts should not wait for late-stage development. They should be incorporated into cellular assays while molecules are still being optimized, so teams can see whether potency comes with the right translational signal and whether improvements in one direction introduce liabilities in another.
That matters especially for target families where a shift in selectivity can turn an apparently effective drug into one that causes harm. Early assay systems can reveal when a candidate is drifting toward an unsafe profile long before a clinical trial exposes the problem in patients. The same principle applies to dose selection and pharmacokinetics. If a program can measure a biomarker that tracks with human response, developers gain a more credible basis for anticipating PK/PD behavior and for narrowing dose ranges before first-in-human studies.
Human-relevant systems also play a larger role in patient stratification. Primary cells and stem cell-derived models can represent different disease stages or pathways, helping teams understand whether a molecule is likely to perform differently across subgroups. In a field where heterogeneity often defines response, that kind of early segmentation can shape who gets studied first and how a development program is designed.
Human Cell Models Can Expose Liability Before the Clinic Does
One of the most concrete examples discussed involved highly human-selective drugs, particularly antibodies and cell therapies. These agents often cannot be understood fully in animals, because the biology they are designed to engage is human-specific. To reduce that blind spot, IQVIA Laboratories has used dozens of human primary and induced pluripotent stem cell-derived models spanning the heart, kidney, skin, lung, brain, liver and vasculature.
The logic is straightforward. Oncology and autoimmune candidates are supposed to act on cancer cells or on defined immune populations, not on healthy organs. When those same agents show activity in non-target human cells, the result can be a safety signal that changes the development path before an investigational new drug filing. Fischer said clients have used safety data from those human in vitro screens to support IND submissions that allowed them to begin clinical trials.
This is where discovery becomes more than a screen for potency. It becomes an early map of biological liability. A candidate that preserves target activity while avoiding heart, liver or kidney effects has a better chance of surviving the transition from concept to clinic.
Organizational Design Can Either Reinforce or Reduce Translational Risk
Frearson argued that integration is not a single operational fix. Different organizations may need different structures, from embedding clinical and translational specialists directly in project teams to building a cross-portfolio translational medicine function. Another option is governance organized around the target and the patient, rather than around the traditional split between preclinical and clinical silos.
This is a subtle but consequential shift. When governance is built around patient benefit, everyone involved is responsible for the same question: does this program have a believable route to help a defined population? That framing affects experiment design, assay choice, go/no-go criteria and the timing of biomarker work. It also forces teams to confront failure earlier, before the cost of a weak program becomes embedded in later-stage development.
For IQVIA Laboratories, the proposed model combines clinical and translational expertise with human-relevant science and discovery workflows so candidate selection can be judged on human evidence from the outset.
Lower Failure Rates Are the Economic Argument Behind the Biology
The economic case followed naturally from the scientific one. Drug development cost is driven not only by the programs that succeed, but by the programs that fail after years of work. If discovery can reduce safety liabilities, identify the most responsive patient population and establish measurable biomarkers earlier, fewer molecules should advance on false premises.
That would not make development cheap. It would make it more selective.
Frearson and Fischer both emphasized decision quality over volume. The goal is not to run more discovery programs. It is to create decision-grade programs with a better chance of reaching patients safely and effectively. In their view, that begins with human insight at the point where a target is first chosen, a modality is first matched to the disease, and a biomarker plan is first drafted.
The unresolved issue is not whether translational thinking belongs in discovery. It already does. The harder question is how many companies are willing to redesign governance, assay strategy and candidate prioritization around that reality before the next round of expensive failures locks in.