Why discovery-stage compounds fail later and how WuXi AppTec says it can reduce the risk

WuXi AppTec says many discovery-stage compounds fail later because potency alone does not prove a molecule can reach the right tissue, work at tolerated exposures or be manufactured consistently. The company argues that earlier DMPK, toxicology, analytical, formulation and process input can expose liabilities sooner and help teams advance stronger IND candidates.

Why it matters: - Discovery teams can spend heavily on a molecule that looks strong in early assays but still fails in development because exposure, safety, formulation or manufacturability problems were not addressed soon enough. - The practical risk is not only scientific. Late-stage failures can consume time, budget and candidate slots before a program reaches IND-enabling work. - WuXi AppTec says earlier cross-functional input can help teams stop weaker series sooner and improve the odds that a candidate can advance with a workable development plan.

What happened: - WuXi AppTec published an explainer on why promising discovery-stage compounds fail later in development and how earlier integration across research and development functions can reduce that risk. - The article says medicinal chemistry programs at WuXi AppTec increasingly bring biology, pharmacology and DMPK/ADME data into lead optimization earlier. - Tao Guo, Ph.D., senior vice president of Research Chemistry Services, Integrated Program Management at WuXi AppTec, framed translatability as a central early-discovery challenge.

The details: - Potency against an isolated target does not prove a compound can produce the intended effect in human tissue at a tolerated exposure. - Key selection questions include whether a compound can reach the relevant tissue, engage its target and sustain the desired effect. - Candidate selection should weigh potency and selectivity alongside physicochemical properties, ADME, PK/PD and early safety signals. - Cell permeability, target abundance, protein turnover and tissue environment can change how well an assay result predicts biological activity. - No single experimental model can establish clinical predictability. - Complementary models are most useful when each one is matched to a defined question and its limitations are understood. - Earlier development and manufacturing input can surface formulation, analytical and scale-up liabilities while design changes are still possible. - The article points to structurally complex small molecules, targeted protein degraders, peptides, oligonucleotides and conjugates as especially challenging classes. - These modalities can bring distinct requirements for solubility, stability, delivery, bioanalysis, purification and manufacturing. - Sequential handoffs from discovery to development can delay recognition of problems until after substantial resources have been spent. - Bringing DMPK, toxicology, analytical chemistry, formulation, process chemistry and manufacturing into lead optimization can help teams address liabilities before candidate nomination. - An integrated approach can translate a target product profile into selection criteria, define evidence needed at each decision point and align specialists around conflicting findings.

Between the lines: - The article’s core argument is that “better science earlier” matters most when it changes decisions, not just when it adds data. - WuXi AppTec is positioning integrated CRO/CDMO-style support as a way to convert separate discipline outputs into one development path. - The company is also careful to note that integration helps resolve avoidable liabilities, but does not by itself prove higher clinical success rates.

What's next: - The article says teams should use integrated findings to discontinue weaker chemical series earlier and optimize promising candidates with clearer expectations for dose, formulation and manufacturing. - WuXi AppTec points to two program examples as proof points for how that workflow can support IND-enabling development. - In a molecular glue degrader program, conventional target occupancy assays could not reliably measure intracellular target degradation. - WuXi AppTec’s Bioanalytical Services team created a customized assay by screening reagents, optimizing assay conditions and standardizing the workflow. - The resulting data were intended to be read alongside DMPK findings to clarify exposure and biological activity and support further IND-enabling development. - In a separate siRNA–lipid conjugate program, the original route depended on customized lipid-functionalized support, slow procurement, inefficient coupling and difficult purification. - WuXi TIDES redesigned the route so the oligonucleotide was made first and then conjugated to the lipid in solution. - The revised process achieved conversion of up to 95% and reduced two purification steps to one. - Lipid synthesis, analytical development, formulation and manufacturing moved in parallel, and the required CMC activities were completed within eight months to support IND submission.

The bottom line: - WuXi AppTec’s message is simple: potency is a starting point, not a development plan, and the earlier teams test translatability, manufacturability and safety together, the fewer surprises they are likely to face later.