Drug development is often described as “a path where failure is more familiar than success.” In particular, clinical trial design aligned with global standards and communication with overseas regulatory authorities remain among the biggest challenges faced by our companies. To address this need, <The Bio> is launching a new expert column by Hanlim Moon, CEO of MediRama. Dr. Moon is a veteran who has spent decades in global big pharma, experiencing the entire process of oncology drug development and directly navigating the thresholds of leading regulatory agencies. As a physician-scientist and clinical strategist, his deep insights will serve as a guideline for our companies striving to cross the “valley of death” between nonclinical and clinical stages. This monthly column will function as a “practical strategy guide,” covering not only key approaches at each stage of drug development but also analyses of review trends among overseas regulatory authorities. It is highly recommended reading for professionals in the pharmaceutical and biotech industries aiming at the global market.[Editor’s note]
In the era of precision medicine, targeted anticancer therapies are often placed at the center of innovation. Of course, the importance of the drug itself is absolute. However, in reality, it is not the drug but CDx (Companion Diagnostic) that defines the patient population eligible for treatment—those who are expected to benefit while maintaining safety. If a biomarker represents a biological signal, CDx is a regulated medical device that provides essential information for the safe and effective use of a specific therapeutic. This distinction is not merely conceptual. It is a structural variable that determines patient selection strategies, clinical trial design, regulatory pathways, indication scope, and even market size and commercial scalability.
In the early stages of oncology drug development, biomarker definitions are relatively flexible. Exploratory analyses are permitted, assays may be conducted using research-use platforms or in the form of LDTs (Laboratory-Developed Tests), and cutoffs are provisionally defined. The purpose at this stage is hypothesis testing and signal exploration.
However, once the program enters the registrational stage—typically Phase 3 (or sometimes Phase 2)—the situation changes. The assay must demonstrate analytical validity, and a reproducible cutoff that can explain clinical efficacy is required. When used for patient selection, regulatory review, including IDE (Investigational Device Exemption) requirements, becomes fully engaged. At this point, the biomarker is no longer an exploratory variable but transitions into a regulated product that defines the indication. Failure to strategically manage this transition can lead to structural risks in later stages of development, such as assay changes, cutoff redefinition, or additional analysis requirements.
The development of PD-L1 in the era of immuno-oncology provides a representative example of the importance of such strategic alignment. Different clones and platforms, along with varying scoring systems and cutoffs, were developed in parallel, resulting in different patient populations being defined for each drug despite targeting the same marker. PD-L1 is inherently a heterogeneous and dynamic protein; however, regulatory approval required clear dichotomous criteria such as TPS ≥1% or ≥50%. In first-line treatment development for non-small cell lung cancer (NSCLC), MSD chose a strategy of clearly defining the high-expression population (PD-L1 TPS ≥50%) and demonstrating the clinical efficacy of monotherapy in this group. As a result, the KEYNOTE-024 study met its primary endpoint and achieved early approval. In contrast, BMS pursued a broader strategy encompassing a wider PD-L1 expression range, setting PD-L1 ≥5% as the primary analysis population in the CheckMate-026 study, but failed to meet its primary endpoint.
This difference was not merely due to technical variations in antibody clones but stemmed from the level of alignment between patient selection criteria and clinical trial design. The contrast between a strategy that maximized the efficacy signal by restricting to a high-expression population and one that attempted to include a broader patient population clearly demonstrates that CDx is not merely a laboratory technique but a clinical strategy.
Cutoff is also not merely a statistical number. Early Phase 3 trials of FRα-targeted therapies applied a relatively broad criterion of ≥50% tumor cell expression but failed to achieve statistical significance. Subsequently, by raising the cutoff to 75% and redefining a high-expression population, the number of patients decreased, but the efficacy signal became clearer, ultimately leading to approval. In MET overexpression programs, the FDA excluded intermediate-expression populations and recognized only strongly expressing populations as the approvable population. This confirmed that reproducible clinical benefit carries more weight in regulatory decision-making than biological plausibility. The paradox emerges here: while broader definitions make development easier, only restricted definitions enable approval.
In hematologic malignancies, the case of FLT3 illustrates the evolving nature of biomarkers. Initially, it served as a prognostic factor based on allelic ratio, but with the advent of targeted therapies, ITD and TKD mutations transitioned into predictive biomarkers for treatment selection. Early patient selection relied on central PCR-based LDTs, but later, with the introduction of the FDA-approved CDx LeukoStrat® FLT3 Mutation Assay, it became incorporated into the regulatory framework. Furthermore, quizartinib adopted a stratified strategy targeting only FLT3-ITD mutations. This reflects a regulatory preference for clear criteria—such as the presence or absence of a mutation—over complex quantitative models.
Sample strategy is also evolving. While tissue-based testing was once the standard, advances in ctDNA technology have enabled plasma-based diagnostics to be integrated into regulatory frameworks. Key targets such as EGFR, MET exon14, and HER2 activating mutations can now be assessed using both tissue and blood-based approaches. This is not merely a technological advancement but a strategic expansion reflecting clinical realities, including patient accessibility, limitations in tissue acquisition, and the need for rapid decision-making at the time of treatment. CDx is no longer a tool fixed to a specific specimen but is being redefined as a flexible system that selects the optimal sample based on disease characteristics and clinical context.
Regulatory flexibility is also observed in the rare biomarker space. Bizengri (zenocutuzumab), approved in 2025, targets patients with NRG1 fusions, which occur in only 0.1–0.5% of all cancers. Considering this ultra-low prevalence, the FDA allowed additional CDx data to be supplemented for approximately three and a half years post-approval. This does not represent a relaxation of scientific standards but rather a policy decision to rebalance the risk–benefit equation in life-threatening diseases. As biomarkers become rarer, such flexibility becomes an important strategic consideration.
Ultimately, CDx is not a fixed tool but a strategic asset that is designed and refined throughout the entire development process. While biomarker definitions are flexible in early stages, they become fixed at the approval stage, and cutoff, assay, and sample strategies must be aligned between clinical signals and regulatory expectations. Although all answers cannot be known at the outset, continuous monitoring of accumulating clinical data and iterative decision-making are essential. CDx strategy is not a matter of post hoc adjustment but a dynamic structure managed across the full lifecycle. The success or failure of precision oncology development is no longer determined solely by the efficacy of the drug. It depends equally on how patients are defined and how the treatment population is specified through diagnostic systems. While drugs may symbolize innovation, it is CDx that ultimately determines the scope of treatment application. Only companies that understand the strategic evolution of CDx and proactively design it will secure sustainable competitiveness in the era of precision medicine.
