The difference between a successful oncology drug and a failed clinical program often comes down to a single decision made years before regulatory submission. How do you design your biomarker strategy?
With increasingly complex immuno-oncology combinations, and regulatory agency demands for more sophisticated dose optimization under Project Optimus, sponsors face important decisions. Should you enroll only biomarker-positive patients and risk a narrow label? Cast a wider net with an all-comers approach and potentially dilute your signal? Or leverage innovative basket designs that could unlock multiple indications simultaneously?
Now a cornerstone of oncology drug development, biomarkers can support patient selection (e.g., enrichment, stratification, subgroups), serve as prognostic or predictive indicators, and as pharmacodynamic endpoints. Recent studies are also assessing their potential to guide treatment and predict disease recurrence.
We are entering a new wave of targeted therapies in immuno-oncology, where established immune checkpoint inhibitors are increasingly combined with novel immunomodulators. Biomarkers such as PD-L1, Microsatellite Instability-High (MSI-H), or Tumor Mutational Burden (TMB) are used to identify tumors that are more likely to respond to these combinations. Understanding the fundamental biomarker trial design types is essential to designing a program level biomarker strategy that accounts for these complex interactions.
In early-phase studies exploring new pathways, indications, and combinations, the role of biomarkers is often uncertain, making it critical to understand different biomarker trial designs and the impact of assay selection on data precision and accuracy.
Let's break down the four core approaches:
Design: Enroll and randomize only biomarker-positive participants.
Use: Predictive biomarkers to demonstrate treatment effect in the biomarker positive cohort.
Example: EGFR mutations (e.g., T790m) in NSCLC (evaluated in tissue/plasma at enrollment).
If you're weighing enrichment vs. broader designs, align your biomarker strategy early. Coordinate across clinical operations, assay development, and data science to keep sample logistics and endpoints coherent.
Design: Enroll all-comers; randomize within biomarker (+/–) subgroups.
Use: Prognostic biomarkers to isolate treatment effect (removes confounding).
Example: PD-L1 in NSCLC (IHC); where baseline expression correlates with prognosis.
Plan your earlier phase trials to assess the activity of your biomarker: is it prognostic, predictive, or both? This information is essential when designing your pivotal study, and may support the use of a stratified randomization approach
Design: Enroll both biomarker + and – without stratification; assess biomarker effect retrospectively (e.g., subgroup analysis).
Use: Hypothesis generation for future studies.
Examples: DNA alterations, TMB, MSI, circulating tumor DNA (ctDNA).
Begin developing your biomarker strategy from day one of the program. Consider the use of adaptive designs that allow for expansion into promising subgroups based on an early positive biomarker signal. If more than one biomarker is expected to be active (for example, with a combination therapy), plan for the potential impact this will have on sample size if interactions need to be studied.
Design: Patients with biomarker-positive tumors from different cancer types are enrolled into separate study arms. These trials are often non-comparable but may use Bayesian methods to share information across cohorts, enhancing statistical efficiency when appropriate.
Use: Therapies that are tumor-agnostic with strong predictive/prognostic biomarker.
Examples: Assessed efficacy across non-small cell lung cancer (NSCLC), colorectal cancer, thyroid cancer, and other cancer types for patients with BRAF V600 mutation.1
If you're considering a basket approach, look for a partner with direct, hands-on experience designing adaptive, biomarker-intensive programs, not just theoretical knowledge.
If you are planning to transition your Clinical Trial Application (CTA) to include a companion diagnostic, be sure to engage assay development teams and regulatory authorities early, as timelines, sample handling and validation requirements are critical to success.
Choosing the correct assay is often the difference between success and failure. Sponsors carefully evaluate whether tissue or liquid biopsy best matches the clinical question, ensure that assay cut-points are aligned with biomarker-drug response, and confirm that sample integrity can be maintained, especially if the assay will later be transitioned to a companion diagnostic. These decisions directly impact patient enrollment, statistical validity, and regulatory acceptance.
From a sponsor perspective, the right questions can save months of time.
Early answers to these questions can help ensure that critical trial design decisions are made proactively to avoid unexpected issues later in development.
Clinical programs often falter not because the science is flawed, but because the execution wasn't properly aligned. Biomarkers that appear promising on paper may prove unreliable when assays are scaled across trial sites. Subgroup analyses may be underpowered, leading to ambiguous results. Operational breakdowns, such as fragmented sample handling or delayed regulatory engagement, can further compromise trial integrity. These challenges are avoidable with early planning, cross-functional coordination, and strategic foresight.
At Precision for Medicine, our biostatistics teams support sponsors both in full-service programs and through flexible FSP models. What differentiates us is the ability to connect trial design with biomarker science and operational detail. We've guided basket and umbrella studies across multiple tumor types, integrating data science, lab expertise, and clinical operations to keep trials on track. This integrated approach ensures biomarker-driven trials are theoretically sound and practically executable.
Biomarker-driven trial designs enable true precision in oncology, but success relies on more than statistics. It requires validated assays, operational foresight, and regulatory collaboration. Innovation is reshaping biomarker-driven oncology trials, with technologies like liquid biopsy, AI-assisted analysis, and adaptive protocols evolving from optional enhancements to essential components for trial success.