Biomarkers in phase 2 clinical trials: patient guide

What is a biomarker in a clinical trial?

A biomarker is a defined and measurable characteristic used as an indicator of normal biological processes, disease processes, or a response to a therapeutic intervention^[1]. This definition is formalised in the FDA-NIH BEST (Biomarkers, EndpointS, and other Tools) framework, which underpins how biomarkers are classified and used across clinical development^[2].

Biomarkers are measured from a variety of sources. These include tumour tissue, whole blood, plasma, serum, bone marrow, urine, or saliva. Some biomarkers measure tumour characteristics, while others measure host characteristics, such as germ-line DNA mutations. Molecular imaging can also serve as a biomarker measurement method^[1].

Quantitative biomarkers give a value on a continuous scale, while qualitative biomarkers give a binary result, either present or absent. A threshold value, also called a cut-off, separates biomarker-positive from biomarker-negative patients. This cut-off is often determined through retrospective analysis of stored samples^[3].

What types of biomarkers are used in clinical trials?

Biomarkers are grouped by the role they play in the trial, generally into prognostic, predictive, and pharmacodynamic biomarkers^[3]. A further category, diagnostic biomarkers, is described in the wider clinical biomarker literature^[2].

• Prognostic biomarkers — give information about the likely course of the disease, regardless of which treatment the patient receives^[3]. They can be used as an enrichment strategy to select patients likely to have a rapidly progressing disease or a particular clinical outcome^[3].
• Predictive biomarkers — give information about how likely a patient is to respond to a specific therapy^[3]. They allow patients to be stratified according to expected response, and can also be used as an enrichment strategy^[3].
• Pharmacodynamic (PD) biomarkers — measure the biological effect of a drug and can give insight into treatment efficacy^[3]. They are mainly used during phase II studies to guide dose selection, in proof-of-concept and dose-ranging studies^[3]. They are rarely used on their own as a basis for regulatory approval^[3].
• Diagnostic biomarkers — help detect the presence or absence of a specific disease and support early, accurate diagnosis^[2].

A biomarker used to identify and recruit a targeted patient population can also be described as a patient stratification biomarker. Current examples include BRAF gene mutations in metastatic melanoma, HER-2 mutations in breast cancer, and ALK gene translocation in non-small-cell lung cancer^[4].

Why do phase II trials rely on biomarkers?

A biomarker in a phase II trial can act as a predictor of response or resistance, a tool for patient enrichment, a correlative endpoint, or a surrogate endpoint. Strategically planned biomarker evaluations in phase II studies may allow for more efficient phase III trial design, and may also improve screening of therapeutics before phase III development begins^[1].

Advances in early phase studies over roughly ten years have focused on collecting more clinical pharmacology data earlier. This informs dose determination, endpoint identification, and patient selection specifically for phase II trials. Phase I investigations already use biomarkers to identify and recruit targeted patient populations. Phase I data is designed to inform phase II endpoint selection and to streamline trials for proof of concept^[4].

The efficiency of a clinical trial can be improved by a protocol that uses biomarkers to stratify patients at enrolment or to monitor treatment response. Biomarkers also allow for sample size reduction by improving patient selection. Using biomarkers is generally easier and less expensive than direct measurement of the final clinical endpoint, and biomarker measurement is generally faster than measuring the final clinical endpoint^[3].

How does biomarker-based patient stratification work in trial design?

Several phase II designs extend Simon's two-stage, single-arm design to accommodate a single binary candidate biomarker. These designs aim to ensure that promising drug activity restricted to "test-positive" patients is not missed. At the same time, they avoid requiring excessive patient numbers when the drug's activity is broad enough that the marker is not needed^[5].

A design described by Freidlin and colleagues uses a single binary biomarker in a randomised phase II trial. It allows researchers to determine whether the drug should move to a phase III enrichment trial, an all-comers trial, or be dropped from further development. More complex settings involve multiple candidate biomarkers or no known cut-point in advance, and these settings generally require much larger sample sizes than a traditional phase II trial. They also require pretreatment tumour tissue from participants, which can make such trials expensive and time-consuming^[5].

If a biomarker test is used to select which patients enter the trial, the test must be clinical-grade and performed in a certified laboratory in real time. An investigational device exemption from a regulator may be required for this purpose^[5].

What did the BATTLE I trial show about adaptive biomarker designs?

The BATTLE I trial in non-small-cell lung cancer (NSCLC) evaluated four different biomarker tests against four different drug regimens. Treatment assignment among the four regimens was randomised, but the randomisation weights changed as the trial progressed, adjusted according to which treatment performed best within each of the four biomarker strata. The trial used freedom from progressive disease at week 8 as the endpoint. A fresh tumour biopsy was required as an entry requirement^[5].

Two objectives drove the adaptive randomisation. The first was to efficiently screen four treatments across four predetermined NSCLC patient strata. The second was to give patients a trial that adapted to assign them the regimen considered best for their form of the disease. Korn and Freidlin have questioned whether such response-adaptive randomisation designs actually reduce the number of patients receiving a less active regimen. It remains unclear whether this approach was more efficient than optimal two-stage designs for each drug-biomarker combination^[5].

What is the I-SPY 2 design in breast cancer?

The I-SPY 2 phase II trial in breast cancer also uses an adaptive design with prespecified biomarker strata and multiple treatments. A key challenge for developing multi-arm adaptive trials that match tumours to drugs is access to a broad menu of drugs that potently inhibit their target pathways^[5].

How are biomarkers selected and validated for a phase II trial?

Selecting the right biomarker for an early clinical study requires balancing scientific validity, operational feasibility, and regulatory requirements. The FDA-NIH BEST framework provides a structured approach to this decision^[2].

• Biological rationale — mechanistic evidence must link the biomarker to the drug's mechanism of action or to the disease process under study^[2].
• Analytical feasibility — the biomarker must be reliably measurable in the available sample type, such as tumour biopsy, blood, FFPE tissue, or liquid biopsy^[2].
• Pre-analytical stability — the biomarker must survive the collection, processing, and shipping conditions of a multisite trial^[2].
• Assay turnaround time — if the biomarker is used for patient stratification, results must return within the enrolment window^[2].
• Regulatory precedent — prior acceptance of this biomarker class in submissions for similar indications strengthens its use^[2].

Not every biomarker assay requires the same depth of validation. The fit-for-purpose principle matches validation stringency to the intended use of the biomarker. Exploratory biomarkers used in phase I for hypothesis generation require only minimum validation, including acceptable precision, basic stability, and absence of gross matrix effects. A biomarker intended to select or exclude patients in a pivotal phase III trial, by contrast, is held to a much higher standard, since an error in that assay directly changes who receives an experimental treatment^[2].

Common biomarker analysis technologies include quantitative polymerase chain reaction (qPCR), enzyme-linked immunosorbent assay (ELISA), flow cytometry, and RNA sequencing (RNA-seq). The choice between them depends on accuracy needs, sample type, and research goals^[2].

Why do promising phase II results often not carry over to phase III?

Between 1978 and 2015, 148 phase II clinical trials for first-line systemic treatment in metastatic pancreatic cancer were conducted and published. Of these, only 15% went on to phase III testing. Only 11% of the 148 trials tested new investigational agents^[6].

Of the 37 trials (25% of the total) that tested biological agents, only 1 was enriched for biomarkers. Seven of these 37 trials included biomarkers only as an exploratory or translational endpoint. The final analysis, covering 148 studies and 7,505 patients, found that reporting of prognostic factors was limited across trials. Baseline factors showed significant heterogeneity between trials, and the proportion of patients with locally advanced disease ranged from 0% to 80% across the trials^[6].

The primary endpoint was defined in 68.9% of the 148 trials. The most common primary endpoint was objective tumour response, used in 41.2% of trials. Progression-free survival was used in 15.5% of trials. Only 10% of trials used overall survival as the primary endpoint^[6].

A separate systematic review of 32 phase III trials in advanced metastatic pancreatic cancer, covering 13,675 patients, found that these trials produced three clinically meaningful treatment regimens. A benchmark of a 50% improvement in overall survival in phase III trials accurately predicted a clinically meaningful phase III trial in 78% of cases^[6].

The National Cancer Institute's Trials Planning Meeting on Pancreas Cancer Treatment has published recommendations for pilot trial design. These recommendations include restricting trials to patients with good performance status, using predictive biomarkers for enrichment, and using survival as the preferred primary endpoint. For the transition to phase III, the recommendations call for considering input from multiple studies and starting phase III testing only after pilot data show robust signals^[6].

How do earlier trial phases feed biomarker data into phase II?

Translational medicine is the discipline that translates laboratory findings into the design of early-stage clinical trials. It uses pre-clinical data from in vivo, in vitro, and in silico research to help design trials, determine methods, and choose biomarkers. Data from clinical studies also feeds back into pre-clinical experiments to improve future drug discovery^[4].

In phase 0 trials, microdoses of an experimental drug are given to volunteers, at a dose expected to be well below toxicity and efficacy thresholds. These microdoses allow an initial assessment of pharmacokinetics (PK) and pharmacodynamics (PD). Positron emission tomography (PET) and accelerated mass spectrometry (AMS) imaging are used to assess drug distribution in phase 0. Phase 0 biomarker evaluations can guide understanding of drug targets and mode of action, making data collection in phase I more selective^[4].

If preclinical research has not identified patient stratification biomarkers before human trials begin, phase I and phase II data may be used to discern them after treatment. Oncology has traditionally been at the forefront of biomarker development and personalised medicine. Fields such as neuroscience and immunological, inflammatory, and metabolic disease research are expanding their use of biomarkers as technology advances^[4].

What phase II biomarker trial designs exist, and what do they require from patients?

Design or trial example	Biomarker role	What is required from patients	Source
Extended Simon two-stage, single-arm design	Single binary predictive biomarker	Pretreatment tumour tissue sample	[5]
Randomised design (Freidlin et al.)	Single binary biomarker used to decide phase III path (enrichment, all-comers, or drop)	Biomarker testing before randomisation	[5]
BATTLE I (non-small-cell lung cancer)	Four biomarker tests linked to four drug regimens, adaptive randomisation	Fresh tumour biopsy required for entry	[5]
I-SPY 2 (breast cancer)	Prespecified biomarker strata, multiple treatments, adaptive design	Biomarker profiling before treatment assignment	[5]

When should a patient or caregiver ask about biomarker testing before joining a trial?

If a trial requires a biomarker test to decide whether you can enrol, ask whether the test is performed in a certified laboratory in real time, since this affects reliability of the result used to include or exclude you^[5]. Ask whether the biomarker is prognostic, predictive, or pharmacodynamic, since each type answers a different question about your disease or your likely response to treatment^[3]. Ask what sample is required (blood draw, tumour biopsy, liquid biopsy) and whether a fresh biopsy is a condition of enrolment, as seen in the BATTLE I design^[5].

Ask which endpoint the trial is measuring, since objective tumour response and progression-free survival do not always reflect overall survival benefit^[6]. Discuss any specific concerns about your test results, their meaning, and how they may affect your treatment options with the trial investigator or your treating clinician before you consent to participate.

Summary

Biomarkers give phase II trials a way to identify likely responders, adjust dosing, and monitor treatment effects earlier than survival data alone would allow^[1]. Trial designs built around biomarkers, from extended Simon two-stage designs to adaptive multi-arm trials such as BATTLE I and I-SPY 2, aim to match patients to the regimen most likely to help them while limiting exposure to ineffective treatment^[5].

At the same time, real-world data from pancreatic cancer trials show that biomarker enrichment has been used in only a small fraction of published phase II studies, and that many phase II results do not translate into phase III success^[6]. Patients and caregivers considering a trial should ask what type of biomarker is being used, what samples are required, and which endpoint the trial is designed to measure^[3].

1. [1] McShane LM, Hunsberger S, Adjei AA. Effective Incorporation of Biomarkers into Phase II Trials. Clin Cancer Res. 2009;15(6):1898-1905. doi:10.1158/1078-0432.CCR-08-2033 (accessed 2 June 2026) — https://pmc.ncbi.nlm.nih.gov/articles/PMC2874890/
2. [2] Biomarkers in clinical trials: Types, selection, and validation (accessed 2 June 2026) — https://ohmx.bio/biomarkers-in-clinical-studies-which-to-choose/
3. [3] Antonini P. Clinical trials, how biomarkers help research (accessed 2 June 2026) — https://www.meditrial.net/2022/09/clinical-trials-how-biomarkers-help-research/
4. [4] Translational Medicine and Biomarkers. Applied Clinical Trials. 2013;22(9). (accessed 2 June 2026) — https://www.appliedclinicaltrialsonline.com/view/translational-medicine-and-biomarkers
5. [5] Simon R, Polley E. Clinical Trials for Precision Oncology Using Next-Generation Sequencing - Phase II Trials. Personalized Medicine. 2013;10(5):485-495. (accessed 2 June 2026) — https://www.medscape.com/viewarticle/808674_3
6. [6] Nelson R. Phase 2 Trials in Pancreatic Cancer: Missing the Mark (accessed 2 June 2026) — https://www.medscape.com/viewarticle/874801