T TNBC Atlas

For researchers & clinicians

Synthesis: Biomarker-stratified trial design

Biomarker-stratified clinical trial design has become essential to TNBC drug development because the actionable subgroups (PD-L1+, BRCA-mutant, HER2-low) require trial designs that can detect treatment effects within these subgroups rather than diluting the signal across all-comer populations. This page covers the principal design approaches — enrichment, stratification, basket, umbrella, master protocols — and the statistical and operational considerations. The history includes both successes (KEYNOTE-355 hierarchical biomarker analysis, OlympiAD BRCA enrichment) and cautionary tales (IMpassion130's SP142 IC ≥ 1% biomarker that didn't replicate in IMpassion131).

Evidence grades (GRADE-adapted): A high — multiple well-conducted RCTs or systematic reviews converge. B moderate — single pivotal RCT or consistent observational evidence. C limited — single observational study, mechanistic, or expert consensus. D preclinical / hypothesis-generating.

The biomarker-stratification trade-off

Modern targeted therapies often have substantial benefit in biomarker-positive subgroups and minimal benefit in biomarker-negative subgroups. Trial design must balance:

Each approach has been used successfully in TNBC, and each has produced failures. The right choice depends on the biological hypothesis, biomarker characterization quality, and the size of the expected treatment effect in biomarker-positive vs negative patients.

Enrichment trial design

Enrichment trials enroll only biomarker-positive patients. Examples in TNBC:

Advantages: simpler design, statistical power concentrated on the biomarker-positive subgroup, smaller required trial size. Limitations: no evidence about biomarker-negative patients; the biomarker definition must be operationalized correctly at the trial outset.

Stratification with hierarchical analysis

Stratification trials enroll all-comers but pre-specify hierarchical analyses by biomarker subgroup. Examples in TNBC:

Advantages: broader evidence base; the biomarker can be evaluated post hoc against alternative definitions. Limitations: larger trial size; risk of biomarker subgroup being underpowered if the hypothesis is wrong.

Basket trials

Basket trials enroll patients with a specific biomarker across multiple cancer histologies (e.g., NTRK fusion patients across all tumor types). The basic question: does this targeted therapy work in this biomarker subset regardless of histology? Examples relevant to TNBC:

In TNBC specifically, basket-trial-supported approvals provide treatment options for the small biomarker-positive subsets that wouldn't have been adequately tested in TNBC-only trials.

Umbrella trials

Umbrella trials enroll patients with a single tumor histology and assign to treatment based on biomarker profile within that histology. The basic question: which biomarker-matched targeted therapy works best for each subset of this disease? Examples in TNBC:

Master protocols

Master protocols are over-arching trial frameworks that can include basket, umbrella, or platform designs under a unified administrative structure. The advantage: shared infrastructure, statistical methodology, and regulatory documentation across multiple investigational sub-studies. NCI-MATCH was an early oncology master protocol; multiple commercial sponsors are now developing master protocols for specific tumor types or indications.

The IMpassion130 / IMpassion131 cautionary tale

The atezolizumab program in metastatic TNBC illustrates the risks of biomarker-stratified design:

The lessons: biomarker-stratified subgroup analyses can produce false-positive results that fail to replicate; chemotherapy-backbone choices interact with biomarker analyses in ways that pre-specification can miss; confirmatory trial requirements that the FDA imposed worked exactly as intended in protecting patients from a treatment whose benefit was uncertain.

Statistical considerations

Biomarker-stratified design requires careful statistical methodology:

Evidence table

Trial Design Biomarker Outcome
OlympiAD Enrichment Germline BRCA1/2 Positive; approved
EMBRACA Enrichment Germline BRCA1/2 Positive; approved
DESTINY-Breast04 Enrichment HER2-low Positive; approved
KEYNOTE-355 Stratification + hierarchy PD-L1 CPS Positive at CPS ≥ 10; approved
IMpassion130 Stratification + hierarchy PD-L1 SP142 IC Positive subgroup; later withdrawn
IMpassion131 Same biomarker, different chemo backbone Same as IMpassion130 Negative; withdrawal trigger
Larotrectinib NTRK basket Basket NTRK fusion Positive; tissue-agnostic approval
FUTURE series Umbrella FUSCC subtype Higher response than historical controls

Open questions and active investigation


For specific biomarker-stratified TNBC trials, see the first-line metastatic synthesis (KEYNOTE-355, OlympiAD, EMBRACA), the DESTINY-Breast04 synthesis, and the IMpassion/KEYNOTE umbrella synthesis. For pCR endpoint considerations relevant to enrichment-trial readouts, see the endpoint design synthesis.

References

Each citation links to the original publication via DOI. The same records are searchable in the evidence library by title or DOI.

  1. Woodcock J, LaVange LM. Master Protocols to Study Multiple Therapies, Multiple Diseases, or Both. N Engl J Med. 2017;377(1):62–70. doi:10.1056/NEJMra1510062.
  2. Park JJH, Siden E, Zoratti MJ, et al. Systematic review of basket trials, umbrella trials, and platform trials. Trials. 2019;20(1):572. doi:10.1186/s13063-019-3664-1.

Last reviewed: 2026-06-04. Researcher-layer synthesis page. Evidence grades follow the GRADE-adapted rubric defined at the top of this page.