Precision oncology promises tailored therapies based on individual tumor characteristics. Yet the evidentiary pathway from biomarker discovery to clinical adoption remains fraught with gaps. Real-World Evidence (RWE) offers a bridge—connecting the controlled environment of clinical trials to the complex reality of clinical practice.
The Evidentiary Gap in Precision Medicine
The Promise of Biomarkers
The precision oncology paradigm rests on identifying predictive biomarkers that match patients to targeted therapies:
EGFR mutations → EGFR inhibitors in NSCLCHER2 amplification → HER2-targeted therapy in breast cancerBRCA mutations → PARP inhibitors in ovarian cancerThe Evidence Problem
Yet many biomarker-driven hypotheses fail to translate from discovery to clinical benefit:
Biomarkers identified in retrospective studies often fail in prospective trialsClinical trial populations are highly selected, limiting generalizabilityRare biomarker subsets make traditional trial designs impracticalReal-World Evidence: Definition and Sources
What is RWE?
FDA defines RWE as:
> "Clinical evidence about the usage, potential benefits, or risks of a medical product derived from analysis of real-world data (RWD)."
Sources of Real-World Data
RWD can originate from:
Electronic Health Records (EHRs): Clinical encounters, diagnoses, treatmentsClaims data: Billing records, healthcare utilizationRegistries: Disease-specific or product-specific databasesPatient-generated data: PROs, wearables, patient appsApplications in Precision Oncology
Treatment Effectiveness
RWE can demonstrate:
Comparative effectiveness of different treatment sequencesLong-term outcomes beyond trial follow-upEffectiveness in populations underrepresented in trialsBiomarker Validation
RWD enables:
External validation of biomarker performanceAssessment in diverse patient populationsReal-world positive/negative predictive valuesRegulatory Decision-Making
FDA increasingly accepts RWE for:
Post-market commitments: Safety surveillanceLabel expansions: New indications based on RWEConditional approvals: Accelerated access with RWE requirementsMethodological Challenges
Confounding and Selection Bias
Observational RWE is vulnerable to:
Confounding by indication: Sicker patients may receive different treatmentsSelection bias: Treatment choices influenced by unmeasured factors immortal time bias: Time-dependent exposure misclassificationData Quality
RWD often suffers from:
Incomplete data: Missing values, inconsistent recordingMisclassification: Diagnostic codes may not reflect clinical realityLack of standardization: Different systems, different definitionsAnalytical Solutions
Modern approaches address these challenges:
Propensity score methods: Matching, weighting, stratificationInstrumental variables: Addressing unmeasured confoundingTarget trial emulation: Designing observational studies as trialsCase Study: RWE in Oncology
Example: CDK4/6 Inhibitors in HR+ Metastatic Breast Cancer
Clinical trials demonstrated:
Significant improvement in progression-free survivalOverall survival benefit in some studiesConsistent benefit across subgroupsRWE has complemented this evidence by:
Demonstrating effectiveness in older patientsShowing real-world tolerability patternsIdentifying predictors of discontinuationLife-Cycle Approach to Evidence Generation
Pre-Trial Evidence
Biomarker discovery: Genomic profiling in clinical careHypothesis generation: Retrospective RWE studiesTarget identification: Observational studies of treatment patternsDuring Clinical Development
External control arms: Synthetic control arms from RWDPragmatic trials: Embedded designs in routine careInterim analyses: Futility and efficacy assessmentsPost-Market
Phase IV commitments: Safety and effectiveness surveillanceRegistry studies: Long-term follow-up in routine practiceComparative effectiveness: Against standard of careImplications for Biostatisticians
New Skill Requirements
Precision oncology RWE requires:
Causal inference methods: Beyond traditional epidemiologyHigh-dimensional data: Genomic features, EHR-derived variablesBayesian thinking: Prior evidence synthesis, borrowing strengthRegulatory Literacy
Understanding regulatory expectations for RWE:
FDA RWE Framework: Guidance on acceptable methodsEMA perspective: Real-World Data in regulatory decisionsICH harmonization: Emerging global standardsConclusion
The evidentiary gaps in precision oncology are real—but they are bridgeable. RWE offers a pathway from biomarker discovery to clinical impact, connecting the controlled environment of clinical trials to the complex reality of patient care.
Realizing this potential requires:
Methodological rigor: Causal inference, bias mitigation, transparencyData infrastructure: Quality, standardization, interoperabilityRegulatory evolution: Acceptance of RWE as legitimate evidenceCultural shift: Viewing RWE as complement, not competitor, to trialsThe life-cycle approach to evidence generation—integrating RWE from discovery through post-market surveillance—represents the future of precision oncology. Biostatisticians who master these methods will be essential partners in bringing effective therapies to the patients who need them.
---
Key Points
RWE bridges the gap between clinical trials and routine practiceSources include EHRs, claims data, registries, and patient-generated dataMethodological challenges require advanced causal inference methodsRegulatory agencies increasingly accept RWE for decision-makingLife-cycle approach integrates RWE across all phases of drug developmentReferences
FDA. Real-World Evidence: Framework for FDA's Support of Drug Approval. 2023.European Medicines Agency. Data Quality Framework for Real-World Data. 2024.Khozin et al. Real-World Evidence in Oncology. Journal of Clinical Oncology. 2024.4.湘雅医学. Life-Cycle Approach to Biomarker Validation. Nature Medicine. 2026.