Nearly half (43%) of the 198 cancer drugs approved by the FDA between 1998 and 2022 rely on molecular biomarker tests to identify patients. The rate of these precision oncology (PO) approvals accelerated sharply after 2017, from an average of one a year from 1998-2017 to an average of eight a year from 2017-2022.1 Regulators are approving PO products, including gene and chimeric antigen receptor (CAR) T-cell therapies, at ever-earlier stages of development to get them to patients faster. For example, first-in-human Phase 1 cancer trials, which in the 1980s yielded a five percent overall response rate, today produce 15-30% response rates, with even higher rates for biomarker-targeted agents.2
Unfortunately, expedited clinical development and regulatory approval have not necessarily translated into expedited utilization of PO for patients. Health Technology Assessment (HTA) agencies, payers (national governments and private insurers), and prescribers still want a precise evidence-based profile of a new product's benefits and risks.
However, the development routes for PO therapies typically leave significant gaps in the scientific evidence that can delay reimbursement and market access. Pharmaceutical sponsors often utilize expedited regulatory pathways, resulting in less comprehensive comparative safety and efficacy data. Pivotal efficacy trials are frequently single-arm and open-label for ethical and practical reasons.3 They may use surrogate endpoints, including biomarkers of response, such as changes in circulating DNA (ctDNA) levels, rather than traditional clinical outcomes.
The explosion of molecular targets has fragmented once-large diagnostic categories, such as non-small cell lung cancer (NSCLC), into ever smaller subgroups defined by the presence or absence of actionable biomarkers such as PD-1, EGFR, ALK, BRAF, KRAS, and NTRK. As a result, randomized clinical trials (RCTs) have become less feasible and relevant.4 Real-world evidence (RWE)—derived from high-quality, real-world data (RWD)—has become an increasingly important means for developers, regulators, and payers to confirm efficacy and safety.
Challenges of collecting and analyzing RWD in precision oncology
Establishing effectiveness through RWD is a methodological challenge. Secondary databases such as electronic health records (EHRs), insurance claims data, and disease registries are heterogeneous and complex to standardize. And there is a lack of well-established statistical guardrails for generating high-quality RWE.
Secondary data sources are uniquely challenging to curate and analyze in precision oncology. RWD is collected through routine clinical care, but precision oncology demands non-standard biomarker and outcome data. For example, cancer patients do not routinely get a next-generation sequencing panel test to determine the genomic and metabolic profiles of their tumors. Some of the most informative EHR data needed for a precision approach—including tumor histology and treatment outcomes—reside in unstructured, scanned documents and notes ancillary to the core EHR record, posing retrieval and curation challenges.5 To accelerate the utilization of PO, these metrics should be incorporated into regular clinical care and the EHR for all cancer patients, irrespective of whether they enroll in a clinical trial.
At Parexel, we have worked with hundreds of sponsors to overcome challenges and develop RWE strategies throughout the product lifecycle to inform clinical development, reduce risk, improve the patient experience, and fulfill regulatory and payer requirements. Here are five ways we have used RWE to support sponsors:
1. Post-marketing studies to precisely quantify benefits and risks
The FDA recently added a boxed warning on all CAR T-cell therapies, underscoring the importance of long-term, real-world studies to quantify the benefit-to-risk ratio of targeted therapies.6 In the United States, the FDA's Risk Evaluation and Mitigation Strategies (REMS) document and post-marketing commitments and requirements (PMCs/PMRs) in the approval letter address post-marketing activities. The EU’s EMA stipulates them by voluntary or mandatory post-authorization safety and effectiveness studies (PASS and PAES). Other regulatory authorities also have post-marketing requirements to evaluate the safety and efficacy of approved treatments.
Collaborating with a sponsor, we recently designed a non-interventional study that fulfilled regulatory requirements for post-marketing safety and effectiveness while gathering data to support future payer requirements. A pharmaceutical sponsor received approval for a precision medicine treating a rare cancer. The condition involves multiple tumor types and necessitates consulting numerous experts and specialists, which complicates patients' diagnosis and treatment pathways. For example, a single patient might visit a neurologist, gynecologist, gastroenterologist, and oncologist before receiving a confirmed diagnosis. Although the product has been approved with a post-marketing requirement, HTA agencies still need to review it for reimbursement, so patients have yet to receive it outside the United States.
Our hybrid approach was to design a traditional prospective study in the United States (where the product was first approved and prescribed) with a plan to augment that data with existing registry sources from another country (where the drug has been approved but not yet reimbursed) to evaluate effectiveness outcomes related to medication exposure.
Epidemiology expertise is integral to designing post-marketing studies such as PASS, PAES, and long-term follow-up (LTFU). These studies can also support product uptake by assessing care pathways, including measuring how unequal access to genetic or metabolomic profiling tests and lags between diagnosis, testing, and treatment impact patient outcomes.
2. Protocol design and endpoint optimization
Molecular and genetic markers identify target populations for PO clinical trials. If an intervention only works in a unique subset and that target population is not tightly defined, a trial can fail to demonstrate efficacy. RWD can be used to identify and fully characterize the target population, including treatment patterns, standard of care (SoC), lines of therapy, and some relevant endpoints.
Enhertu (fam-trastuzumab – deruxtecan-nxki) for breast cancer patients with low HER2 expression offers a classic example of how RWE can inform trial design and benefit patients. After the drug was approved for HER2-positive patients only, an analysis of RWD showed that many patients classified as HER2-negative were HER2-low—and might benefit from treatment. Using RWE to refine the threshold for HER2-low patient selection and select endpoints, the sponsor demonstrated significant improvements in progression-free and overall survival in HER2-low patients.7
Epidemiology studies are used to establish the association between biomarkers and outcomes, especially if either is novel. A challenge is that the relevant molecular markers and endpoints are not often included in many RWD sources, requiring additional validation studies to identify appropriate proxies.
Some pharmaceutical and device companies incorporate RWE as a core component of the overall evidence-generation strategy by planning Phase IV and observational studies early or supplementing a pivotal RCT through the addition of RWD, either as an external comparator arm (ECA) or as a supplement to an existing arm of a trial.
3. External control arms
Sponsors often conduct single-arm studies when a concurrent placebo or SoC arm in a PO clinical trial would be unethical, impractical, or untimely. This creates a challenge for regulators to evaluate a drug's effect without a comparator. Determining whether the observed results are genuine or spurious in small, uncontrolled trials can be difficult. Yet it is critical to differentiate a valid safety signal from one that occurred due to chance or as part of the natural course of disease.
One solution is to generate an ECA based on RWE. An ECA is an innovative process for collecting and analyzing clinical data external to the clinical trial as a comparator. The data could come from traditional RWD, registries, other prospective studies, or other RCTs—as long as the population meets the criteria for the control arm.
The design and conduct of a trial with an ECA should be as rigorous as those of a traditional clinical trial. It must use the external data to explicitly emulate the ideal (hypothetical) randomized trial arm that would answer the same question if circumstances allowed. While the FDA may not routinely accept an RWE-based ECA instead of an RCT to prove clinical efficacy, the agency's viewpoint on ECAs is evolving. Recently, the agency published a draft guidance on the conduct of external control arms.8
An ECA often requires patient-level data from a cohort of patients matched one-to-one with an active trial arm, with clinical endpoints measured the same way in both. A historical control may allow for patient matching, but those included are not contemporaneous with the randomized controlled trial protocol. In PO, historical control arms are problematic because the SoC evolves rapidly, and a historical control may not emulate the control arm of the planned trial.
Meeting the agency's exacting standards for data completeness and consistency demands specialized epidemiologic, biostatistical, and clinical skills.
At Parexel, we have worked with multiple clients to evaluate the feasibility, design, and execution of ECAs. In one project, we generated an ECA for a targeted oncology drug to delineate its contribution to the efficacy of a combination regimen under review by the FDA. We estimated the experimental agent's additional treatment effect using retrospective patient data and a matching methodology that balanced cohorts for nine prespecified prognostic baseline covariates. We also helped a sponsor design a hybrid ECA for a registrational Phase 3 trial of a targeted treatment for recurrent Glioblastoma (rGBM). The design, approved by the FDA, reduced the number of prospective control patients by 2/3.
4. Indication prioritization and target product profiling
The speed with which some PO products progress through clinical development increases the risks and raises the stakes of product advancement decisions. RWE can enable better decision-making by developers.
Recently, a sponsor asked us to create a model that could prioritize cancer indications for its current and future PO asset portfolio. They wanted to use the model independently, so it was built to be transferable. We combined our partner Partex's extensive collection of proprietary, public, and biological IP (including -omics and sequencing data) with our clinical and RWD expertise to provide a unique solution.
Normalizing data is a core challenge. For example, the lack of standardized terms and disease codes in EHR data has hampered the use of RWD in multiple sclerosis (MS). We created an algorithm based on EHR notes and claims data that enabled one of our clients to identify patients for their key MS asset. The HER algorithm was 99% accurate in identifying MS patients.9
Parexel relies on a team of health data strategists, epidemiologists, physicians, and advanced analytics experts to ensure scientific, clinical, and data integrity for quality RWE that can inform development decisions.
Recently, a sponsor asked us to create a model that could prioritize cancer indications for its current and future PO asset portfolio. We combined our partner Partex’s extensive collection of proprietary, public, and biological IP (including -omics and sequencing data) with our clinical and RWD expertise to provide a unique solution.
5. De-risking clinical trial enrollment with registry data
An established disease registry can also provide data about patients unexposed to medications or receiving the SoC, allowing regulators and payers to contextualize clinical trial data. Some sponsors think long-term when working on a pan-solid tumor or pan-biomarker product and proactively look for existing databases that contain patients with the relevant cancers and biomarkers to have data at the time of trial initiation.
Disease registries ideally enroll patients as close to their diagnosis as possible, providing a long-term evaluation of the course of the disease. Patients usually enroll for altruistic reasons because these studies do not provide treatment. However, they may allow patients to enroll in clinical trials later.
At Parexel, we collaborate with several disease registries and patient advocacy groups to accelerate patient identification and enrollment in clinical trials. While there are data-sharing and intellectual property challenges in this space, it is a fast-evolving field in which transparency and collaboration are advancing science.
RWE is the key to success in precision oncology
Regulators understand that RWE may provide essential information that RCTs cannot, but they want it to meet evidentiary standards and be meaningful and defensible. In precision oncology, RWE must be a sustained strategy, not a tactical battle fought at product launch. Sponsors can deploy high-quality RWE throughout the product lifecycle to optimize product planning, analyze the patient care pathway, accelerate clinical trial recruitment, de-risk development decisions, and drive market access. A key benefit is reducing patient and site burden by replacing traditional outcomes data collection with more efficient RWD collection strategies.
Regulators understand that RWE may provide essential information that RCTs cannot, but they want it to meet evidentiary standards and be meaningful and defensible. In precision oncology, RWE must be a sustained strategy, not a tactical battle fought at product launch.
Contributing Expert
