Real-World Data Successfully Integrated as External Control Arm in HER2+ Breast Cancer Trial
核心洞察
Researchers successfully demonstrated the feasibility of using real-world data to create an external control arm for a phase 2 trial of tucatinib plus liposomal doxorubicin in HER2 (搜索)-positive breast cancer patients.
The pilot study expanded an 8-patient phase 2 trial to a 40-patient simulated dataset and matched it with a 77-patient real-world control arm, achieving 82% successful matching.
Baseline characteristics were well-balanced between cohorts, with mean ages of 61.2 and 60.4 years respectively, and similar prior treatment distributions.
A pilot study has demonstrated the successful integration of real-world data as an external control arm in a phase 2 clinical trial, potentially offering a pathway to accelerate drug development in HER2 (搜索)-positive breast cancer. The research, presented at the 2025 ASCO Annual Meeting, showed that investigators could effectively match real-world patients to those enrolled in a phase 2 trial of tucatinib (Tukysa) plus liposomal doxorubicin.
Study Design and Methodology
The pilot study focused on a phase 2 trial (NCT05748834) evaluating tucatinib plus liposomal doxorubicin in patients with HER2 (搜索)-positive locally advanced or metastatic breast cancer. Investigators used real-world proxies for phase 2 eligibility criteria to create a simulated dataset, expanding the original 8-patient enrollment to a 40-patient dataset. The real-world external control arm contained 77 patients.
"One of the achievements of this [research] was that we demonstrated that we were able to do this quickly and in lockstep with the [phase 2] trial as it enrolled," said Jessica Paulus, ScD, senior director of Observational Research at Ontada (搜索).
Patient Matching and Baseline Characteristics
Following propensity score matching, the study achieved an 82% success rate, with 33 out of 40 patients in the simulated cohort successfully matched. The baseline characteristics between the two cohorts showed generally good balance across multiple parameters.
The mean age in the phase 2 cohort was 61.2 years (SD, 10.8) compared with 60.4 years (SD, 11.9) in the real-world cohort (SD, 0.03). Median ages were 61 years (range, 57-64) and 62 years (range, 57-67), respectively.
Prior treatment exposure showed similar distributions between cohorts. Patients who received one prior line of therapy represented 21% of the phase 2 cohort compared with 27% in the real-world patient cohort (SD, 0.09). Notably, 48.5% of patients in both cohorts had prior tucatinib exposure (SD, 0).
Areas of Imbalance
While most baseline characteristics were well-matched, some imbalances were observed. Prior exposure to fam-trastuzumab deruxtecan-nxki (T-DXd; Enhertu) showed rates of 45.5% in the phase 2 cohort compared with 33.3% in the real-world cohort.
Paulus noted that balance in baseline characteristic matching was achieved in approximately 4 out of 6 baseline characteristics, with the remaining 2 variables being modestly out of balance. She emphasized that the findings presented were from an interim analysis, and as more patients are added to both the phase 2 study and the real-world arm, these differences are expected to be minimized.
Clinical Development Implications
The successful demonstration of this methodology could have significant implications for future clinical development strategies. Paulus and colleagues believe that proving the validity of this approach in phase 2 studies could show that assembling real-world cohorts offers a way to speed the clinical development pipeline.
"We hope that demonstrating [the validity of] this approach for phase 2 studies [shows] that assembling these real-world cohorts would offer a way to speed the clinical development pipeline," Paulus explained. This approach would enable faster and more evidence-driven decision-making, particularly when proceeding to phase 3 studies.
Operational Advantages
The study highlighted several operational benefits of this approach. Investigators demonstrated they could add patients quickly and simultaneously with the phase 2 trial enrollment using this methodology. While gathering real-world data can be faster than traditional control arms, Paulus noted that significant effort was required for patient matching to ensure proper data elements were extracted and patients were accurately matched to the phase 2 trial criteria.