Paulus on Leveraging Real-World Control Arms in Future HER2+ Breast Cancer Trials

“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.”

Jessica Paulus, ScD, the senior director of Observational Research at Ontada, discussed the potential clinical implications of using real-world data to add an external control arm to a phase 2 trial (NCT05748834) of tucatinib (Tukysa) plus liposomal doxorubicin in patients with HER2-positive locally advanced or metastatic breast cancer.

Data from a pilot study, which were presented during the 2025 ASCO Annual Meeting showed that investigators were able to create a simulated dataset using real-world proxies for phase 2 eligibility criteria. This expanded the 8 patients enrolled in the phase 2 study to a 40-patient dataset and the real-world control arm enrolled 77 patients. Eighty-two percent of patients in the simulated cohort were successfully matched.

The baseline characteristics of the 2 cohorts were generally well balanced; the mean age in the phase 2 and real-world cohorts was 61.2 years (SD, 10.8) and 60.4 years (SD, 11.9), respectively (SD, 0.03). The median ages were 61 years (range, 57-64) and 62 years (range, 57-67), respectively. Patients in the phase 2 cohort received 1 prior line of therapy at a rate of 21% compared with 27% in the real-world patient cohort (SD, 0.09). Additionally, 48.5% of patients in both cohorts received prior tucatinib (SD, 0). However, there was some imbalance in terms of prior exposure to fam-trastuzumab deruxtecan-nxki (T-DXd; Enhertu), with respective rates of 45.5% and 33.3%.

Paulus and colleagues hope that by demonstrating the validity of this approach in phase 2 studies, the use of real-world control cohorts could accelerate the clinical development pipeline, Paulus explained. This would enable faster and more evidence-driven decision-making, particularly when proceeding to phase 3 studies, she concluded.