Early results from a novel pilot study showed that real-world external control arms can be dynamically assembled in parallel with phase 2 clinical trial patient enrollment, potentially accelerating the drug development pipeline for HER2-positive breast cancer in particular, according to Jessica Paulus, ScD.
At the 2025 ASCO Annual Meeting, Paulus and colleagues presented interim data from a pilot study exploring the feasibility of developing a real-world data–based external control arm in parallel with enrollment to an ongoing single-arm phase 2 trial (NCT05748834) investigating tucatinib (Tukysa) plus liposomal doxorubicin in patients with HER2-positive locally advanced or metastatic breast cancer.1 At the time that this interim analysis was conducted, the phase 2 trial had enrolled 8 patients. Using multiple imputation, the investigators expanded this population into a 40-patient simulated dataset based on the distribution of baseline characteristics. This simulated phase 2 trial cohort was then matched with the real-world data external control arm (n = 77). Following propensity score matching, 82% of the simulated phase 2 cohort was successfully matched with the external control cohort. Covariate matching was generally achieved between the baseline characteristics outlined in these 2 cohorts. Specifically, balance was observed regarding mean age (standard deviation [SD], 0.03), number of prior treatments (SD, 0.09), and prior tucatinib exposure (SD, 0).
Notably, the phase 2 trial has a target enrollment of 36 patients, and investigators expect the balance between the cohorts to improve as trial enrollment continues. The phase 2 trial is enrolling patients at least 18 years of age with an ECOG performance score of 0 or 1 who have measurable disease per RECIST 1.1 criteria and have received prior treatment with at least 1 line of HER2-directed therapy for locally advanced or metastatic disease or have relapsed within 6 months of completing adjuvant HER2-directed therapy.2 Overall response rate serves as the primary end point. Key secondary end points include safety and progression-free survival.
“The big-picture reason we’re doing this [research] is in the hopes of demonstrating that this approach can be valid for phase 2 studies and that assembling these kinds of real-world cohorts would offer a way to speed up the clinical development pipeline,” Paulus said in an interview with OncLive®.
During the interview, Paulus discussed the unique properties of investigating the use of an external control arm in relation to phase 2 trials, contextualized the interim findings from the study within the larger planned scope of the research, and emphasized how the findings from this pilot study might ignite positive future changes for drug development.
Paulus is vice president of Real World Research at Ontada in Boston, Massachusetts.
OncLive: What was the rationale for accelerating phase 2 clinical development in the HER2-positive metastatic breast cancer arena with real-world data from an external control arm?
Paulus: At least in the past decade, there’s been a lot of interest in the power of real-world data generally, but also specifically to support the development of external control arms to help aid the conclusions we draw from clinical trials, in particular clinical trials that only have a single arm with no comparator arm. There are often ethical or clinical development reasons why a clinical trial might only have a single arm. Although that’s often ethically reasonable or justifiable, [the lack of] contrast challenges our ability to know whether the experimental therapy is efficacious, safe, etc.
To date, most of the work with external control arms has been done for phase 3 trials. What we sought to do in this investigation was extend the methods for external control arms to phase 2 trials. This was a challenge because phase 2 trials are inherently smaller in sample size than phase 3 trials, but they also tend to be single-arm studies, so the value of a comparator arm is perhaps even more important.
How is this real-world external control arm being incorporated into the phase 2 trial?
We are assembling a real-world, data-derived external control arm for an ongoing phase 2 clinical trial of metastatic breast cancer that’s being sponsored by the Sarah Cannon Research Institute [in Nashville, Tennessee]. The results that we presented at ASCO 2025 were interim, so the trial is not done enrolling. However, we were seeking to dynamically develop a valid external control arm for that phase 2 trial; as that trial enrolls, [we plan to assemble] the external control arm in lockstep.
Our ASCO 2025 data are based on only the first [8] patients enrolled in the clinical trial. Anchoring around those [8] trial patients, we developed a real-world cohort of 36 patients who were matched on approximately 6 baseline characteristics from the trial patients. What we were seeking to do was show comparability between the trial patients and the real-world patients as a justification for the validity of the external control arm patients to serve as a reference point.
How feasible has it been to integrate this external control arm into this phase 2 research?
One of the achievements of this work is that we demonstrated that we could [create this external control arm] quickly and in lockstep with the trial as it enrolls. Work with real-world data can often be quicker compared with trial data, but there is still a lot of effort required on the real-world data side regarding assembling the right medical oncology and epidemiology expertise to ascertain exactly how to match those real-world patients to the trial patients to make sure the right data elements are abstracted. We were able to do that, and we did it quickly.
We achieved great balance [between the cohorts] on approximately 4 out of approximately 6 of the baseline characteristics, with 2 of the variables being modestly out of balance. However, this is intended to be an interim look at the data, as the trial [still needs to] fully enroll 36 patients. As we triple the enrollment on the real-world data side, I expect to see those differences minimized—if they still exist at all at the end of the study.
As the prospective cohort grows, how will the real-world cohort grow with it?
We’re aiming for an approximate 3-to-1 ratio [between the external control arm and the phase 2 trial population]. As the trial [population] gets bigger, the real-world cohort will get bigger. Because of statistical probability sampling, we expect the balance [between the populations] to get better as the sample size gets closer to 100 overall.
If this research continues to demonstrate the feasibility of using this type of real-world data within the context of a phase 2 prospective trial, what might be the future implications of these findings?
This would allow the life sciences sector to hopefully make faster and more evidence-driven decisions about proceeding to phase 3 clinical trials, for example. It would add another piece of evidence to support decision-making along the clinical development pipeline.
References
- Brown-Bickerstaff C, Charles M, Windish H, et al. Accelerating phase 2 clinical development with real-world data (RWD): an external control arm (ECA) pilot in HER2-positive (HER2+) metastatic breast cancer (mBC). J Clin Oncol. 2025;43(suppl 16):11172. doi:10.1200/JCO.2025.43.16_suppl.11172
- Study of tucatinib and doxil in participants with human epidermal growth factor receptor 2 positive (HER2+) metastatic breast cancer. ClinicalTrials.gov. Updated January 31, 2025. Accessed August 26, 2025. https://clinicaltrials.gov/study/NCT05748834