Abstract PO5-25-11: Appropriate use of interventional cohort studies with threshold-crossing designs for de-escalation trials in breast cancer

Abstract For some populations of patients with early breast cancer (BC), clinical outcomes have improved such that the focus has shifted to identifying aspects of therapy that can be reduced or omitted to spare patients toxicity without compromising outcomes. For example, in patients with HER2+ disease, introduction of HER2-targetted therapy significantly reduced event rates but at increased exposure to toxicity. For those with a pathological complete response (pCR) to neo-adjuvant anti-HER2 therapy, the absolute gain from continued adjuvant therapy is likely to be small and the event rate sufficiently low such that any deviation from the expected prognosis would be sufficient to infer failure of a reduced treatment strategy. Previous studies would usually adopt a non-inferiority (NI) trial randomising patients between standard of care (SOC) and reduced therapy. The reduced therapy would be declared non-inferior if the difference between the groups does not exceed a pre-agreed NI margin. Some populations, such as those described, have a very low rate of recurrence and thus large sample sizes are required to gain sufficient power resulting in several issues. In these situations, threshold-crossing designs, also known as interventional cohort studies (ICS), provide an alternative to traditional NI trials. In ICS, a single cohort is recruited and receives the reduced therapy. The event rate is compared to a pre-defined fixed threshold, above which recurrence rates would be considered unacceptable. When conducted appropriately, such studies are statistically and logistically efficient, and should be considered definitive. ICS can be used when 1. A stable background of SOC exists with robust data from which an appropriate crossing-boundary can be defined. Critically, the historical control population used to define the boundary must closely reflect the study population and those seen in clinical practice. 2. The event rate is sufficiently low and consequently has low variability, so a concurrent control group is not required because an accurate estimate of the event rate with SOC already exists. 3. The primary endpoint of interest is reliable, sensitive to treatment and timely to allow early stopping if interim analyses indicate an issue. The well-defined excellent prognosis of patients with HER2+ BC who have a pCR to contemporary neo-adjuvant therapy, coupled with the pattern of relapse rates dropping off rapidly after 3 years, makes this an appropriate setting for implementing such designs. Three recently initiated studies testing reduced therapy in this patient population, HER2-RADiCAL (NIHR131362), CompassHER2 (NCT04266249) and DECRESCENDO (NCT04675827), all have an ICS design. An ICS approach is also valid for trials testing avoidance of radiotherapy in very low-risk populations, where the absolute benefit is likely low and occurrence of late local events is salvageable so as not to impact on overall survival, such as in PRIMETIME (ISRCTN 41579286). In contrast, there are several settings where these designs should not be used. For example, in the assessment of systemic therapy in ER+ HER2- BC the recurrence rate is constant beyond 3 years meaning events cannot be picked up in a timely manner and a fixed threshold defined at 3/5 years would not allow preclusion of a later drop-off in efficacy. It would also be premature to use ICS in TNBC following immunotherapy as outcome data is not yet sufficiently mature to accurately define a suitable threshold. Finally, if relapse rates are not sufficiently low, natural variation in control group incidence rates would prevent setting a suitable fixed threshold. We will discuss the key requirements and design features for using ICS to efficiently evaluate avoidance of treatment in BC and highlight the importance of patient involvement in the development of such trials. Citation Format: Holly Tovey, Laura Finneran, Lucy Kilburn, Iain Macpherson, Andrew Wardley, Hilary Stobart, Lesley Turner, Mairead MacKenzie, Charlotte Coles, Cliona Kirwan, Judith Bliss. Appropriate use of interventional cohort studies with threshold-crossing designs for de-escalation trials in breast cancer [abstract]. In: Proceedings of the 2023 San Antonio Breast Cancer Symposium; 2023 Dec 5-9; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2024;84(9 Suppl):Abstract nr PO5-25-11.