https://meetinglibrary.asco.org/record/160753/abstract
Authors:
Background:Traditional clinical trials in Oncology require radiology assessments to be performed within a rigid timeframe that is burdensome and not indicative of real world practice. Pragmatic clinical trials may incorporate a more flexible schedule of radiographic assessments reflective of usual care patterns. However, there is concern that this could introduce surveillance bias, particularly if the schedules differ systematically between treatment arms. We aimed to understand the impact of flexible scanning schedules on treatment effect estimates.Methods:We conducted simulations to estimate the impact of flexible assessment schedules on determining clinical evidence of progression in metastatic breast cancer (mBC). We investigated two scan assessment schedules, one traditional and one more flexible, and within each schedule we tested ‘window sizes’ of every 12 weeks +/- 1, 2, 3 and 4 weeks. To simulate a worst-case scenario, we set the scan frequencies to be systematically different by arm. Simulation parameters were based on the mBC literature and retrospective electronic health record data. We used a representative range of hazard ratios (HR) and baseline progression-free survival (PFS) rates.Results:Using any of the window sizes in either of the assessment schedules resulted in only a small amount of bias in the treatment effect estimates. Among all HR settings in the simulation, the median bias ranged from -.01 to .10 (-1.0% to 5.3% bias). In an example using a flexible assessment schedule and a PFS of 15 months, we saw a lack of effect of scanning interval on treatment effect estimates (Table).Conclusions:Increased flexibility of radiographic assessment has minimal impact on comparative assessment of PFS in simulated randomized mBC clinical trials. A less rigid assessment schedule in the setting of a pragmatic clinical trial could ease patient and physician burden, and produce study results more generalizable to real world clinical practice, without compromising analytic integrity.
Sources:
ASCO Annual Meeting