More than just the median: Calculating survival times for patients with HER2 positive, metastatic breast cancer using...

To estimate worst-case, typical and best-case scenarios for survival as a communication aid for managing patients with HER2-positive metastatic breast cancer (MBC) starting HER2-targeted therapies.

We sought randomised trials of HER2-targeted therapies and recorded the following percentiles (representative scenarios) from each OS curve: 90th (worst-case), 75th (lower-typical), 50th (median), 25th (upper-typical) and 10th (best-case). We then tested whether we could estimate these percentiles for each OS curve by multiplying its median by four simple multiples: 0.25 (to derive the 90th percentile), 0.5 (75th), 2 (25th) and 3 (10th). Estimates were deemed accurate if within 0.75-1.33 times the actual value.

We identified 15 trials with 4798 patients. For first-line, single-agent HER2-targeted therapy (15 treatment groups), the median (interquartile range [IQR]) for median OS was 33.3 months (29.1-38.4), and for each percentile was: 90th 9.5 months (7.7-11.0); 75th 19.2 months (16.4-20.8); and 25th 50.6 months (47.1-63.3). The 10th percentile was unavailable for all treatment groups. For first-line dual HER2-targeted therapy (1 treatment group), the median OS was 56.5 months. Simple multiples of the median OS accurately estimated the: 90th percentile in 79%; 75th percentile in 100%; and 25th percentile in 89% of OS curves.

Surprisingly little is known of survival beyond the median for HER2-positive MBC. Longer trial follow-up is required to help clinicians estimate and explain the best-case scenario. Simple multiples of the median OS provide a reasonable framework for estimating then explaining survival times to patients.