In a meta-analysis from Australia, Dr. Kiely and colleagues analyzed first-line chemotherapy randomized controlled trials involving patients with metastatic breast cancer, published between 1999 and 2009. They looked at disease and patient factors, as well as parameters of survival curves including overall survival (OS) and progression-free survival (PFS) from each trial. From each curve, they extracted the following percentiles: 90th (worst case), 75th (lower typical), 25th (upper typical), and 10th (best case). The results showed that the mean median PFS was 7.5 months and mean median OS was 21.5 months, with a ratio of median OS to PFS of approximately 3. Means for each OS scenario were as follows:
Worst case (10th percentile), 6.25 months;
Lower typical (25th percentile), 12 months;
Upper typical (75th percentile), 36 months; and
Best (90th percentile), 56 months.
Simple multiples of the median gave accurate estimates in excess of 95% of the OS curves for all scenarios, except for the worst-case scenario, for which simple multiples of the median gave accurate estimates in 73%.
Estimating survival probabilities for metastatic malignancies is a challenge for physicians. Considering the heterogeneity of disease and patient factors, it is desirable to approximate a potential prognosis not only to inform a patient and her loved ones but also to help guide therapeutic decisions.
This elegant analysis for patients eligible for first-line therapy for metastatic breast cancer likely included patients of a higher performance status and lower tumor burden than the average patient with metastatic breast cancer. In addition, many of the patients who were part of the meta-analysis had not received the community standard adjuvant therapies offered to patients today after initial diagnosis. As such, the estimates for survival from this analysis would likely be higher than we would expect in the average population of patients with metastatic breast cancer presenting in the community today.
Being able to provide this information as well as worst-case, median, and best-case scenarios to patients and their family members is important to help them plan for the future and make choices for treatment and supportive care.
I hope that these and other investigators will work to derive similar information for clinicians to use in other palliative situations such as advanced lung cancer, glioblastoma, advanced pancreatic cancer, and other metastatic situations.