Alternative approaches to evaluating the impact of local control on survival

An important aspect of evaluating treatments aimed at local tumor control is to assess the effect of local control on survival. A common approach to this problem is to compare all patients who have failed locally with those who have not using a logrank test. However, this method of analysis does not take into account the effect of time on a patient's local disease status. Since patients in the local failure group must survive long enough to fail, this group may be biased in favor of longer survival times. The Mantel-Byar extension of the logrank test avoids this source of bias by utilizing the data on time to local failure in defining the local control and local failure groups. In this paper, the application of this technique is illustrated using data from recurrent breast cancer patients and contrasted with the results obtained using the standard logrank test.