The use of recursive partitioning analysis grouping in patients with brain metastases from non-small-cell lung cancer.

Background: This study evaluates the use of recursive partitioning analysis (RPA) grouping in an attempt to predict the survival probabilities in patients with brain metastases from non-small-cell lung cancer (NSCLC).

Methods: Seventy-two patients with brain metastases from NSCLC treated with radiation therapy were included in the study. Sixty-three patients were male and nine patients were female. Their median age was 57 years and their median Karnofsky performance status was 70. At the time of brain metastases, there was no evidence of the intrathoracic disease in 27 patients and the extrathoracic disease was limited to the intracranial disease in 42 patients. In accordance with RPA grouping, 12 patients were in Group 1, 24 patients were in Group 2, and 36 patients were in Group 3. Radiation therapy was delivered to the whole brain at a dose of 30 Gy in 10 fractions in most of the patients.

Results: The median survival time was 7 months for Group 1, 5 months for Group 2 and 3 months for Group 3. The survival probability at 1 year was 50% for Group 1, 26% for Group 2 and 14% for Group 3.

Conclusions: This study presents evidence supporting the use of RPA grouping in an attempt to predict the survival probabilities in patients with brain metastases from NSCLC.