Recursive Partitioning Analysis Groups II-III brain metastases of non-small cell lung cancer: a phase II randomized...

To compare the response rates and safety profiles of two investigational chemotherapies that were delivered concurrently with whole-brain radiotherapy in a population of patients with chemonaive non-small cell lung cancer.

Eligible patients included those presenting with brain metastases belonging to the Recursive Partitioning Analysis of prognostic factors from the Radiation Therapy Oncology groups II or III, who were not able to undergo surgery or stereotactic radiotherapy. Other main eligibility criteria were age <75 years and Eastern Cooperative Oncology Group performance status = 0 to 2. The study design was as follows: all patients received whole-brain radiotherapy in three split courses of 18 gy/10 fractions. They were randomly (1:1) assigned to regimen A, consisting of a triplet cisplatin-vinorelbine-ifosfomide, or to regimen B, consisting of high-dose single-drug ifosfamide. In both groups, chemotherapy was delivered on a 4-week cycle, for three courses. Each course was delivered concurrently with radiotherapy. Brain and other tumor lesion assessments were performed in both groups at the end of the three courses (RECIST). Neurologic symptoms were evaluated quantitatively at each step of the treatment program. All analyses were carried out in an intention to treat basis, and statistical tests were two sided.

Seventy patients were randomly allocated into groups A (n = 37) and B (n = 33). With regards to the whole lesions, overall response rates did not significantly differ between the groups (group A: 45.9%; group B: 33.3%; chi(2); p = 0.28). When brain-target lesions were separately analyzed, respective response rates were 59.5% and 48.5%; (chi(2); p = 0.0.53). Febrile neutropenia was more frequently observed in the former group (n = 19, 54.29%) than in the latter (n = 12, 36.36%; p = 0.13), and a similar difference was also observed regarding documented infections. Red blood cell transfusions and readmission for antibiotic infusions significantly affected more patients in group A than in group B. The longitudinal evaluation of neurologic symptoms (by means of Generalized Estimating Equation) slightly improved during the treatment program, and there was no difference between the groups. Median overall survival did not significantly differ between the two groups (months [95% confidence interval], 8.5 [6.4-10.8] and 5.7 [4.6-11.9] in groups A B, respectively; p= 0.82). In the Cox model, a high neuron-specific enolase serum level was the only significant prognostic determinant.

Both regimens were active and induced a high rate of response, particularly for brain-target lesions. Myelotoxicity jeopardizes the acceptability of both regimens. Despite such an aggressive approach, none of the regimens suggested a putative overall improvement of outcome in this poor prognosis presentation of metastatic non-small cell lung cancer. The search for alternative therapies to chemotherapy, such as targeted therapy, is urgently warranted in this setting.