ATLANTA – Patients with non–small cell lung cancer and fewer than four brain metastases treated with stereotactic radiosurgery had better overall survival than did similar patients treated with whole-brain irradiation in a nonrandomized observational study.
The study of 413 patients who were eligible for either treatment showed that the median overall survival was 9.0 months for those treated with stereotactic radiosurgery (SRS) alone, versus 3.9 months for those treated with whole-brain radiation therapy (WBRT) alone, reported Dr. Lia M. Halasz, assistant professor of radiation oncology at the University of Washington in Seattle.
Dr. James B. Yu
The findings suggest the need for a randomized clinical trial comparing the two treatment strategies in patients with non–small cell lung cancer and up to three brain metastases, she said at the annual meeting of the American Society for Radiation Oncology.
"This observational data may better reflect real-world practice; however, the caveat is that all of these patients were treated at large NCCN [National Comprehensive Cancer Network] institutions, and may not reflect practices all across the United States," she said.
Dr. James B. Yu, a therapeutic radiologist and cancer outcomes researcher at Yale University, New Haven, Conn., commented that the study shows "at the very least, NCCN sites are doing a very good job at selecting patients for radiosurgery." Dr. Yu, the invited discussant, was not involved in the study.
There have been no randomized clinical trials directly comparing SRS alone vs. WBRT alone in patients with newly diagnosed brain metastases, and the optimal treatment for such patients is unknown, Dr. Halasz said. The investigators therefore undertook an observational study to determine whether one strategy had a therapeutic advantage over the other.
They identified 413 patients diagnosed with brain metastases without leptomeningeal disease from an NCCN longitudinal database from November 2006 through January 2010. The patients had all received radiation therapy with no neurosurgical resection within 60 days of diagnosis.
Of this group, 118 (29%) underwent SRS, 295 (71%) had WBRT; and 13 patients (3%) had both as initial treatment.
Patients with three or fewer metastases were significantly more likely to receive SRS than WBRT, whereas those with four or more metastases were more likely to receive WBRT (P less than .001). Other factors associated with choice of SRS were smaller metastases (P = .036) and one or no sites of extracranial disease, compared with two or more (P = .013).
The authors analyzed a subset of 197 patients with fewer than four brain metastases and all metastatic sites smaller than 4 cm, all of whom were eligible for either treatment, and 48% of whom underwent SRS alone. As noted before, the unadjusted overall survival in this group was 9.0 months for the SRS-treated patients, and 3.9 months for those treated with WBRT.
To compensate for patient-selection biases, the authors then performed a propensity score analysis in which they stratified patients by their propensity to receive radiosurgery. In this analysis, the estimated treatment effect of SRS on overall survival was a hazard ratio (HR) of 0.62 (P = .018). Factors significantly associated with overall survival included SRS vs. WBRT, number of brain metastases, extent of extracranial disease, institution, and year of treatment.
In an analysis using a standardized mortality-ratio weighing method, they found that the estimated treatment effect of SRS on overall survival was an HR of 0.67 (P = .007).
Additionally, the authors performed a sensitivity analysis of potential unmeasured confounders, assuming that patients who underwent WBRT were three times more likely to have a Karnofsky performance score less than 70, and that the HR for that poor performance status was 2.13, based on recursive partitioning analysis (RPA) status. In this analysis, the HR favoring SRS was 0.64 (P = .037).
Finally, they performed a companion analysis with breast cancer data, and found a similar HR in favor of SRS (HR, 0.59; P = .036)
The funding source for the study was not reported. Dr. Halasz and Dr. Yu reported having no conflicts of interest to disclose.