Since the types of patients, prognosis, and treatment strategy are different between patients with a single brain metastasis compared to those with multiple brain metastases, studies addressing these two groups of patients were examined separately. The studies were further divided by study design, based on the question the trials were intended to address. The quality of the studies was assessed using the Jadad quality assessment tool.
Study characteristics, including inclusion criteria, intervention, number analysed, types of outcomes reported, and results, were extracted in duplicate. Specifically, data on outcomes of interest, including survival, intracranial progression-free duration, response of brain metastases to therapy, quality of life, symptom control, neurological function, and toxicity, were extracted.
The proportion of patients with brain response and progression is dependent on the imaging modality used (computed tomography [CT] or magnetic resonance imaging [MRI]). Similarly, neurological symptom response and quality of life are sensitive to the tool used for evaluation. These details were tabulated.
For the evaluation of dose response, many different dose fractionation schedules were compared. The most commonly employed "control" regimen was 3,000 cGy in 10 fractions. The concept of Biological Equivalent Dose (BED) was used to facilitate comparison among different dose fractionation regimens. BED can be calculated using the equation BED = nd (1+ d/alpha/beta) where n = number of fractions, d = dose per fraction, and alpha/beta = 10 for tumour. For the purpose of assessing dose response, studies were divided into those comparing lower doses to 3,000 cGy in 10 fractions, and higher doses compared with 3,000 cGy in 10 fractions. As 2,000 cGy in five fractions is most commonly employed in Canada, and this is the second most commonly employed standard regimen, outcomes comparing 2,000 cGy in five fractions versus 3,000 cGy in 10 fractions are also presented.
For the pooled analysis of brain tumour response, the number of patients with a complete or partial response was abstracted from the tables or text in published reports. Tumour response was determined by the proportion of patients achieving complete response (CR) or partial response (PR). Patients were considered to have responded (CR + PR) if there was a 50% or greater decrease in lesion size and they were on a stable or decreasing dose of corticosteroids. Intracranial progression-free duration was defined as the duration during which there was no intracranial tumour growth and no new brain metastases.
Mortality data were obtained by estimating, from the Kaplan-Meyer probability curves presented in each report, the number of patients who died within six months after randomization.
The statistical package Revman 4.1 (Metaview © Update Software) provided by the Cochrane Collaboration was used for all analyses. Relative risk (RR) with 95% confidence intervals (CI) using the random effects model was reported as the more conservative estimate of effect. Analyses were primarily conducted on an intention-to-treat basis; however, when the number of patients randomized per study arm was not reported, the number of patients evaluable was analyzed. For tumour response, a RR >1.0 indicates that the patients in the experimental treatment group experienced better response compared with those in the control group. For mortality analyses, a RR <1.0 indicates that the patients in the experimental treatment group experienced fewer deaths compared with those in the control group.