Brain metastases are a common site of distant failure for epithelial malignancies. They are often a harbinger of the final phase of the cancer patient's course. The appropriate management approach for a patient with a brain metastasis depends on the patient's clinical status, the extent of the disease, and the patient's wishes. Very few prospective or randomized studies have evaluated the best treatment approach for a patient with a brain metastasis. For example, no study has prospectively compared radiation therapy to best supportive care. However, the clear efficacy of radiation therapy in controlling, at least temporarily, the neurologic symptoms of brain metastases makes it difficult to advocate withholding radiation, except in the most extreme of situations . Corticosteroids alone, although a temporizing measure, quickly loose their efficacy, and their side effects may negatively impact on the patient's quality of life.
With these issues in mind, a review of the English language literature was undertaken in an to attempt to (1) determine the efficacy of radiation therapy for the treatment of brain metastases, (2) identify patient factors that have prognostic significance, and (3) ascertain whether treatment technique has any effect on outcome. Using Medline and journal references, relevant articles were reviewed and abstracted. Randomized trials were preferentially selected. The chosen articles were analyzed for treatment techniques, response, and survival.
Several randomized trials have looked at the effect of fractionation and dose on response and survival. Harwood and Simpson of the Princess Margaret Hospital in Toronto randomized 108 patients to receive either 30 Gy in ten 3-Gy fractions over 2 weeks or 10 Gy in one fraction . Among the 101 evaluable patients, there was no statistically significant difference in median survival between the single-fraction and multiple-fraction arms (132 vs 121 days). Nor was there a difference in the frequency of response between the two arms. Death from cerebral metastasis was reported in 70% of patients, with no difference seen in the recurrence rate between the two arms.
RTOG Trials of Whole-Brain Irradiation
The Radiation Therapy Oncology Group (RTOG) completed a series of protocols designed to optimize palliative whole-brain radiation therapy. The first trial, open from January 1971 to November 1973, randomized 993 patients to one of four treatment regimens: 40 Gy in 4 weeks, 40 Gy in 3 weeks, 30 Gy in 3 weeks, or 30 Gy in 2 weeks . A total of 910 patients were evaluable. A follow-up study, which ran from November 1973 to February 1976, randomized 1,001 patients to treatment with 40 Gy in 3 weeks, 30 Gy in 2 weeks, or 20 Gy in 1 week.
The results of these two trials were published together . The response to treatment was equivalent among all of the arms, even when patients were stratified by initial performance status. In both studies, patients with the shortest treatment regimens responded more quickly. For example, in the second study, 64% of the patients receiving 20 Gy in 1 week responded within 2 weeks, as compared with only 54% of the patients on the other arms.
There was no difference in the duration of improvement among the dose schedules. Similarly, there was no difference by treatment group for the time to progression. Overall, median survival was 18 weeks in the first study and 15 weeks in the second study. Death was attributed to brain metastasis in 49% of the patients in the first trial and 31% of the patients in the second.
The authors defined a "palliative index," which was the percentage of remaining life spent in an improved or stable neurologic state. For all patients in the study, 75% to 80% of the remaining life was palliated, using this marker. Again, there was no time or dose effect.
Ultra-Rapid High-Dose Arms--At the same time that these two studies were run, individual participating centers had the option to also randomize patients to ultra-rapid high-dose arms . In the first study, 26 patients were offered 10 Gy in a single fraction, and in the second study, 33 patients were treated with 12 Gy in two fractions. The percentage of these patients showing neurologic improvement was the same as those receiving more protracted fractionation (46%). The promptness of improvement was also consistent with the previous studies.
In contrast, the duration of improvement was less satisfactory. In the first study, the median duration of improvement (time to failure or death) was 10 weeks for the longer-fractionation arms and 4 weeks for the single 10-Gy fraction. However, in the second study, patients given 12 Gy in two fractions had a 10-week duration of improvement, which was the same as that seen in patients given the other fractionation schemes. The response of symptoms to treatment and treatment morbidity were the same on all arms. Median survival times for all of the arms also were equivalent, as was the percentage of deaths attributed to the brain metastases.
The authors concluded that the duration of improvement, time to progression, and complete disappearance of headache was generally poorer in both the 10-Gy arm and the 12-Gy arm. A review of the data indicates that this conclusion appears to hold true for the 10-Gy arm but is more equivocal with regard to the 12-Gy-in-two-fraction arm.
Factors Predicting Improved Survival--The complete study was analyzed to identify patients having a favorable survival . Patients with breast cancer and no soft-tissue metastases, ambulatory lung cancer patients with an absent primary or no extracerebral metastases, and other ambulatory patients with no extracerebral metastases had a median survival of 28 weeks, as compared with 11 weeks for the remaining patients (see Table 1). There was no outcome advantage to any fractionation scheme among even this selected subgroup of patients. The median time to progression was 24 weeks for the breast cancer patients and 13 weeks for the lung cancer patients.
Follow-up of Favorable Patients--A follow-up study by the RTOG enrolled 309 patients with a favorable prognosis (ie, those who had a positive brain isotope scans with no other evidence of disseminated disease, no severe neurologic impairment, and a controlled primary) . These patients were randomized between 30 Gy in ten 3-Gy fractions and 50 Gy in twenty 2.5-Gy fractions. The primary tumor was located in the lung in 80% of patients and in the breast in 7%. (In the remaining patients, the primary was either unknown or at another site.) Of the patients in the 30-Gy arm, 93% completed treatment. In contrast, 21% of the patients in the 50-Gy arm failed to get full therapy, receiving a median dose of 30 Gy (range, 2.5 to 47.5 Gy).
The median survival of patients in the 30-Gy arm was 18 weeks and that of patients in the 50-Gy arm was 17 weeks. The 1-year survival rates of both arms were approximately 25%. The two arms were equivalent with regard to the percentage of patients achieving response, time to achieving response, duration of response, and time to progression. Overall, 47% of the patients died with an improved or stable neurologic function.
Radiation With and Without a Radiosensitizer
Between 1979 and 1983, the RTOG then randomized 859 patients between 30 Gy in 10 fractions and 30 Gy in 6 fractions over 3 weeks, with or without the administration of the putative radiosensitizer misonidazole. Again, among the 779 analyzable cases, there were few differences in outcome among the arms . The overall median survival was 3.9 months, with 60% of patients alive at 3 months, 35% at 6 months, and 15% at 1 year. Brain metastases were considered the cause of death in one-third of the patients.
As a measure of a palliative index, 44% of the patients spent 90% to100% of their remaining survival time in a stable or improved Karnofsky status, and 65% maintained a stable or improved neurologic function. The median time to deterioration of performance status was 1.8 months.
Analyses for Prognostic Factors--Two papers have been published analyzing the patients who participated in this study (see Table 1). The first paper, by Diener-West and colleagues, identified four factors associated with improved survival: Karnofsky performance status of 70 to 100, an absent or controlled primary, age less than 60 years, and metastatic spread limited to the brain . Patients with all four charac-teristics had a 200-day survival of 52%; those with three characteristics, a survival of 38%; those with two characteristics, 24%; those with one characteristic, 18%; and those with no characteristics, 8%.
The second paper, by Swift and colleagues, looked at the CT characteristics of patients in the study in an attempt to identify a favorable subgroup . Pretreatment CT scans were available for over 750 patients. CT scan quality was variable, with 7% judged excellent; 45%, good; 34%, fair; 11%, poor; and 3%, absent or useless. Among the 408 patients who had at least one follow-up scan, 50% were responders and 21% were complete responders. Of the patients without a follow-up CT scan, 59% died within 6 weeks of initiation of treatment and 80%, within 12 weeks. The only prognostic factors that were statistically correlated with increased survival were one to three lesions (median survival of 4.0 months, vs 3.2 months in those with more than three lesions) and the absence of mid-line shift (4.3 months, vs 3.7 months in those with such a shift).