The role of whole brain radiation therapy in the management of newly diagnosed brain metastases: a systematic review and evidence-based clinical practice guideline

Seven studies met the eligibility criteria for this treatment comparison, and of these six were unique and one was a companion study [6‐11, 30]. Three of these studies were prospective randomized trials [6‐8] (Table 1). Given that the treatment modalities being compared included surgical resection in only one arm of each trial, all of the RCTs were non-blinded. In a randomized trial performed at the University of Kentucky [6], 48 patients with known systemic cancer were treated with either biopsy of the suspected brain metastasis plus WBRT or complete surgical resection of the metastasis plus WBRT. The radiation doses were the same in both groups and consisted of a total dose of 36 Gy given in 12 daily fractions of 3 Gy each. Patients had to be capable of caring for themselves independently, with a Karnofsky performance score (KPS) of at least 70. Patients were ineligible if they had a need for immediate treatment to prevent acute neurologic deterioration, or if they had tumors considered to be relatively radiosensitive [small cell lung cancer (SCLC), germ-cell tumors, lymphoma, leukemia, and multiple myeloma]. Patients were not excluded based on the extent of systemic disease. Randomization was performed by computer-generated random numbers. Information on allocation concealment was not reported. All the patients in the surgical group were considered to have had complete resection as assessed by postoperative computerized tomography (CT) scanning. Follow up brain CT or magnetic resonance imaging (MRI) scans were required every 3 months. There was a statistically significant increase in survival in the surgical group (40 vs. 15 weeks). In addition, the time to recurrence of brain metastases, freedom from death due to neurologic causes, and duration of functional independence were significantly longer in the surgical resection group. The 1 month mortality was 4% in each group, indicating that there was no extra mortality from surgery. Although surgical resection was the only variable positively associated with maintaining performance status, the extent of systemic disease and increased age were associated with poor performance status post-treatment.

A second randomized study [8], conducted as a multi-institutional trial in the Netherlands, contained 63 evaluable patients. Patients with single brain metastases were randomized to complete surgical resection plus WBRT or WBRT alone. Randomization was performed centrally by telephone. The WBRT schedules were the same for both treatment arms and consisted of 40 Gy given in a non-standard fractionation scheme of 2 Gy twice per day for 2 weeks (10 treatment days). Patients had to have a reasonable quality of life and neurological status, defined as spending no more than 50% of their time in bed and not requiring continuous nursing care or hospitalization. Excluded histologies were SCLC and lymphoma. Information is not given regarding the extent of resection in the surgical group or the use and frequency of imaging in follow up. Survival was significantly longer in the surgical group (10 vs. 6 months). There was also a non-significant trend toward longer duration of functional independence in the surgically treated patients. No data concerning recurrence of brain metastases were provided. The 1 month mortality rates were 9% in the surgery group and 0% in the WBRT alone group, a statistically insignificant difference. The authors concluded that the addition of surgery to WBRT provided a survival benefit except to those patients who were 60 years of age or older, or those patients with progressive systemic disease in the 3 months prior to the diagnosis of the brain metastasis.

A third randomized trial, conducted as a multi-center trial in Canada by Mintz et al. [7], failed to find a benefit from surgical treatment. In that study, 84 patients with a single brain metastasis were randomized to receive radiotherapy alone (30 Gy given in 10 daily fractions of 3 Gy) or surgery plus radiotherapy. Randomization was performed centrally by telephone. Eligible patients had to be less than 80 years of age, and they had to have a KPS of at least 50, i.e., they could be spending more than 50% of their time in bed but had to be able to care for some personal needs. Patients were not eligible if they had leukemia, lymphoma, or SCLC. A CT scan was done in the first postoperative week to assess the extent of tumor removed. Follow up CT scans were performed monthly for 6 months and every 3 months after that. A gross total resection was achieved in 38 of the 40 patients in the surgical group. No difference was found in overall survival; the median survival time was 6.3 months in the radiotherapy alone group and 5.6 months for the surgical group. There was also no difference in causes of death or quality of life.

It is unclear why the Canadian study was not in agreement with the other two trials. In all three studies, the control arms (the radiation alone arms) had median lengths of survival in the 3–6 months range—within the expected range for patients treated with radiotherapy alone. The major difference in the studies was the poor results obtained in the surgical arm of the Canadian trial. That study contained a higher proportion of patients with extensive systemic disease and lower performance scores. It is possible that these factors resulted in more patients dying of their systemic cancer before a long term benefit of surgery was seen. Additionally, MRI was not mandatory in the Canadian study, and it is theoretically possible that patients with additional lesions not detected by CT may have been included in the study.

All three of the evidence class II studies [9‐11] demonstrated a survival benefit for patients who underwent surgical resection followed by WBRT as compared to WBRT alone. Ampil et al., published a single institution experience with either surgery plus WBRT (11 patients) versus WBRT alone (34 patients) in 45 patients who had a cerebellar metastasis. The majority of the patients in the WBRT alone arm had additional supratentorial brain metastases. Although there was a significant difference in survival noted between the two groups, the authors concluded that the outcome of patients with a brain metastasis within the cerebellum was improved with surgical resection if the primary was not from the lung.

Twenty-three studies met the eligibility criteria for this question, and of these, 17 were unique [12‐28] and six [31‐36] were companion publications (Table 2). The 17 unique studies fell into three AANS/CNS evidence class categories as follows: ten class I studies (nine RCTs [12‐20] and one randomized phase I/II trial [21]), six class II studies [22‐24, 26‐28] (retrospective cohort studies) and one class III study [25] (prospective cohort study with historical controls).

The radiation dosages have been expressed in terms of the tumor response biologically effective dose (BED) in order to quantitatively capture the observed biological effect between treatment arms. This was calculated from the linear quadratic equation:\( {\text{BED}} = nd\left[ {1 + d/(\alpha /\beta )} \right] \) where n = number of treatments, d = dose per fraction; the calculation assumes α/β = 10 Gy for tumor effects of each schedule, however it is uncorrected for treatment regimes of two fractions/day. The BED units are referred to in terms of Gy₁₀ to indicate that the BED values are single-point calculations for tumor response [37, 38] and no correction for accelerated repopulation was made.

Expressing radiation dosages in terms of the BED takes into account the total dose of radiation, fraction size, and overall time to deliver the radiation, and presumed repair of irradiated tissue. The analyses are stratified by low or high dose versus control dose. The control group consists of patients treated with 30 Gy in 10 fractions for a BED = 39 Gy₁₀ (therefore assigning the low dose regimens as a BED < 39 Gy₁₀, and high dose regimens as a BED > 39 Gy₁₀).

None of the trials demonstrated a meaningful improvement in any endpoint relative to dose; specifically, survival was not improved. There was considerable overlap in terms of survival even at the same dose level in different trials, underscoring the significance of host-specific variables in determining survival.

The data from the randomized trials, stratified by the BED are shown in Table 3.

The meta-analyses were stratified by low or high dose versus control dose. Studies lacking a comparable control were excluded from meta-analyses, due to the inherent dosing variability amongst the study’s control group.

Figure 2 indicates the RR of mortality at 6 months in the low dose (BED < 39 Gy₁₀) group compared to that in the WBRT control group (BED = 39 Gy₁₀). Only data from two trials (Chantani et al. [13] and Priestman et al. [20]) were robust enough to be considered for this end-point. When combined, no difference (P = 0.52) was found for 6 month mortality (RR 1.05; 95% CI 0.90, 1.23). Figure 3 presents the same analysis, but for the comparison of the high dose (BED > 39 Gy₁₀) group to the WBRT control group (BED = 39 Gy₁₀). Data from five trials were robust enough to be considered for this end-point. Combining data for Chantani et al. [13, 14], Komarnicky et al. [17], Kurtz et al. [18], and Murray et al. [19], yielded no difference (P = 0.39) in 6 month mortality (RR 1.05; 95% CI 0.94, 1.18). A 1981 trial by Borgelt et al. [12] was excluded from the meta-analyses for lack of a comparable control. Since the administered dosage ranged from 20 to 40 Gy over 1–4 weeks, the resulting permutations and combinations of plausible radiation schedules in this group varied too substantially (22–72 Gy₁₀) to function as a true control. The inherent dosing variability disqualified this control group thereby making the trial ineligible for comparison.

Similar comparisons were made for overall survival and neurologic function, and no dose–effect was identified for either end-point. Figures 4 and 5 depict the overall survival for the respective comparisons of low and high BED values versus the control BED. In Fig. 4, the pooling of two studies, Chantani et al. [13] and Priestman et al. [20], for the comparison of low dose WBRT (BED < 39 Gy₁₀) to a control dose (BED = 39 Gy₁₀) yielded no difference (P = 0.07) in overall survival (RR 0.86; 95% CI 0.73, 1.01). In Fig. 5, overall survival data from Chantani et al. [13, 14], Komarnicky et al. [17], Kurtz et al. [18], and Murray et al. [19], were pooled for the comparison of high dose WBRT (BED > 39Gy₁₀) versus WBRT control dose (BED = 39 Gy₁₀). No difference (P = 0.26) in overall survival was found (RR 1.06; 95% CI 0.96, 1.18). The hazard plots for neurologic function are not presented.

Using the AANS/CNS evidence classification, no papers with class I or II evidence were identified by the systematic literature review that met the eligibility criteria for this question. In fact, only one paper, providing class III evidence, met the inclusion criteria (Table 4). The only data is provided by a single institution retrospective analysis with 75 cases (Sundstrom et al. [29]). In that study, there were 15 cases that were of classic “radio-resistant” histology. There were no statistically significant differences in overall survival by tumor histology. Local control by tumor type was not reported.

Based on the available class I and class II evidence, surgical resection followed by WBRT is an effective treatment for patients with single, surgically accessible, brain metastases who have controlled extra-cranial disease and are in good general condition. Good general condition in the relevant studies was defined as functional independence and spending less than 50% of time in bed. Patients with disease progression in the 3 months preceding the diagnosis of the brain metastases have a relatively poor survival and poor functional status but still had a significant improvement in survival as a result of surgical resection. Due to the risk of fourth ventricle compression and the subsequent increase in intracranial pressure, surgery should be particularly considered if there is a brain metastasis situated within the posterior fossa.

The surgical resection guideline paper by Kalkanis et al. [1] outlines in detail the evidence supporting the addition of WBRT after surgical resection. Please refer to this paper for a further discussion of why the combined modalities of surgical resection followed by WBRT represent a superior treatment option, in terms of improving tumor control at the original site of the metastasis and in the brain overall, when compared to surgical resection alone

There is class I evidence that altered dose/fractionation schedules of WBRT do not result in significant differences in median survival, local control or neurocognitive function when compared to “standard” WBRT dose/fractionation. (i.e., 30 Gy in 10 daily fractions or a BED of 39 Gy₁₀).

Only one small case series (class III evidence) met the eligibility criteria for the question addressing the impact of histopathology on WBRT treatment outcomes. Further studies in this area are needed before any recommendations can be made.

There have been numerous studies comparing various dose/fractionation schemes for WBRT. Unless future studies incorporate more sophisticated measures of neurocognitive outcome there is little need to repeat these studies. There are, however, very few studies evaluating the effectiveness of WBRT for one histopathological type versus another. Future studies of WBRT explicitly addressing treatment outcomes for well-defined patient groups by different tumor types are needed.

The following is a list of major ongoing or recently closed randomized trials pertaining to the use of whole brain radiation therapy that evaluate treatment comparisons addressed by this guideline paper for the management of newly diagnosed brain metastases.