Single brain metastasis: whole-brain irradiation plus either radiosurgery or neurosurgical resection

Authors


Abstract

BACKGROUND:

The current study was conducted to compare neurosurgical resection (NR) followed by whole-brain irradiation (WBI) (NR + WBI) with WBI followed by radiosurgery (WBI + RS) in patients with a single brain metastasis.

METHODS:

The outcome of 41 patients treated with WBI + RS was retrospectively compared with 111 patients who received NR ;+ WBI with respect to local control of the treated metastasis and survival. Eleven additional potential prognostic factors were investigated, including WBI schedule, patient age, patient gender, Karnofsky performance score (KPS), primary tumor type, extracerebral metastases, recursive partitioning analysis (RPA) class, interval between the first diagnosis of cancer to the treatment of brain metastasis, metastatic site, maximum diameter of the metastasis, and graded prognostic assessment (GPA) score.

RESULTS:

The 1-year local control rates were 87% after WBI + RS and 56% after NR + WBI (P = .001). Using the Cox proportional hazards model, the treatment regimen remained significant (risk ratio [RR], 2.46; 95% confidence interval [95% CI], 1.29-5.17 [P = .005]). On the multivariate analysis, local control was also found to be associated with the maximum diameter of the metastasis. The 1-year survival rates were 61% after WBI + RS and 53% after NR + WBI (P = .16). Acute and late toxicities were similar in both groups. On the multivariate analysis, KPS, extracerebral metastases, RPA class, and the GPA score were found to be independent predictors of survival.

CONCLUSIONS:

The use of WBI + RS resulted in significantly better local control of the treated metastasis than NR + WBI. Survival was not found to be significantly different in either group. Because WBI + RS is less invasive than NR + WBI, it appears to be preferable for many patients with a single brain metastasis. These results should be confirmed in a randomized trial. Cancer 2012;. © 2011 American Cancer Society.

INTRODUCTION

For patients with a single brain metastasis, neurosurgical resection (NR) followed by whole-brain irradiation (WBI) (NR + WBI) has been established as standard treatment. In 2 randomized trials, survival was found to be significantly better after NR + WBI than after WBI alone.1, 2

Radiosurgery (RS) has also been compared with NR + WBI.3 In the randomized trial reported by Muacevic at al, who compared NR + WBI with RS alone, local control of the treated metastasis was not significantly different between the groups, but there was a trend favoring the use of RS alone (P = .06).3 The rate of cerebral metastases distant from the treated lesion was found to be significantly higher after RS alone than after NR + WBI (P = .04). The study by Muacevic et al has some limitations: it was stopped prematurely because of poor accrual after the inclusion of 64 patients and the primary endpoint, improvement in survival with the addition of WBI to RS, was not met (P = .8).3 However, it appears likely that the addition of WBI to RS would improve intracerebral and local control when compared with RS alone.

Randomized studies have examined WBI + RS versus RS alone, and the data indicated that WBI in addition to RS improves local control but not survival.4-6 The recently published randomized trial by Kocher et al compared patients who had received either NR alone (N = 79) or RS alone (N = 100) with patients who had received either NR + WBI (N = 81) or WBI + RS (N = 99).7 In this trial, the addition of WBI significantly improved local control without improving survival in the entire cohort as well in the surgery group and the RS group. In contrast to the results of the randomized studies, a nonrandomized study of 236 patients suggested a trend (P = .08) toward an improved median survival for the use of WBI + RS versus RS alone in patients without evidence of extracranial disease.8

The use of WBI in addition to NR or RS is still controversial. Additional WBI may increase the risk of late toxicity such as neurocognitive dysfunction.9, 10 Conversely, a local failure was suggested to be associated with increased deterioration in neurocognitive function.11, 12 The results of the available studies are difficult to interpret. The studies by Aoyama et al and Regine et al, which favored the use of WBI + RS versus RS alone, used the Mini-Mental State Examination (MMSE) for the evaluation of neurocognitive function, which is not a valid tool for neurocognition.11, 12 The randomized trial by Chang et al, which also compared WBI + RS with RS alone, used the Hopkins Verbal Learning Test, a more appropriate tool for neurocognition.6 In the study by Chang et al, evaluation of neurocognitive function was performed at 4 months. However, although local control at 1 year was found to be significantly better in the group treated with WBI +RS, neurocognitive function was not tested at that time. Thus, we do not really know whether long-term neurocognitive function is better or worse with additional WBI.

Another question that still remains to be answered is whether NR + WBI or WBI + RS is superior for the treatment of a single brain metastasis. Two studies could be identified to date that have compared NR with RS for the treatment of patients with a single brain metastasis. In 2000, Schoggl et al reported on 133 patients who received either NR in combination with WBI or RS (median dose, 17 gray [Gy]) in combination with WBI.13 The latter regimen resulted in significantly better local control but not survival rates. One may speculate that the results in the RS group could have been better if a RS dose higher than the median of 17 Gy was administered. In 2003, O'Neill et al presented a series of 74 patients treated with NR and 23 patients treated with RS and reported a significantly better local control for the RS group.14 However, 18% of the NR group and 4% of the RS group had not received additional WBI. Thus, more studies comparing NR + WBI with WBI + RS are required to better define the potential benefit of RS when compared with NR in patients with a single brain metastasis.

In the current study, which compared NR + WBI with WBI + RS, all patients had received WBI, and the median RS dose was 21 Gy. Both treatment groups were compared with respect to local control of the treated metastasis and survival.

MATERIALS AND METHODS

The outcome of 41 patients who received WBI + RS for the treatment of a single brain metastasis between 1999 and 2009 was retrospectively compared with that of 111 patients who were treated with NR + WBI with respect to local control of the treated metastasis, survival, and toxicity. In the group treated with WBI + RS, RS was administered as a boost after WBI, whereas in the group treated with NR + WBI, resection of the metastasis preceded WBI. Criteria for inclusion in this study were a single brain metastases (diameter ≤ 4 cm), recursive partitioning analysis (RPA)15 class 1 or 2, confirmation of metastasis by magnetic resonance imaging (MRI), and the administration of dexamethasone during WBI (at a dose of 12-32 mg/day). WBI was administered with a linear accelerator and 6- to10-megavolt photon beams, either delivering 5 × 4 Gy in 1 week (n = 18), 10 × 3 Gy in 2 weeks (n = 83), or 20 × 2 Gy in 4 weeks (n = 51). RS was performed as linac-based (n = 36) or Gamma Knife RS (n = 5). The median RS dose was 21 Gy (range, 15 Gy-25 Gy), which was prescribed to the periphery of the metastatic lesion. Depending on the location of the metastasis, the dose was 20 to 25 Gy for lesions measuring ≤ 2.5 cm, 18 to 20 Gy for lesions measuring 2.6 to 3.5 cm, and 15 to 18 Gy for lesions measuring > 3.5 cm. Treatment selection for the individual patient was based on current treatment policies at the contributing centers, the availability of either surgery or RS (waiting lists), and the decision of the treating physicians. Neurosurgical resection was performed by various experienced surgeons under ultrasound guidance and as an en bloc resection of each lesion whenever possible. The extent of resection was assessed by histology and MRI. Patient data were obtained from patients, treating physicians, and patient files (Table 1).

Table 1. Patient Characteristics of Both Treatment Groups
CharacteristicWBI+RSNR+WBIP
(N=41) No. (%)(N=111) No. (%)
  1. Abbreviations: FD: first diagnosis; GPA, graded prognostic assessment; Gy, gray; NR: neurosurgical resection; RPA, recursive partitioning analysis; RS: radiosurgery; WBI: whole-brain irradiation.

WBI schedule   
5 × 4 Gy (N=18)5 (12)13 (12) 
10 × 3 Gy (N=83)22 (54)61 (55) 
20 × 2 Gy (N=51)14 (34)37 (33).99
Age, y   
≤60 (N=78)22 (54)56 (50) 
>60 (N=74)19 (46)55 (50).92
Gender   
Female (N=86)25 (61)63 (57) 
Male (N=66)16 (39)48 (43).84
Karnofsky performance score   
70 (N=58)17 (41)41 (37) 
≥80 (N=94)24 (59)70 (63).84
Primary tumor   
Breast cancer (N=34)10 (24)24 (22) 
Non-small cell lung cancer18 (44)54 (49) 
(N=72)   
Other tumors (N=46)13 (32)33 (30).97
Extracerebral metastases   
No (N=90)25 (61)65 (59) 
Yes (N=62)16 (39)46 (41).93
RPA class   
1 (N=64)17 (41)47 (42) 
2 (N=88)24 (59)64 (58).97
Interval from FD of cancer to treatment of brain metastasis, mo
<12 (N=71)18 (44)53 (48) 
≥12 (N=81)23 (56)58 (52).86
Metastatic site   
Supratentorial (N=116)33 (80)83 (75) 
Infratentorial (N=36)8 (20)28 (25).79
Maximum diameter of the metastasis, cm
≤2.5 (N=75)22 (54)53 (48) 
>2.5 (N=77)19 (46)58 (52).74
GPA score, points   
1.5-2.5 (N=92)24 (59)68 (61) 
3.0-4.0 (N=60)17 (41)43 (39).92

In addition to the treatment regimen, 11 potential prognostic factors were investigated with respect to treatment outcomes. These factors included the WBI schedule (5 × 4 Gy vs 10 × 3 Gy vs 20 × 2 Gy), age (≤ 60 years vs > 60 years), gender, Karnofsky performance score (KPS) (70 vs ≥ 80), primary tumor type (breast cancer vs non-small cell lung cancer vs other tumors), extracerebral metastases (no vs yes), RPA class (1 vs 2),15 interval from tumor diagnosis to WBI (< 12 months vs ≥ 12 months), metastatic site (supratentorial vs infratentorial), maximum diameter of the brain metastasis (≤ 2.5 cm vs > 2.5 cm), and graded prognostic assessment (GPA) score (1.5-2.5 points vs 3.0-4.0 points).16 Both treatment groups were well balanced with regard to these 11 factors (Table 1).

Patients were followed until death or for a median of 12 months (range, 6 months-57 months) in surviving patients. Local control was defined as no progression or recurrence of the treated brain metastasis. Local failures were confirmed by MRI. To allow for better differentiation between local failure and radiation-induced necrosis in the WBI + RS group, the MRI scans were reviewed by specialized neuroradiologists. Time to endpoints was measured from the completion of treatment. Univariate analyses of local control and survival were performed using the Kaplan-Meier method17 and the Wilcoxon test. Factors found to be significant on the univariate analysis (P < .05) were included in a multivariate analysis that was performed with the Cox proportional hazards model. Because the RPA class and the GPA score included age, KPS, and extracerebral metastases, these were considered to be confounding variables. Therefore, if the RPA class or the GPA score and any of the other 3 factors were found to be significant on univariate analysis, separate multivariate analyses were performed. The comparison of the treatment groups for patient characteristics was performed with the chi-square test.

RESULTS

A local recurrence of the treated brain metastasis was observed in 59 patients (39%). On univariate analysis, improved local control was found to be significantly associated with the type of treatment, favoring WBI + RS (P = .001) (Table 2), and with a maximum diameter of the metastasis of ≤ 2.5 cm (P = .016) (Table 2). In addition, using the Cox proportional hazards model, both treatment regimen (risk ratio [RR], 2.46; 95% confidence interval [95% CI], 1.29-5.17 [P = .005]) and diameter of the metastasis (RR, 2.03; 95% CI, 1.19-3.52 [P = .009]) remained significant. For lesions measuring > 2.5 cm, the local control rates at 6 months were 94% after WBI + RS and 67% after NR + WBI, respectively (P = .22); the local control rates at 12 months were 86% and 42%, respectively (P = .005). For lesions measuring ≤ 2.5 cm, the local control rates at 6 months were 100% after WBI + RS and 84% after NR + WBI, respectively (P = .07); the local control rates at 12 months were 86% and 69%, respectively (P = .09).

Table 2. Results of Univariate Analysis With Respect to Local Control
VariableAt 6 Months, %PAt 12 Months, %P
  1. Abbreviations: FD: first diagnosis; GPA, graded prognostic assessment; Gy, gray; NR: neurosurgical resection; RPA, recursive partitioning analysis; RS: radiosurgery; WBI: whole-brain irradiation.

Treatment regimen    
WBI+RS (N=41)97 87 
NR+WBI (N=111)74.02156.001
WBI schedule    
5 × 4 Gy (N=18)94 79 
10 × 3 Gy (N=83)78 61 
20 × 2 Gy (N=51)81.3864.33
Age, y    
≤60 (N=78)83 69 
>60 (N=74)78.7259.29
Gender    
Female (N=86)82 63 
Male (N=66)78.7068.88
Karnofsky performance score 
70 (N=58)73 64 
≥80 (N=94)85.1365.10
Primary tumor    
Breast cancer (N=34)87 69 
Non-small cell lung cancer (N=72)75 62 
Other tumors (N=46)86.2564.33
Extracerebral metastases    
No (N=90)80 65 
Yes (N=62)81.9964.86
RPA class    
1 (N=64)84 69 
2 (N=88)78.7661.24
Interval from FD of cancer to treatment of brain metastasis, mo 
<12 (N=71)78 61 
≥12 (N=81)83.5568.34
Metastatic site    
Supratentorial (N=116)84 62 
Infratentorial (N=36)74.2871.19
Maximum diameter of the metastasis, cm 
≤2.5 (N=75)89 74 
>2.5 (N=77)74.1354.016
GPA score, points    
1.5-2.5 (N=92)78 64 
3.0-4.0 (N=60)86.3964.42

The median survival was 14 months for the group treated with WBI + RS compared with 13 months for the group treated with NR + WBI (P = .6) and 13 months for the entire cohort. On univariate analysis, improved survival was found to be significantly associated with age ≤ 60 years, a KPS ≥ 80, the absence of extracerebral metastases, RPA class 1, supratentorial location of the metastasis, and a GPA score of 3.0 to 4.0 points (Table 3). On multivariate analyses, KPS (RR, 1.66; 95% CI, 1.04-2.60 [P = .033]), extracerebral metastases (RR, 2.15; 95% CI, 1.24-3.74 [P = .007]), RPA class (RR, 1.87; 95% CI, 1.22-2.92 [P = .004]), and GPA score (RR, 1.68; 95% CI, 1.09-2.65 [P = .018]) remained significant. A trend was observed for age (RR, 1.64; 95% CI, 0.97-2.79 [P = .065]). The location of the metastasis was not found to be significant (RR, 1.39; 95% CI, 0.87-2.18 [P = .17]).

Table 3. Results of Univariate Analysis With Respect to Survival
VariableAt 6 Months, %PAt 12 Months, %P
  1. Abbreviations: FD: first diagnosis; GPA, graded prognostic assessment; Gy, gray; NR: neurosurgical resection; RPA, recursive partitioning analysis; RS: radiosurgery; WBI: whole-brain irradiation.

Treatment regimen    
WBI+RS (N=41)88 61 
NR+WBI (N=111)68.0653.16
WBI schedule    
5 × 4 Gy (N=18)78 59 
10 × 3 Gy (N=83)76 51 
20 × 2 Gy (N=51)69.7058.99
Age, y    
≤60 (N=78)83 66 
>60 (N=74)64.02743.011
Gender    
Female (N=86)80 59 
Male (N=66)66.1948.06
Karnofsky performance score 
70 (N=58)64 43 
≥80 (N=94)80.0962.003
Primary tumor    
Breast cancer (N=34)82 70 
Non-small cell lung cancer (N=72)74 51 
Other tumors (N=46)67.4348.12
Extracerebral metastases    
No (N=90)81 66 
Yes (N=62)63.04639<.001
RPA class    
1 (N=64)84 73 
2 (N=88)66.04441<.001
Interval from FD of cancer to treatment of brain metastasis, mo 
<12 (N=71)73 50 
≥12 (N=81)74.9758.48
Metastatic site    
Supratentorial (N=116)78 57 
Infratentorial (N=36)58.0646.040
Maximum diameter of the metastasis, cm 
≤2.5 (N=75)76 58 
>2.5 (N=77)71.7151.22
GPA score, points    
1.5-2.5 (N=92)64 43 
3.0-4.0 (N=60)88.00772<.001

In a subgroup analysis of the 111 patients treated in the NR + WBI group, complete resection of the metastasis resulted in significantly better 1-year local control (vs; P = 0.) and 1-year survival (58% vs 25%; P = .001) compared with incomplete resection.

Acute toxicity of grade ≥ 2 (according to National Cancer Institute Common Toxicity Criteria, version 2.0)18 such as headache, nausea, and fatigue occurred in 15% of patients receiving WBI + RS and 15% of those treated with NR + WBI (P = .95). Rates of late toxicities of grade ≥ 2 (Radiation Therapy Oncology Group [RTOG] criteria)19 such as neurocognitive deficits and vision or hearing problems occurred in 5% and 7% of patients, respectively (P = .93).

DISCUSSION

Uncertainty exists concerning the optimal treatment of patients with a single brain metastasis, in particular with regard to the pros and cons of using either NR or RS. A randomized trial of 95 patients demonstrated that the addition of WBI to NR can improve local control of the treated metastasis. A local recurrence was observed in 46% of patients after treatment with NR alone and in 10% after NR + WBI (P < .001).2

Several studies suggested a better local control rate if WBI was added to RS for patients with a very limited number of brain metastases. In 1992, a retrospective study of 27 patients with 1 to 4 brain metastases suggested better local control after treatment with WBI + RS than after RS alone, although the median survival times were similar.20 In another retrospective study of 135 patients with 1 to 5 brain metastases that was published in 2000, the 1-year local control rates were 80% after WBI + RS and 59% after RS alone (P = .034).21 However, in that study, the median survival times were not significantly different (10.5 months vs 6.4 months; P = .08). In a randomized trial of 132 patients with 1 to 4 lesions that was published in 2006, treatment with WBI + RS resulted in significantly better 1-year local control rates (89% vs 73%; P = .002) but not survival rates (39% vs 28%; P = .42) when compared with RS alone.5 In a more recent randomized trial of 58 patients, 73% of patients were free from an intracerebral recurrence 1 year after treatment with WBI + RS compared with only 27% after RS alone (P < .001).6

There may be a question of whether improvement in the local control of brain metastases can be considered important if it does not lead to improved survival. The addition of WBI to NR or RS is likely to increase treatment-related toxicity, particularly neurocognitive deficits. In the randomized trial by Chang et al, significantly more patients who were treated in the WBI + RS group demonstrated a decline in learning and memory function according to the Hopkins Verbal Learning Test compared with patients treated with RS alone at 4 months after treatment (96% vs 52%).6 Because of these results, the data monitoring committee stopped the trial after the inclusion of only 58 patients. Conversely, several authors have stated that local control is important to reduce neurocognitive dysfunction. Regine et al reported that patients with controlled brain metastases after WBI are more likely to have stable MMSE scores.12 Aoyama et al reported a significant drop in MMSE scores in 24% of patients at 1 year after treatment with WBI + RS and in 41% of patients after RS alone.11 At 2 years after treatment, the rates were 31% and 48%, respectively. Median times to deterioration of neurocognitive function were 16.5 months and 7.6 months, respectively (P = .05).11 Thus, WBI in addition to RS may be beneficial for patients with a relatively favorable survival prognosis. This accounts for many patients with a limited number of brain metastases, in particular patients with a single lesion. To define the best available treatment for such patients, it appears reasonable to compare NR + WBI with WBI + RS.

The current study compared treatment with NR + WBI with WBI + RS in patients with a single lesion. According to the results, WBI + RS resulted in better local control than NR + WBI without improving survival. The improvement in local control was independent of the size of the metastatic lesion. The retrospective design of the current study must be taken into account when interpreting these results. However, both treatment groups were well balanced with regard to potential prognostic factors, which considerably reduced the risk of selection bias. Furthermore, when compared with the data from the study by Kocher et al,7 the local control rates after NR + WBI reported in the current study are worse. In the study by Kocher et al,7 the 1-year local control rate after treatment with NR + WBI (81 patients, 98% of whom had a single metastasis) was 75%, compared with 56% in the current study. This may be explained by the fact that complete surgical resection of the metastases was a criterion for inclusion in the study by Kocher et al,7 whereas in the current study, 16% of the patients treated in the NR + WBI group had not undergone a complete resection. In the current study, complete resection led to significantly better treatment outcomes compared with incomplete resection.

The findings of the current study are consistent with the 2 previous studies that compared NR and RS in patients with a single brain metastasis. The study by Schoggl et al compared 66 patients who received NR + WBI with 67 patients receiving WBI + RS. The 1-year local control rates were 95% in the RS group and 83% in the NR group (P < .05).13 The 1-year survival rates were 52% and 44%, respectively (P = .55). One may question whether the results in the group treated with WBI + RS could be improved further by increasing the median RS dose beyond 17 Gy, as was done in the current study. However, despite the higher median dose of 21 Gy in our study, survival was not found to be significantly better after treatment with WBI + RS compared with after NR + WBI. The second previous study that compared NR with RS for the treatment of patients with a single brain metastasis was presented by O'Neill et al in 2003.14 The authors compared 74 patients treated with NR with or without WBI with 23 patients treated with RS with or without WBI. A local recurrence was observed in 58% of patients in the NR group and in none of the patients in the RS group (P = .020). The 1-year survival rates were 62% and 56%, respectively (P = .15). However, the local control results from the study by O'Neill et al may have been confounded to some degree by the finding that 18% of patients in the NR group and 4% of patients in the RS group had not received additional WBI. In contrast to these studies, the results of an earlier retrospective matched pair (1:2) study from 1996, which compared 31 patients treated with RS with or without WBI with 62 patients treated with NR with or without WBI, suggested that NR with or without WBI resulted in better median survival because of better local control (16.4 months vs 7.5 months; P < .001).22 However, the results of treatment with RS with or without WBI may have been confounded due to an inappropriate RS dose, because doses as low as 12 Gy have been used (median dose, 20 Gy; range, 12 Gy-22 Gy).

Understanding of prognostic factors, particularly those that help predict survival, can aid the physician in appropriate treatment selection and the design of future trials. Proper stratification based on known prognostic factors decreases the risk of uncontrolled biases fouling the results of trials. In the current study, KPS, extracerebral metastases, RPA class, and GPA score were found to be independent predictors of survival. These findings are in keeping with the data from the literature, in particular with regard to the prognostic factors included in scoring systems developed to predict the survival of patients with brain metastases.15, 16, 23

Treatment with WBI + RS appeared to be superior to that with NR + WBI in terms of better local control of the treated metastasis. Survival was not found to be significantly different between the treatment groups. Because WBI + RS is less invasive and appears to result in better local control, it appears preferable to NR + WBI for many patients with a single brain metastasis. These results should be confirmed in a randomized trial designed to include the significant prognostic factors in stratification.

FUNDING SUPPORT

No specific funding was disclosed.

CONFLICT OF INTEREST DISCLOSURES

The authors made no disclosures.

Ancillary