Stereotactic radiosurgery alone versus resection plus whole-brain radiotherapy for 1 or 2 brain metastases in recursive partitioning analysis class 1 and 2 patients

Authors


Abstract

BACKGROUND.

The objective of this study was to compare stereotactic radiosurgery (SRS) alone with resection plus whole-brain radiotherapy (WBRT) for the treatment of patients in recursive partitioning analysis (RPA) class 1 and 2 who had 1 or 2 brain metastases.

METHODS.

Two hundred six patients in RPA class 1 and 2 who had 1 or 2 brain metastases were analyzed retrospectively. Patients in Group A (n = 94) received from 18 grays (Gy) to 25 Gy SRS, and patients in Group B (n = 112) underwent resection of their metastases and received 10 × 3 Gy/20 × 2 Gy WBRT. Eight other potential prognostic factors were evaluated regarding overall survival (OS), brain control (BC), and local control (LC) of treated metastases: age, sex, performance status, tumor type, number of brain metastases, extracranial metastases, RPA class, and interval from tumor diagnosis to treatment of brain metastases.

RESULTS.

A comparison of the 2 treatment groups did not reveal significantly different OS (P = .19), BC (P = .52), or LC (P = .25). In RPA subgroup analyses, outcome also did not differ significantly for either RPA class of patients (P values from .21 to .83). On multivariate analysis, improved OS was associated with age ≤60 years (relative risk [RR], 1.75; P = .002), better performance status (RR, 1.67; P = .015), no extracranial metastases (RR, 2.84; P < .001), interval from tumor diagnosis to treatment >12 months (RR, 1.70; P = .003), and RPA class 1 (RR, 1.51; P = .016). Improved BC was associated with a single metastasis (RR, 1.54; P = .034) and an interval from tumor diagnosis to treatment >12 months (RR, 1.58; P = .019), and improved LC was associated with an interval from tumor diagnosis to treatment >12 months (RR, 1.59; P = .047).

CONCLUSIONS.

SRS alone appeared to be as effective as resection plus WBRT in the treatment of 1 or 2 brain metastases for patients in RPA class 1 and 2. Patient outcomes were associated with age, Karnofsky performance status, number of brain metastases, extracranial metastases, RPA class, and interval from tumor diagnosis to treatment. Cancer 2007. © 2007 American Cancer Society.

Brain metastases occur in 20% to 40% of adult cancer patients.1, 2 The survival of patients with brain metastases often is limited to a few months.3, 4 Patients who have 1 or 2 brain metastases appear to have better outcomes than patients who have >2 lesions.5, 6 Therefore, the treatment of patients who have 1 or 2 brain metastases often is more aggressive than the treatment of patients who have a greater number of metastases. Whole-brain radiotherapy (WBRT) alone is the most common treatment for multiple brain metastases. Stereotactic radiosurgery (SRS) often is administered to patients who have a good performance status and ≤3 metastases of suitable size (≤4 cm).7, 8 SRS is administered in a single fraction and has some potential benefits compared with conventional radiotherapy like WBRT. SRS is more precise, requires less treatment time, and reduces the volume of surrounding normal brain tissue being irradiated. A reduction in the treatment volume may result in reduced toxicity.

It has been demonstrated that SRS alone is an effective treatment for a limited number of brain metastases. Results from a study at the University of Pittsburgh suggested that brain metastases are controlled well with SRS alone and that WBRT in addition to SRS may be omitted to reduce the risk of radiation-related toxicity.9 The authors of that report stated that SRS alone is a reasonable alternative to WBRT followed by an SRS boost. Those data were supported by 2 retrospective studies, which indicated that the omission of WBRT in the initial management of patients who underwent SRS did not compromise survival or intracranial control in patients who had up to 4 brain metastases.10, 11 Therefore, SRS alone can be considered an effective therapy for a limited number of brain metastases and may be administered without additional WBRT.

Surgical resection also reportedly had an important role in the treatment of 1 (or 2) brain metastases. Results from 2 randomized trials12, 13 indicated that resection of metastasis followed by WBRT resulted in better survival and local control than WBRT alone for solitary metastasis. In contrast, the results from a third trial did not indicate a significant difference in survival between surgery plus WBRT and WBRT alone.14 However, the findings from that study may have been confounded by an unequal distribution of primary tumors between the groups and nonuniform calculation of the survival times. In addition, the majority of patients (73%) in that study had poorly controlled extracranial disease and may not have been expected to benefit from aggressive therapy of their brain metastases.15

Resection alone appeared to be less effective than resection followed by WBRT in terms of controlling brain disease and possibly survival. In a retrospective study, the brain recurrence rates were 85% after resection alone and 21% after resection plus WBRT16: The median survival times were 11.5 months and 21 months, respectively. A prospective trial that included patients who underwent gross resection of a single brain metastasis indicated that there was an improvement in local control for patients who received postoperative WBRT (82% vs 30%).17 Although overall survival did not differ significantly in that trial, the risk of death from neurologic causes was decreased with postoperative WBRT (14% vs 37%). Therefore, it was recommended that resection generally should be followed by WBRT.

It may be questioned whether SRS alone or resection followed by WBRT is superior with respect to survival or intracranial control. Very few studies have compared SRS with surgical resection; and, to our knowledge, no randomized trials have addressed this issue. The results from the available retrospective series are controversial. In a series of 75 patients, the median survival was 16.4 months after surgical resection alone versus 7.5 months after SRS.18 However, underdosage may have confounded the results after SRS. Another retrospective study of 108 patients compared surgery plus WBRT versus SRS for patients who had single lesions that measured ≤3.5 cm in greatest dimension.19 The 1-year survival and 1-year local control rates were similar in both groups. A third retrospective study of 133 patients compared SRS plus WBRT versus surgery plus WBRT in a series of 133 patients.20 The authors reported no significant difference in median survival but significantly better local control in the SRS group. A retrospective study from the Mayo Clinic suggested better local control after SRS but similar survival in 97 patients who had solitary brain metastasis.21 However, in that study, 96% of patients who underwent with SRS and 82% of patients who underwent resection received additional WBRT. Thus, there is a lack of studies comparing SRS alone with resection plus WBRT in patients with a limited number of brain metastases. In the current study, we compared both treatment regimens in a series of 206 patients who had 1 or 2 lesions.

The prognosis for patients with brain metastases is associated strongly with the recursive partitioning analysis (RPA) class. Using data from 3 Radiation Therapy Oncology Group (RTOG) brain metastases trials, RPA was performed to define prognostic factors for survival.22 Karnofsky performance status (KPS), age, status of the primary tumor (controlled vs uncontrolled), and extracranial metastases were the most relevant prognostic factors. Based on these factors, 3 prognostic classes were defined: RPA class 1 (KPS ≥70%, age <65 years, controlled primary, no extracranial metastases), RPA class 2 (KPS ≥70%, age ≥65 years, and/or uncontrolled primary, and/or extracranial metastases), and RPA class 3 (all patients with KPS <70%). The median survival for patients in RPA classes 1, 2 and 3 was 7.1 months, 4.2 months, and 2.3 months, respectively.22 It has been demonstrated that the RPA classes are applicable for both SRS and resection of brain metastases.23, 24 The maximal benefit from resection of metastases was observed for patients in RPA classes 1 and 2, whereas patients in RPA class 3 had poor survival regardless of treatment.15 Therefore, in the current study, included only patients in RPA classes 1 and 2. The appropriate treatment regimen may vary with RPA class. Therefore, the outcome of the 2 investigated therapies were compared for the entire cohort and for both RPA class 1 and RPA class 2.

MATERIALS AND METHODS

Data related to 206 patients who were treated for 1 or 2 brain metastases between 1994 and 2006 were analyzed retrospectively. These patients underwent either SRS alone (Group A; n = 94) or resection plus WBRT (Group B; n = 112). Only patients in RPA class 1 and 2 were included in the current study, because patients undergoing brain surgery or radiosurgery generally have an appropriately favorable performance status (KPS ≥70). In addition, patients in RPA class 3 do not appear to benefit from aggressive therapy.15 Group A patients received 18 grays (Gy) to 25 Gy of SRS (median dose, 23.5 Gy) either as linear accelerator-based SRS (n = 67 patients) or as γ-knife SRS (n = 27 patients). Group B patients underwent resection of metastases followed by 10 × 3 Gy or 20 × 2 Gy of WBRT administered with a linear accelerator and 6-megavolt (MV) to 10-MV photon beams. Further criteria for inclusion were as follows: no prior radiotherapy to the brain, confirmation of metastases by computed tomography or magnetic resonance imaging, size of metastases ≤4 cm, and administration of dexamethasone (12–32 mg/day) during radiotherapy. The prescribed dose of SRS represented the dose surrounding the tumor. The data for the analysis were obtained from the patients, their general practitioners, treating oncologists, and patient files. The patient characteristics are summarized in Table 1.

Table 1. Patient Characteristics of the 2 Treatment Groups
CharacteristicNo. of patients (%)P
SRS, n = 94OP + WBRT, n = 112
  1. SRS indicates stereotactic radiotherapy; OP + WBRT, resection plus whole-brain radiotherapy; PS, performance status; RPA, recursive partitioning analysis; DX, diagnosis; RT, radiotherapy.

Age, y
 ≤6044 (47)54 (48) 
 >6050 (53)58 (52).93
Sex
 Women52 (55)56 (50) 
 Men42 (45)56 (50).67
Karnofsky PS
 70–8072 (77)87 (78) 
 90–10022 (23)25 (22).94
Primary tumor
 Breast cancer15 (16)24 (21) 
 Lung cancer33 (35)44 (39) 
 Other tumors46 (49)44 (39).62
No. of brain metastases
 156 (60)78 (70) 
 238 (40)34 (30).44
Extracranial metastases
 No52 (55)65 (58) 
 Yes42 (45)47 (42).86
RPA class
 Class 134 (36)47 (42) 
 Class 260 (64)65 (58).65
Interval from DX of tumor to RT, mo
 ≤1242 (45)57 (51) 
 >1252 (55)55 (49).60

In addition to the treatment regimen, 8 other potential prognostic factors were evaluated regarding overall survival, brain control (freedom from recurrent brain metastases), and local control of the treated metastases: These factors included age (≤60 years vs >60 years), sex, KPS (70–80 vs 90–100), primary tumor type (breast cancer vs lung cancer vs other tumors), number of brain metastases (1 vs 2), presence of extracranial metastases at the time of radiotherapy, RPA class (class 1 vs class 2), and interval from tumor diagnosis to treatment of brain metastases (≤12 months vs >12 months). Both treatment groups were balanced with respect to these factors (Table 1). Subgroup analyses were performed for the 2 RPA classes individually to evaluate whether patients in either RPA class would benefit from 1 of the 2 compared treatment regimens.

Patients were followed until death or for 6 months to 57 months (median, 9 months) in survivors. Brain control was defined as the absence of new or recurrent brain metastases. Local control was defined as the absence of recurrence of the treated brain metastases. Local or brain failure was confirmed by computed tomography or magnetic resonance imaging. Time to any endpoint was measured from the completion of radiotherapy. Survival, brain control, and local control rates were calculated using the Kaplan-Meier method.25 The differences between the Kaplan-Meier curves were determined with the log-rank test (univariate analysis). The prognostic factors that were identified as significant (P < .05) were included in a multivariate analysis, which was performed with the Cox proportional-hazards model.

RESULTS

On univariate analysis, improved survival was associated significantly with age ≤60 years (vs >60 years), KPS from 90 to 100 (versus KPS from 70 to 80), lack of extracranial metastases, RPA class 1 (vs RPA class 2), and an interval from tumor diagnosis to treatment of brain metastases >12 months (vs ≤12 months) (Table 2). On multivariate analysis, age (relative risk [RR], 1.75; 95% confidence interval [95% CI],1.23–2.48; P = .002), KPS (RR, 1.67; 95% CI, 1.10–2.56; P = .015), extracranial metastases (RR, 2.84; 95% CI, 1.68–5.04; P < .001), RPA class (RR, 1.51; 95% CI, 1.08–2.14; P = .016), and the interval from tumor diagnosis to treatment (RR, 1.70; 95% CI, 1.20–2.39; P = .003) maintained significance.

Table 2. Results of the Univariate Analysis of Survival
Variable% of patientsP
At 6 monthsAt 12 monthsAt 18 monthsAt 24 months
  1. SRS indicates stereotactic radiotherapy; OP + WBRT, resection plus whole-brain radiotherapy; PS, performance status; RPA, recursive partitioning analysis; DX, diagnosis; RT, radiotherapy.

Treatment
 SRS, n = 9468543620 
 OP+WBRT, n = 11261383014.19
Age, y
 ≤60, n = 9876553920 
 >60, n = 10854372614.012
Sex
 Women, n = 10864453214 
 Men, n = 9565453319.69
Karnofsky PS
 70–80, n = 15960422713 
 90–100, n = 4781584927.012
Primary tumor
 Breast cancer, n = 3972624017 
 Lung cancer, n = 7763453619 
 Other tumors, n = 9062382613.38
No. of brain metastases
 1, n = 13466493420 
 2, n = 7262382912.15
Extracranial metastases
 No, n = 11776574225 
 Yes, n = 894931196<.001
RPA class
 Class 1, n = 8178574322 
 Class 2, n = 12556382513.017
Interval from DX of tumor to RT, mo
 ≤12, n = 9960372814 
 >12, n = 10768533720.030
All patients, n = 20664453317 

A recurrence anywhere in the brain (local or distant intracerebral failure) occurred in 107 patients (52%) after a median interval of 6 months (range, 0–47 months) after radiotherapy. The potential prognostic factors in relation to brain control are shown in Table 3 (univariate analysis). On univariate analysis, improved brain control was associated significantly with a single brain metastasis (vs 2 metastases) and with an interval from tumor diagnosis to treatment >12 months (vs ≤12 months). On multivariate analysis, both the number of brain metastases (RR, 1.54; 95% CI, 1.03–2.26; P = .034), and the interval from tumor diagnosis to treatment (RR, 1.59; 95% CI, 1.08–2.33; P = .019) remained significant.

Table 3. Results of the Univariate Analysis of Brain Control
Variable% of patientsP
At 6 monthsAt 12 monthsAt 18 monthsAt 24 months
  1. SRS indicates stereotactic radiotherapy; OP + WBRT, resection plus whole-brain radiotherapy; PS, performance status; RPA, recursive partitioning analysis; DX, diagnosis; RT, radiotherapy.

Treatment
 SRS, n = 9474492819 
 OP+WBRT, n = 11267443415.52
Age, y
 ≤60, n = 9875543623 
 >60, n = 10865382510.07
Sex
 Women, n = 10871473522 
 Men, n = 9569452714.64
Karnofsky PS
 70–80, n = 15966413116 
 90–100, n = 4782593322.11
Primary tumor
 Breast cancer, n = 3983583518 
 Lung cancer, n = 7766483415 
 Other tumors, n = 9067392620.38
No. of brain metastases
 1, n = 13473513620 
 2, n = 7265362113.032
Extracranial metastases
 No, n = 11772503720 
 Yes, n = 8966402316.09
RPA class
 Class 1, n = 8173483317 
 Class 2, n = 12568452917.51
Interval from DX of tumor to RT, mo
 ≤12, n = 996437255 
 >12, n = 10775543526.018
All patients, n = 20670463117 

Local recurrence of the treated brain metastases was observed in 76 patients (37%) after a median interval of 6 months (range, 0–27 months). The potential prognostic factors in relation to local control are summarized in Table 4 (univariate analysis). On univariate analysis, improved local control was associated significantly only with an interval from tumor diagnosis to treatment >12 months (vs ≤12 months). On multivariate analysis, the interval from tumor diagnosis to treatment maintained significance (RR, 1.59; 95% CI, 1.01–2.50; P = .047).

Table 4. Results of the Univariate Analysis of Local Control
Variable% of patientsP
At 6 monthsAt 12 monthsAt 18 monthsAt 24 months
  1. SRS indicates stereotactic radiotherapy; OP + WBRT, resection plus whole-brain radiotherapy; PS, performance status; RPA, recursive partitioning analysis; DX, diagnosis; RT, radiotherapy.

Treatment
 SRS, n = 9484644934 
 OP+WBRT, n = 11274564729.25
Age, y
 ≤60, n = 9883675038 
 >60, n = 10875524522.10
Sex
 Women, n = 10879595134 
 Men, n = 9578604430.93
Karnofsky PS
 70–80, n = 15976585236 
 90–100, n = 4786654026.85
Primary tumor
 Breast cancer, n = 3983664829 
 Lung cancer, n = 7774584728 
 Other tumors, n = 9081574740.78
No. of brain metastases
 1, n = 13478605033 
 2, n = 7279584231.42
Extracranial metastases
 No, n = 11780605032 
 Yes, n = 8976584332.45
RPA class
 Class 1, n = 8179584530 
 Class 2, n = 12579615035.99
Interval from DX of tumor to RT, mo
 ≤12, n = 9975493917 
 >12, n = 10782685442.041
All patients, n = 20679604732 

The subgroup analyses of each of the 2 RPA classes did not reveal a significant difference between the 2 compared treatment regimens with respect to overall survival, brain control, or local control. These results are summarized in Table 5.

Table 5. Subgroup Analyses of Each of the 2 Recursive Partitioning Analysis Classes Evaluated Individually for Survival, Brain Control, and Local Control: Potential Impact of the Treatment Schedule
Variable% of patientsP
At 6 monthsAt 12 monthsAt 18 monthsAt 24 months
  1. RPA indicates recursive partitioning analysis; SRS, stereotactic radiotherapy; OP + WBRT, resection plus whole-brain radiotherapy.

Survival
 RPA class 1
  SRS, n = 3485654928 
  OP+WBRT, n = 4772513819.21
 RPA class 2
  SRS, n = 6058482816 
  OP+WBRT, n = 6553292510.32
Brain control
 RPA class 1
  SRS, n = 3480473220 
  OP+WBRT, n = 4768503313.40
 RPA class 2
  SRS, n = 6070512519 
  OP+WBRT, n = 6566393920.83
Local control
 RPA class 1
  SRS, n = 3483614932 
  OP+WBRT, n = 4776564228.35
 RPA class 2
  SRS, n = 6085684836 
  OP+WBRT, n = 6573565628.48

The grade 3 acute toxicity rates according to the Common Toxicity Criteria (version 2.0) were 2% in Group A and 6% in Group B. Surgery-related complications, such as brain abscess, occurred in 2% of Group B patients. Grade 3 late toxicity rates according to the RTOG criteria were 4% in Group A and 3% in Group B.

DISCUSSION

The results of treatment for patients with brain metastases generally are poor. Most patients have multiple brain metastases, receive WBRT alone, and live for only 3 to 6 months. Patients who have a limited number of lesions often receive more aggressive treatment, because their prognosis appears to be better than the prognosis for patients who have multiple lesions.5, 6 Resection plus WBRT has been associated with a better outcome than either resection alone or WBRT alone for patients in RPA class 1 and 2 who have a single metatasis.12, 13, 15–17 SRS alone and WBRT plus an SRS boost reportedly were effective therapies for up to 3 or 4 brain small metastases.9–11, 26–28 Three studies compared SRS alone with WBRT plus an SRS boost and suggested that the omission of WBRT in the initial management of patients who underwent SRS did not compromise survival or intracranial control.9–11 Therefore, SRS alone appears to be effective in the treatment of a limited number of brain metastases. WBRT in addition to SRS may be omitted potentially to reduce radiation-related toxicity.

To our knowledge, only 1 previous study compared resection of metastases followed by WBRT with SRS alone. That retrospective study included 108 patients.19 Neither survival nor local control differed significantly for the treatment groups, which we also observed in the current study.

The current analysis included 206 patients, and both survival and local control appeared to be slightly better after SRS alone than after resection plus WBRT. However, statistical significance was not achieved. Both treatment groups were balanced for the other potential prognostic factors, which reduced the risk of a selection bias. However, the retrospective design of the current study should be taken into account when interpreting the results. Acute and chronic toxicity rates were relatively low in both treatment groups, which is in accordance with the findings from other investigators.15, 29 In the subgroup analyses, the results of both treatment options appeared similar for patients in both RPA class 1 and RPA class 2.

Thus, it appears that patients in RPA classes 1 and 2 who have 1 or 2 brain metastases may be treated adequately without brain surgery. These patients may be treated well with SRS alone, which is less invasive. Furthermore, SRS, which is administered as a single radiation fraction (treatment time = 1 day), is less time-consuming and more cost-effective than resection of metastases followed by a period of recovery and at least 2 weeks of WBRT.

Patient outcome was influenced by other potential prognostic factors. Improved survival was associated significantly with younger age, better performance status, lack of extracranial metastases, lower RPA class, and a longer interval from tumor diagnosis to treatment of brain metastases. Improved brain control was associated significantly with a single brain metastasis and a longer interval from tumor diagnosis to treatment. Improved local control was associated significantly only with a longer interval from tumor diagnosis to treatment.

These findings are in accordance with an RTOG RPA report regarding the survival of patients with brain metastases in which it was demonstrated that age, KPS, and lack of extracranial metastases were the most relevant prognostic factors.22 The number of metastases and the interval from tumor diagnosis to treatment of metastases were significant in the univariate analysis of that study. The negative prognostic impact of a shorter interval from tumor diagnosis to the diagnosis and treatment of the brain metastasis likely reflects the faster growth of more aggressive tumors and has also been demonstrated for other clinical situations, such as metastatic spinal cord compression.30

In conclusion, for patients in RPA class 1 and 2 who underwent either SRS alone or resection of metastases plus WBRT for 1 or 2 brain metastases, the prognosis was associated with age, KPS, the number of brain metastases, extracranial metastases, RPA class, and interval from tumor diagnosis to treatment. The treatment regimen (SRS alone vs resection plus WBRT) had no significant impact on outcome. Thus, for patients in RPA class 1 and 2 who have 1 or 2 brain metastases, SRS alone is a reasonable alternative that is less invasive, less time-consuming, and more cost-effective than resection plus WBRT.

Ancillary