- Email: [email protected]
Brain Metastases From Breast Carcinoma: Validation of the Radiation Therapy Oncology Group Recursive Partitioning Analysis Classification and Proposition of a New Prognostic Score
Int. J. Radiation Oncology Biol. Phys., Vol. 69, No. 3, pp. 839–845, 2007 Copyright Ó 2007 Elsevier Inc. Printed in the USA. All rights reserved 0360-3016/07/$–see front matter
doi:10.1016/j.ijrobp.2007.04.024
CLINICAL INVESTIGATION
Brain
BRAIN METASTASES FROM BREAST CARCINOMA: VALIDATION OF THE RADIATION THERAPY ONCOLOGY GROUP RECURSIVE PARTITIONING ANALYSIS CLASSIFICATION AND PROPOSITION OF A NEW PROGNOSTIC SCORE ROMUALD LE SCODAN, M.D.,* CHRISTOPHE MASSARD, M.D.,* EMMANUELLE MOURET-FOURME, M.D.,y JEAN MARC GUINEBRETIERRE, M.D., PH.D.,z CHRISTINE COHEN-SOLAL, M.D.,* BRIGITTE DE LALANDE, M.D.,* PATRICIA MOISSON, M.D.,* CHRISTELLE BRETON-CALLU, M.D.,* MIRIAM GARDNER, M.D., PH.D.,* ALAIN GOUPIL, M.D.,x NICOLE RENODY, M.D.,* JEAN LOUIS FLOIRAS, M.D.,* AND ALAIN LABIB, M.D.* Departments of *Radiation Oncology, y Medical Statistics, z Pathology, and x Medical Oncology, Centre Rene´ Huguenin, Saint Cloud, France Purpose: To validate the Radiation Therapy Oncology Group Recursive Partitioning Analysis (RTOG RPA) classification and determine independent prognostic factors, to create a simple and specific prognostic score for patients with brain metastases (BM) from breast carcinoma treated with whole-brain radiotherapy (WBRT). Methods and Materials: From January 1998 through December 2003, 132 patients with BM from breast carcinoma were treated with WBRT. We analyzed several potential predictors of survival after WBRT: age, Karnofsky performance status, RTOG-RPA class, number of BM, presence and site of other systemic metastases, interval between primary tumor and BM, tumor hormone receptor (HR) status, lymphocyte count, and HER-2 overexpression. Results: A total of 117 patients received exclusive WBRT and were analyzed. Median survival with BM was 5 months. One-year and 2-year survival rates were 27.6% (95% confidence interval [CI] 19.9–36.8%) and 12% (95% CI 6.5–21.2%), respectively. In multivariate analysis, RTOG RPA Class III, lymphopenia (#0.7 109/L) and HR negative status were independent prognostic factors for poor survival. We constructed a three-factor prognostic scoring system that predicts 6-month and 1-year rates of overall survival in the range of 76.1–29.5% (p = 0.00033) and 60.9–15.9% (p = 0.0011), respectively, with median survival of 15 months, 5 months, or 3 months for patients with none, one, or more than one adverse prognostic factor(s), respectively. Conclusions: This study confirms the prognostic value of the RTOG RPA classification, lymphopenia, and tumor HR status, which can be used to form a prognostic score for patients with BM from breast carcinoma. Ó 2007 Elsevier Inc. Brain metastases, Breast cancer, RTOG RPA classification, Prognostic factors, Lymphopenia.
(WBRT) is considered the standard treatment, providing symptom relief and increase median and overall survival (7–11). Despite the use of WBRT, the prognosis for patients with BM remains poor, with a median survival time of approximately 5 months (3, 10, 12–14). In 1997, the Radiation Therapy Oncology Group (RTOG) analyzed the relative contribution of pretreatment variables and the influence of treatment variations to the survival of patients with BM from all primary tumor types, to identify patient subgroups that could influence stratification and interpretation of clinical trials, regarding the possible impact of patient selection on benefits of
INTRODUCTION Brain metastases (BM) are an important cause of morbidity and mortality, occurring in approximatively 15% of patients with metastatic breast cancer (1, 2). A higher risk of developing BM after the diagnosis of breast carcinoma has been associated with pulmonary metastases, HER-2 overexpression, and estrogen receptor–negative tumors, with an incidence varying from 25% to 40% (1–6). For patients with BM occurring in an eloquent area or too large, numerous, or disseminated for surgery or radiosurgery, whole-brain radiotherapy
Conflict of interest: none. Acknowledgment—The authors thank Mme O. Nemec for her technical assistance. Received Jan 24, 2007, and in revised form April 12, 2007. Accepted for publication April 13, 2007.
Reprint requests to: Romuald Le Scodan, M.D., Department of Radiation Oncology, Centre Rene´ Huguenin, 35 rue Dailly, 92210 Saint Cloud, France. Tel: (+33) 1-47-11-23-59; Fax: (+33) 1-4711-15-87; E-mail: [email protected] Presented in part at the 42nd Annual Meeting of the American Society of Clinical Oncology, June 2–6, 2006, Atlanta, GA. 839
I. J. Radiation Oncology d Biology d Physics
840
new treatment strategies (3). In the RTOG study, the recursive partitioning analysis (RPA) estimates three prognostic classes, defined by age, Karnofsky performance status (KPS), and disease status, with longest median survival for RPA Class I patients (patients aged <65 years, KPS score $70, no other sites of metastases, and a controlled primary tumor). In this analysis, only 12% of patients had BM from breast carcinoma, and no general biologic factors or specific parameters for breast carcinoma, such as tumor hormone receptor (HR) status or HER-2 overexpression, were analyzed. The purpose of the study reported here was to evaluate the prognostic impact of RTOG RPA score and to determine independent general and biologic prognostic factors correlated with outcome. We demonstrate that RTOG RPA Class III, lymphopenia, and HR negative status are independent predictors of poor survival and can be used to construct a simple and specific prognostic score for patients with BM from breast carcinoma treated with WBRT. METHODS AND MATERIALS Patient characteristics at the time of BM diagnosis Between January 1998 and December 2003, 132 consecutive women with BM from breast carcinoma were treated at the Department of Radiation Oncology of the Centre Rene´ Huguenin (Saint Cloud, France). Fourteen patients underwent surgery, and 1 patient had stereotactic radiosurgery before WBRT. One hundred seventeen patients received WBRT only and were analyzed for the study. The median age at BM diagnosis was 53 years (mean, 54.9 years; range, 26–87 years), and 25.8% of patients were aged $65 years. The KPS was $70 in 52.6% of patients. Median time from initial breast cancer to BM diagnosis was 39 months (range, 0–282 months). Brain was the only site of metastases in 7 patients, and only 2 patients were in RTOG RPA Class I. Twenty-two patients (18.8%) had BM at tumor diagnosis. Of the 83 tested patients, 32 (38.6%) overexpressed HER-2, and 14 had received trastuzumab-based therapy before BM. Tumor characteristics and patient characteristics at the time of BM diagnosis are presented in Tables 1 and 2.
Treatment schedules All patients underwent computed tomography (CT) and/or magnetic resonance imaging (MRI) for diagnosis of BM. The median dose was 30 Gy in 10 3-Gy daily fractions (mean, 27.5 Gy; range, 0–44 Gy), distributed as follows: 30 Gy in 10 fractions in 94 patients; <30 Gy in 3-Gy daily fractions in 17 patients; and >30 Gy in 2-Gy daily fractions in 6 patients. All patients completed the full course of WBRT, except for 17 who discontinued the treatment because of deterioration in either systemic or brain disease. Six patients were given a boost with external beam RT to a dominant focus of disease. Most patients were administered corticosteroids before WBRT was started.
Prognostic factors identification for survival The following variables were examined to determine their prognostic value for overall survival after WBRT: age at BM, KPS (<70 vs. $70), RTOG RPA Class (I–II vs. III) according to Gaspar et al. (3), presence of extracranial systemic metastases, site of other systemic metastases (bone vs. lung vs. liver vs. multiple), number of BM (single vs. multiple), interval between primary tumor and BM (<2 years vs. $2 years), neurologic function score (0–2 vs. 3–4)
Volume 69, Number 3, 2007
Table 1. Tumor characteristics and patient characteristics at BM diagnosis Histology of primary carcinoma (n = 111) Ductal carcinoma 101 (91) Lobular carcinoma 6 (5.4) Other 4 (3.6) Tumor differentiation of primary carcinoma (n = 97) SBR I 2 (2.1) SBR II 38 (39.2) SBR III 57 (58.8) Tumor HR status (n = 112) Positive 65 (58) Negative 47 (42) HER-2 overexpression (n = 83) Yes 32 (38.6) No 51 (61.4) Age at BM diagnosis (y), median (range) 53 (26–87) <65 87 (74.3) $65 30 (25.7) Karnofsky performance status (n = 114) <70 54 (47.4) $70 60 (52.6) Neurologic function scale (n = 113) 0–2 57 (50.4) 3–4 56 (49.6) No. of BM Single 18 (7.7) Multiple 90 (76.9) Meningitis 9 (7.7) No. of other metastatic sites 0 (BM alone) 7 (6) 1 30 (25) 2 40 (34) >2 40 (34) Concomitant and other metastases Liver 56 (47.8) Lung 56 (47.8) Bone 58 (49.6) Other 61 (52.1) BM at breast carcinoma diagnosis 22 (18.8) Previous chemotherapy Yes 92 (78.6) No 25 (21.4) RTOG RPA class (n = 114) I 2 (1.7) II 58 (50.9) III 54 (47.4) Lactate dehydrogenase (n = 94) Range 127–3469 #500 U/L 47 (50) >500 U/L 47 (50) Lymphocyte count (n = 107) Range 40–3016 #0.7 g/L 43 (40.2) >0.7 g/L 64 (59.8) Vital status Alive 22 (18.8) Dead 95 (81.2) Abbreviations: BM = brain metastases; SBR = Scarff-BloomRichardson grade; HR = hormone receptor; RTOG = Radiation Therapy Oncology Group; RPA = recursive partitioning analysis. Values are number (percentage), unless otherwise noted. (3), total dose of WBRT (<30 Gy vs. $30 Gy), histology of primary breast carcinoma (ductal vs. lobular and other), Scarff-BloomRichardson grade, tumor HR status, lymphocyte count (#0.7 vs. >0.7 109/L), serum lactate dehydrogenase (LDH) level (#500
Brain metastases from breast carcinoma: A prognostic score d R. LE SCODAN et al.
0.8 0.0
100
150
time (months)
200
250
Fig. 1. Overall survival from breast cancer.
2-year survival rates with BM were 5 months (range, 0–52 months), 27.6% (95% CI, 19.9–36.8%), and 12% (95% CI, 6.5–21.2%), respectively (Fig. 2). Because most patients had active disease outside the brain, it was difficult to determine whether patients died of intracranial or extracranial metastatic disease.
1.0
Prognostic factors Univariate analyses. In the univariate analysis, patients with KPS <70 or RTOG RPA Class III, a 3–4 neurologic function score, negative HR status, serum LDH level >500 U/L, and lymphopenia at BM diagnosis (lymphocyte count #0.7 109/L) had significantly worse overall survival. No prognostic value was found for age at BM, site of other systemic metastases, prior chemotherapy, number of BM, interval between primary tumor and BM, total dose of WBRT,
0.2
Patients were followed every month during the 6 months after the end of treatment and every 2 months beyond 6 months. Overall survival with breast cancer was calculated from the time of initial diagnosis of breast cancer to death from any cause. Overall survival with BM was defined as the time from the date of BM to the date of death or date of last follow-up. Overall survival rates were calculated with the Kaplan-Meier method (16), and survival curves were compared with the log–rank test. The survival rates are presented with their 95% confidence intervals (CIs). Multivariate analysis was used to determine the effect of prognostic factors on overall survival using the Cox proportional hazards model in a backward stepwise procedure (17). Probability (p) values of #0.05 were considered significant. Missing values were not replaced. The construction of the prognostic model started with a univariate assessment of the prognostic effect of each factor. The score was based on the sum of the estimated coefficient for each prognostic factor in the final Cox model. Then the score was used to divide the population into three groups: the first had no pejorative factors, the second only one pejorative factor, and the third had more than one pejorative factor. Statistical analyses were performed using R software (R 2.0.1 Software; Free Software Foundation, Boston, MA).
50
0.8
Follow-up and statistical analysis
0
overall survival
vs. >500 U/L), and HER-2 overexpression. Karnofsky performance status was determined by reviewing each patient’s medical record. Karnofsky performance status was either documented numerically on the chart or was determined according to the Eastern Cooperative Oncology Group (ECOG) performance status, with patients with an ECOG score of 0–1 being considered as having a KPS $70. All biologic determinations at diagnosis were performed in the same laboratory. The upper limit of normal for LDH determination was 500 U/L. Lymphopenia was defined as <0.7 109 lymphocytes per liter at BM diagnosis, because this threshold had previously been shown to be important for predicting survival in patients with BM from breast carcinoma (15). The status of HER-2 in the tumor was assessed for 83 patients in the same laboratory, using a standardized testing procedure. Patients were considered to have HER-2 overexpression if they were 3+ on immunohistochemistry or 2+ on immunohistochemistry with gene amplification identified by fluorescence in situ hybridization testing.
0.2
Abbreviations as in Table 1. Values are number (percentage).
0.6
792 (66) 338 (28) 70 (6)
0.4
2 (1.7) 58 (50.9) 54 (47.4)
overall survival
RTOG database
0.6
I II III
Centre Rene´ Huguenin brain metastases patients
0.4
RTOG RPA class (n = 114)
1.0
Table 2. Comparison between the Centre Rene´ Huguenin brain metastases patients and the RTOG brain metastases database by RPA class
841
Patients outcomes Ninety-five patients were dead at last contact. Five-year and 10-year overall survival rates from initial breast cancer were 50.2% (95% CI, 41–59.3%) and 26% (95% CI, 18.2– 35.6%), respectively (Fig. 1). Median survival and 1- year and
0.0
RESULTS
0
10
20
30
40
time (months)
Fig. 2. Overall survival with brain metastases.
50
1.0 0.6
0.8
Log rank test, p=0.02
0.2
0.4
RTOG-RPA class 1, 2
RTOG-RPA class 3
0.0
histology of primary breast carcinoma, Scarff-BloomRichardson grade, and HER-2 overexpression (Table 3). The prognostic value of the presence of extracranial systemic metastases could not be investigated because only 7 patients had BM only in our series. Patients who had a KPS $70 at time of BM (n = 60) had a median survival of 8 months, compared with 3 months for those with KPS <70 (n = 54) (p = 0.02). Similarly, the first subset, consisting of 60 patients with RPA Class I–II at the time of BM diagnosis, had a median survival of 8 months, compared with 3 months for those with Class III (p = 0.02) (Fig. 3). The second subset, consisting of 65 patients with positive HR tumor status, had a median survival of 8 months, compared with 4 months for those with negative HR tumor status (n = 47) (p = 0.013) (Fig. 4). The third subset, consisting of 64 patients without lymphopenia at BM diagnosis, had a median survival of 7 months, compared with 2 months for those with lymphopenia (n = 43) (p = 0.0011) (Fig. 5). There was no significant correlation between lymphopenia at BM diagnosis and administration of prior chemotherapy. Multivariate analyses. Using multivariate analysis, the independent prognostic predictors for poor survival were RTOG RPA Class III or KPS <70 (p = 0.05), lymphopenia (p = 0.05), and negative HR status (p = 0.026) (Table 4). Our analysis of prognostic factors suggests that the prognosis of a patient at time of BM can be estimated according to several influential factors and that groups of patients with different prognoses can be defined, possibly requiring treatments of different intensities. Three prognostic groups were defined on the basis of factors identified in the multivariate analysis.
Volume 69, Number 3, 2007
overall survival
I. J. Radiation Oncology d Biology d Physics
842
0
10
20
30
40
50
time (months)
Fig. 3. Survival curves according to Radiation Therapy Oncology Group Recursive Partitioning Analysis (RTOG-RPA) class.
The good-prognosis group included patients with no adverse prognostic factors and had a median survival of 15 months (n = 23). The intermediate-prognosis group included patients with only one adverse prognostic factor and had a median survival of 5 months (n = 33). The poor-prognosis group included patients with more than one adverse prognostic factor and had a median survival of 3 months (n = 44). Median, 6month, and 1-year overall survivals according to prognostic score are presented in Table 5. Survival curves according to the three-factor prognostic score are presented in Fig. 6.
Table 3. Prognostic factors for poor survival in patients with brain metastases from breast cancer (univariate analyses)
0.02 0.45 (ns) 0.21 (ns) 0.26 (ns)
6.105
0.013
2.36 0.013
0.12 (ns) 0.91 (ns)
4.8 10.73 0.35 3.18 9.68
0.028 0.0011 0.55 (ns) 0.20 (ns) 0.0019
0.09
0.75 (ns)
5.42
0.02
1.0
5.42 0.559 1.58 2.67
Log rank test, p=0.013 0.8
0.68 (ns)
0.6
0.17
Response to WBRT Of the 117 patients, only 40 had at least one follow-up brain CT scan or MRI by which the radiographic response
0.4
p
overall survival
Alone vs. presence of systemic metastases KPS <70 vs. $70 Age at BM (y) <50 vs. $51 Age at BM (y) <65 vs. $65 Histology Ductal vs. lobular and other Tumor HR status Negative vs. positive SBR grade 1–2 vs. 3 HER-2 overexpression Negative vs. positive LDH (U/L) #500 vs. >500 Lymphocyte count #700 vs. >700 Time interval (y) <2 vs. $2 No. of lesions Single vs. multiple Neurologic function 0–2 vs. 3–4 scale Total radiation #30 vs. >30 dose (Gy) RTOG RPA class I–II vs. III
Log–rank
RH +
0.2
Brain metastases
Comparison
RH -
0.0
Covariate
0
Abbreviations: ns = not significant; KPS = Karnofsky performance status; LDH = lactate dehydrogenase. Other abbreviations as in Table 1.
10
20
30
40
50
time (months)
Fig. 4. Survival curves according to hormone receptor (RH) status.
Brain metastases from breast carcinoma: A prognostic score d R. LE SCODAN et al.
843
1.0
Table 5. Rates of overall survival at 6 months and 1 year according to the three-factor prognostic score Log rank test, p=0.0011
0.6
No adverse prognostic factors One prognostic factor More than one prognostic factor
0.4
Lymphocyte count > 0.7 109/ L
No. Median of survival patients (mo)
Overall survival at 6 mo (%)
Overall survival at 1 y (%)
23
15
76.1
60.9
33
5
38.6
20.7
44
3
29.5
15.9
Chi-square = Chi-square = 16.59 13.85 p = 0.00033 p = 0.0011
0.0
0.2
overall survival
0.8
Prognostic score
Lymphocyte count < 0.7 109/ L 0
10
20
30
40
50
time (months)
Fig. 5. Survival curves according to lymphopenia.
could be assessed. Seventeen patients (42.5%) with followup scans had brain failure, either local (original site) or distant (new brain lesions), and 1 was treated with stereotactic radiosurgery.
DISCUSSION Because several subgroups of patients with metastatic breast cancer are at high risk of developing BM (1, 4) and because systemic therapy, particularly trastuzumab, has limited efficacy to prevent or control intracranial metastases, BM from breast carcinoma are becoming a major problem associated with a detrimental impact on survival and quality of life (2, 18, 19). Identification of subgroups of patients with substantially different outcomes is, thus, mandatory to enable tailoring of therapy and to influence the design, stratification, and interpretation of future clinical trials. The following findings emerged from this study: (1) this study confirms the
prognostic value of the RTOG RPA score, (2) lymphopenia and HR tumor status are additional prognostic factors that predict overall survival at the time of BM diagnosis, and (3) using KPS, lymphopenia, and negative HR status, a prognostic score was developed classifying patients into three subgroups with significantly different overall survival. Whole-brain RT palliates symptoms and prolongs median survival time by approximately 4 months, but poor median survival rates continue to be reported in the range of 5 to 6 months, compared with 4–6 weeks for untreated patients (8–10). Gaspar et al. (3) reviewed the results of three RTOG studies, conducted to evaluate the effectiveness of different time–dose fractionation schemes or radiation sensitizers in the treatment of patients with BM from all primary tumors and reported on the use of RPA to determine prognostic classes in those patients. They proposed three prognostic classes defined by age, KPS, and disease status (3). The best survival (RPA Class I), with a median of 7.1 months, was
Table 4. Multivariate analyses of prognostic factors for poor overall survival from BM in 85 patients Covariate Tumor HR status Positive Negative Lymphocyte count >700 #700 RTOG RPA class I–II III
Coefficient (Cox)
Hazard ratio
95% CI
p
1 1.75
1.07–2.86
0.026
1 1.64
1.0–2.71
0.05
1 1.63
1.0–2.67
0.05
0.5589 0.4974 0.488
Abbreviation: CI = confidence interval. Other abbreviations as in Table 1.
Fig. 6. Survival curves according to the three-factor prognostic score.
844
I. J. Radiation Oncology d Biology d Physics
observed in patients aged <65 years, a KPS of $70, and a controlled primary tumor with the brain as the only site of metastases. The worst survival (RPA Class III), with a median of 2.3 months, was seen in patients with a KPS <70. Remaining patients (RPA Class II) had a median survival of 4.2 months. In this analysis, no general biologic factors or specific parameters for breast carcinoma, such as tumor HR status or HER-2 overexpression, were analyzed. In our study, prognostic independent predictors for poor survival were RTOG RPA Class III or KPS <70 (p = 0.05), lymphopenia (p = 0.05), and negative HR status (p = 0.026). Karnofsky performance status is the most commonly reported prognostic factor for survival for patients with BM from breast cancer (10, 15, 20, 21). Our data corroborate the RTOG RPA classification system in approximating the survival for breast cancer patients with BM. However, only 2 patients in our series were RPA Class I, compared with 66% in the RTOG database, because only 7 patients (6%) had BM only in our series. Therefore, only two groups (RPA Class I–II vs. RPA Class III) with different prognosis could be identified. In the Lyon experience, the analysis regarding the prognostic value of RTOG RPA classes was also performed with Classes I and II considered together, owing to the small number of patients in Class I (15). The majority of patients (50.9%) in our series were RPA Class II, and 47.4% of patients were Class III, compared with 28% and 6%, respectively, in the RTOG database, reflecting differences in patient selection. In the series reported by Claude et al. (15), only 20.3% of patients had BM only, and 57.6 % of patients with BM from breast cancer were RTOG RPA Class III. We believe that our population closely resembles those patients treated in daily practice and represents the natural history of BM from breast carcinoma. Relatively little information is available on the importance of biologic parameters in patients with BM from breast carcinoma. Claude et al. previously reported that lymphopenia was an independent prognostic factor for poor survival in cancer patients with BM, both for chemo-naı¨ve and previously treated patients (15, 22). Negative HR status is predictive for the occurrence of BM in patients with metastatic breast cancer (4, 23, 24) and also seems to be an independent prognostic factor in our series. In 1997 Wronski et al. (25) reported that negative HR status was an independent prognostic factor for patients with BM from breast carcinoma treated with surgery with or without WBRT, with a median survival of 12.5 months for 20 patients with negative HR status, compared with 21.9 months for 22 patients with positive HR status (p < 0.05). Estrogen receptor negative status was also an independent pejorative factor for 45 patients with BM from breast cancer treated at the University of California, San Francisco (26). Those two biologic parameters remain independent prognostic factors even when tested against RTOGRPA class or KPS in multivariate analysis. Although response rates after WBRT vary, partial or complete responses have been documented in more than 60% of patients in randomized controlled studies conducted by RTOG (11). Only 40 patients in our series had at least one
Volume 69, Number 3, 2007
follow-up brain CT scan or MRI, and because of the retrospective nature of our study and the potential selection of patients who were radiographically controlled, the failure rate (42.5%) reported here is possibly overestimated. Several trials tested the dose, timing, and fractionation of WBRT, without clear superiority of a particular treatment schedule (11). Thus, the most widely used WBRT regimen uses 30 Gy in 10 3-Gy fractions. However, this schedule is probably inadequate for long-term tumor control. In a recent analysis, Li et al. (27) studied 208 patients who received 10 daily fractions of 3-Gy WBRT in the control arm of the Phase III trial PCI-P120-9801 evaluating motexafin gadolinium. Neurocognitive function and survival were compared in 135 patients assessable at 2 months with tumor shrinkage below (poor responders) and above (good responders) the population median (45%) on MRI scans. In this report, good responders survived significantly longer than did poor responders (median survival, 300 26 vs. 240 19 days; p = 0.03), with a longer median time to neurocognitive function deterioration, supporting the use of techniques to maximize intracranial control. Our study confirms that RPA classes are valid for stratifying patients with BM from breast carcinoma but suggests that the addition of two biologic parameters may better define prognosis. We constructed a three-factor prognostic scoring system that predicts 6-month rates of overall survival in the range of 76.1–29.5% (p = 0.00033) and 1-year rates of overall survival in the range of 60.9–15.9% (p = 0.0011). Three groups of patients with different prognoses can be defined, and further studies of more aggressive therapy could be proposed for selected patients with no adverse prognostic factors, who could expect a median survival time of 15 months and a 1-year survival rate of 60.9%. We have not tested the value of this three-factor prognostic scoring system on an independent sample of patients with BM from breast carcinoma. A prospective validation of this prognostic score is ongoing in our institution, and we encourage other investigators to validate this score externally.
CONCLUSIONS This retrospective study of a relatively large number of patients confirms the value of established prognostic factors, such as the RTOG RPA score, and some less-wellrecognized factors, such as lymphopenia and tumor HR status. Our three-factor prognostic scoring system uses clinical characteristics that can be easily collected at the time of BM diagnosis and identifies three groups of patients with substantially different outcomes, particularly patients with no adverse prognostic factors, who can expect a median survival time of 15 months and a 1-year survival rate of 60%. The prognostic factors identified may be useful to tailor the therapy for subgroups of patients, to define homogeneous cohorts for prospective randomized trials, and to identify more precisely patients with relative good prognosis who could be treated with innovative approaches.
Brain metastases from breast carcinoma: A prognostic score d R. LE SCODAN et al.
845
REFERENCES 1. Gabos Z, Sinha R, Hanson J, et al. Prognostic significance of human epidermal growth factor receptor positivity for the development of brain metastasis after newly diagnosed breast cancer. J Clin Oncol 2006;24:5658–5663. 2. Miller KD, Weathers T, Haney LG, et al. Occult central nervous system involvement in patients with metastatic breast cancer: Prevalence, predictive factors and impact on overall survival. Ann Oncol 2003;14:1072–1077. 3. Gaspar L, Scott C, Rotman M, et al. Recursive partitioning analysis (RPA) of prognostic factors in three Radiation Therapy Oncology Group (RTOG) brain metastases trials. Int J Radiat Oncol Biol Phys 1997;37:745–751. 4. Slimane K, Andre F, Delaloge S, et al. Risk factors for brain relapse in patients with metastatic breast cancer. Ann Oncol 2004; 15:1640–1644. 5. Clayton AJ, Danson S, Jolly S, et al. Incidence of cerebral metastases in patients treated with trastuzumab for metastatic breast cancer. Br J Cancer 2004;91:639–643. 6. Bendell JC, Domchek SM, Burstein HJ, et al. Central nervous system metastases in women who receive trastuzumab-based therapy for metastatic breast carcinoma. Cancer 2003;97: 2972–2977. 7. Lin NU, Bellon JR, Winer EP. CNS metastases in breast cancer. J Clin Oncol 2004;22:3608–3617. 8. Fokstuen T, Wilking N, Rutqvist LE, et al. Radiation therapy in the management of brain metastases from breast cancer. Breast Cancer Res Treat 2000;62:211–216. 9. Mintz AH, Kestle J, Rathbone MP, et al. A randomized trial to assess the efficacy of surgery in addition to radiotherapy in patients with a single cerebral metastasis. Cancer 1996;78: 1470–1476. 10. Lentzsch S, Reichardt P, Weber F, et al. Brain metastases in breast cancer: Prognostic factors and management. Eur J Cancer 1999;35:580–585. 11. Khuntia D, Brown P, Li J, et al. Whole-brain radiotherapy in the management of brain metastasis. J Clin Oncol 2006;24: 1295–1304. 12. Sundstrom JT, Minn H, Lertola KK, et al. Prognosis of patients treated for intracranial metastases with whole-brain irradiation. Ann Med 1998;30:296–299. 13. Murray KJ, Scott C, Greenberg HM, et al. A randomized phase III study of accelerated hyperfractionation versus standard in patients with unresected brain metastases: A report of the Radiation Therapy Oncology Group (RTOG) 9104. Int J Radiat Oncol Biol Phys 1997;39:571–574.
14. Mehta MP, Rodrigus P, Terhaard CH, et al. Survival and neurologic outcomes in a randomized trial of motexafin gadolinium and whole-brain radiation therapy in brain metastases. J Clin Oncol 2003;21:2529–2536. 15. Claude L, Perol D, Ray-Coquard I, et al. Lymphopenia: A new independent prognostic factor for survival in patients treated with whole brain radiotherapy for brain metastases from breast carcinoma. Radiother Oncol 2005;76:334–339. 16. Kaplan EL, Meier P. Nonparametric estimation from incomplete observations. J Am Stat Assoc 1958;53:457–481. 17. Cox DR. Regression models and life-tables. J R Stat Soc B 1972;34:187–202. 18. Engel J, Eckel R, Aydemir U, et al. Determinants and prognoses of locoregional and distant progression in breast cancer. Int J Radiat Oncol Biol Phys 2003;55:1186–1195. 19. Chang J, Clark GM, Allred DC, et al. Survival of patients with metastatic breast carcinoma: Importance of prognostic markers of the primary tumor. Cancer 2003;97:545–553. 20. Lederman G, Wronski M, Fine M. Fractionated radiosurgery for brain metastases in 43 patients with breast carcinoma. Breast Cancer Res Treat 2001;65:145–154. 21. Mahmoud-Ahmed AS, Suh JH, Lee SY, et al. Results of whole brain radiotherapy in patients with brain metastases from breast cancer: A retrospective study. Int J Radiat Oncol Biol Phys 2002;54:810–817. 22. Ray-Coquard I, Van Glabbeke M, Negrier S. Lymphopenia as an independent prognostic factor for survival in advanced carcinoma, soft tissue carcinoma and lymphoma [Abstract]. Proc Am Soc Clin Oncol 2000;19:2591. 23. Samaan NA, Buzdar AU, Aldinger KA, et al. Estrogen receptor: A prognostic factor in breast cancer. Cancer 1981;47:554–560. 24. Stewart JF, King RJ, Sexton SA, et al. Oestrogen receptors, sites of metastatic disease and survival in recurrent breast cancer. Eur J Cancer 1981;17:449–453. 25. Wronski M, Arbit E, McCormick B. Surgical treatment of 70 patients with brain metastases from breast carcinoma. Cancer 1997;80:1746–1754. 26. Sneed P, Melisko M, Moore D, et al. Predicting survival in breast cancer patients with brain metastases [Abstract]. Proc Am Soc Clin Oncol 2003;22:168. 27. Li J, Bentzen SM, Renschler M, et al. Regression after wholebrain radiation therapy for brain metastases correlates with survival and improved neurocognitive function. J Clin Oncol 2007;25:1260–1266.
doi:10.1016/j.ijrobp.2007.04.024
CLINICAL INVESTIGATION
Brain
BRAIN METASTASES FROM BREAST CARCINOMA: VALIDATION OF THE RADIATION THERAPY ONCOLOGY GROUP RECURSIVE PARTITIONING ANALYSIS CLASSIFICATION AND PROPOSITION OF A NEW PROGNOSTIC SCORE ROMUALD LE SCODAN, M.D.,* CHRISTOPHE MASSARD, M.D.,* EMMANUELLE MOURET-FOURME, M.D.,y JEAN MARC GUINEBRETIERRE, M.D., PH.D.,z CHRISTINE COHEN-SOLAL, M.D.,* BRIGITTE DE LALANDE, M.D.,* PATRICIA MOISSON, M.D.,* CHRISTELLE BRETON-CALLU, M.D.,* MIRIAM GARDNER, M.D., PH.D.,* ALAIN GOUPIL, M.D.,x NICOLE RENODY, M.D.,* JEAN LOUIS FLOIRAS, M.D.,* AND ALAIN LABIB, M.D.* Departments of *Radiation Oncology, y Medical Statistics, z Pathology, and x Medical Oncology, Centre Rene´ Huguenin, Saint Cloud, France Purpose: To validate the Radiation Therapy Oncology Group Recursive Partitioning Analysis (RTOG RPA) classification and determine independent prognostic factors, to create a simple and specific prognostic score for patients with brain metastases (BM) from breast carcinoma treated with whole-brain radiotherapy (WBRT). Methods and Materials: From January 1998 through December 2003, 132 patients with BM from breast carcinoma were treated with WBRT. We analyzed several potential predictors of survival after WBRT: age, Karnofsky performance status, RTOG-RPA class, number of BM, presence and site of other systemic metastases, interval between primary tumor and BM, tumor hormone receptor (HR) status, lymphocyte count, and HER-2 overexpression. Results: A total of 117 patients received exclusive WBRT and were analyzed. Median survival with BM was 5 months. One-year and 2-year survival rates were 27.6% (95% confidence interval [CI] 19.9–36.8%) and 12% (95% CI 6.5–21.2%), respectively. In multivariate analysis, RTOG RPA Class III, lymphopenia (#0.7 109/L) and HR negative status were independent prognostic factors for poor survival. We constructed a three-factor prognostic scoring system that predicts 6-month and 1-year rates of overall survival in the range of 76.1–29.5% (p = 0.00033) and 60.9–15.9% (p = 0.0011), respectively, with median survival of 15 months, 5 months, or 3 months for patients with none, one, or more than one adverse prognostic factor(s), respectively. Conclusions: This study confirms the prognostic value of the RTOG RPA classification, lymphopenia, and tumor HR status, which can be used to form a prognostic score for patients with BM from breast carcinoma. Ó 2007 Elsevier Inc. Brain metastases, Breast cancer, RTOG RPA classification, Prognostic factors, Lymphopenia.
(WBRT) is considered the standard treatment, providing symptom relief and increase median and overall survival (7–11). Despite the use of WBRT, the prognosis for patients with BM remains poor, with a median survival time of approximately 5 months (3, 10, 12–14). In 1997, the Radiation Therapy Oncology Group (RTOG) analyzed the relative contribution of pretreatment variables and the influence of treatment variations to the survival of patients with BM from all primary tumor types, to identify patient subgroups that could influence stratification and interpretation of clinical trials, regarding the possible impact of patient selection on benefits of
INTRODUCTION Brain metastases (BM) are an important cause of morbidity and mortality, occurring in approximatively 15% of patients with metastatic breast cancer (1, 2). A higher risk of developing BM after the diagnosis of breast carcinoma has been associated with pulmonary metastases, HER-2 overexpression, and estrogen receptor–negative tumors, with an incidence varying from 25% to 40% (1–6). For patients with BM occurring in an eloquent area or too large, numerous, or disseminated for surgery or radiosurgery, whole-brain radiotherapy
Conflict of interest: none. Acknowledgment—The authors thank Mme O. Nemec for her technical assistance. Received Jan 24, 2007, and in revised form April 12, 2007. Accepted for publication April 13, 2007.
Reprint requests to: Romuald Le Scodan, M.D., Department of Radiation Oncology, Centre Rene´ Huguenin, 35 rue Dailly, 92210 Saint Cloud, France. Tel: (+33) 1-47-11-23-59; Fax: (+33) 1-4711-15-87; E-mail: [email protected] Presented in part at the 42nd Annual Meeting of the American Society of Clinical Oncology, June 2–6, 2006, Atlanta, GA. 839
I. J. Radiation Oncology d Biology d Physics
840
new treatment strategies (3). In the RTOG study, the recursive partitioning analysis (RPA) estimates three prognostic classes, defined by age, Karnofsky performance status (KPS), and disease status, with longest median survival for RPA Class I patients (patients aged <65 years, KPS score $70, no other sites of metastases, and a controlled primary tumor). In this analysis, only 12% of patients had BM from breast carcinoma, and no general biologic factors or specific parameters for breast carcinoma, such as tumor hormone receptor (HR) status or HER-2 overexpression, were analyzed. The purpose of the study reported here was to evaluate the prognostic impact of RTOG RPA score and to determine independent general and biologic prognostic factors correlated with outcome. We demonstrate that RTOG RPA Class III, lymphopenia, and HR negative status are independent predictors of poor survival and can be used to construct a simple and specific prognostic score for patients with BM from breast carcinoma treated with WBRT. METHODS AND MATERIALS Patient characteristics at the time of BM diagnosis Between January 1998 and December 2003, 132 consecutive women with BM from breast carcinoma were treated at the Department of Radiation Oncology of the Centre Rene´ Huguenin (Saint Cloud, France). Fourteen patients underwent surgery, and 1 patient had stereotactic radiosurgery before WBRT. One hundred seventeen patients received WBRT only and were analyzed for the study. The median age at BM diagnosis was 53 years (mean, 54.9 years; range, 26–87 years), and 25.8% of patients were aged $65 years. The KPS was $70 in 52.6% of patients. Median time from initial breast cancer to BM diagnosis was 39 months (range, 0–282 months). Brain was the only site of metastases in 7 patients, and only 2 patients were in RTOG RPA Class I. Twenty-two patients (18.8%) had BM at tumor diagnosis. Of the 83 tested patients, 32 (38.6%) overexpressed HER-2, and 14 had received trastuzumab-based therapy before BM. Tumor characteristics and patient characteristics at the time of BM diagnosis are presented in Tables 1 and 2.
Treatment schedules All patients underwent computed tomography (CT) and/or magnetic resonance imaging (MRI) for diagnosis of BM. The median dose was 30 Gy in 10 3-Gy daily fractions (mean, 27.5 Gy; range, 0–44 Gy), distributed as follows: 30 Gy in 10 fractions in 94 patients; <30 Gy in 3-Gy daily fractions in 17 patients; and >30 Gy in 2-Gy daily fractions in 6 patients. All patients completed the full course of WBRT, except for 17 who discontinued the treatment because of deterioration in either systemic or brain disease. Six patients were given a boost with external beam RT to a dominant focus of disease. Most patients were administered corticosteroids before WBRT was started.
Prognostic factors identification for survival The following variables were examined to determine their prognostic value for overall survival after WBRT: age at BM, KPS (<70 vs. $70), RTOG RPA Class (I–II vs. III) according to Gaspar et al. (3), presence of extracranial systemic metastases, site of other systemic metastases (bone vs. lung vs. liver vs. multiple), number of BM (single vs. multiple), interval between primary tumor and BM (<2 years vs. $2 years), neurologic function score (0–2 vs. 3–4)
Volume 69, Number 3, 2007
Table 1. Tumor characteristics and patient characteristics at BM diagnosis Histology of primary carcinoma (n = 111) Ductal carcinoma 101 (91) Lobular carcinoma 6 (5.4) Other 4 (3.6) Tumor differentiation of primary carcinoma (n = 97) SBR I 2 (2.1) SBR II 38 (39.2) SBR III 57 (58.8) Tumor HR status (n = 112) Positive 65 (58) Negative 47 (42) HER-2 overexpression (n = 83) Yes 32 (38.6) No 51 (61.4) Age at BM diagnosis (y), median (range) 53 (26–87) <65 87 (74.3) $65 30 (25.7) Karnofsky performance status (n = 114) <70 54 (47.4) $70 60 (52.6) Neurologic function scale (n = 113) 0–2 57 (50.4) 3–4 56 (49.6) No. of BM Single 18 (7.7) Multiple 90 (76.9) Meningitis 9 (7.7) No. of other metastatic sites 0 (BM alone) 7 (6) 1 30 (25) 2 40 (34) >2 40 (34) Concomitant and other metastases Liver 56 (47.8) Lung 56 (47.8) Bone 58 (49.6) Other 61 (52.1) BM at breast carcinoma diagnosis 22 (18.8) Previous chemotherapy Yes 92 (78.6) No 25 (21.4) RTOG RPA class (n = 114) I 2 (1.7) II 58 (50.9) III 54 (47.4) Lactate dehydrogenase (n = 94) Range 127–3469 #500 U/L 47 (50) >500 U/L 47 (50) Lymphocyte count (n = 107) Range 40–3016 #0.7 g/L 43 (40.2) >0.7 g/L 64 (59.8) Vital status Alive 22 (18.8) Dead 95 (81.2) Abbreviations: BM = brain metastases; SBR = Scarff-BloomRichardson grade; HR = hormone receptor; RTOG = Radiation Therapy Oncology Group; RPA = recursive partitioning analysis. Values are number (percentage), unless otherwise noted. (3), total dose of WBRT (<30 Gy vs. $30 Gy), histology of primary breast carcinoma (ductal vs. lobular and other), Scarff-BloomRichardson grade, tumor HR status, lymphocyte count (#0.7 vs. >0.7 109/L), serum lactate dehydrogenase (LDH) level (#500
Brain metastases from breast carcinoma: A prognostic score d R. LE SCODAN et al.
0.8 0.0
100
150
time (months)
200
250
Fig. 1. Overall survival from breast cancer.
2-year survival rates with BM were 5 months (range, 0–52 months), 27.6% (95% CI, 19.9–36.8%), and 12% (95% CI, 6.5–21.2%), respectively (Fig. 2). Because most patients had active disease outside the brain, it was difficult to determine whether patients died of intracranial or extracranial metastatic disease.
1.0
Prognostic factors Univariate analyses. In the univariate analysis, patients with KPS <70 or RTOG RPA Class III, a 3–4 neurologic function score, negative HR status, serum LDH level >500 U/L, and lymphopenia at BM diagnosis (lymphocyte count #0.7 109/L) had significantly worse overall survival. No prognostic value was found for age at BM, site of other systemic metastases, prior chemotherapy, number of BM, interval between primary tumor and BM, total dose of WBRT,
0.2
Patients were followed every month during the 6 months after the end of treatment and every 2 months beyond 6 months. Overall survival with breast cancer was calculated from the time of initial diagnosis of breast cancer to death from any cause. Overall survival with BM was defined as the time from the date of BM to the date of death or date of last follow-up. Overall survival rates were calculated with the Kaplan-Meier method (16), and survival curves were compared with the log–rank test. The survival rates are presented with their 95% confidence intervals (CIs). Multivariate analysis was used to determine the effect of prognostic factors on overall survival using the Cox proportional hazards model in a backward stepwise procedure (17). Probability (p) values of #0.05 were considered significant. Missing values were not replaced. The construction of the prognostic model started with a univariate assessment of the prognostic effect of each factor. The score was based on the sum of the estimated coefficient for each prognostic factor in the final Cox model. Then the score was used to divide the population into three groups: the first had no pejorative factors, the second only one pejorative factor, and the third had more than one pejorative factor. Statistical analyses were performed using R software (R 2.0.1 Software; Free Software Foundation, Boston, MA).
50
0.8
Follow-up and statistical analysis
0
overall survival
vs. >500 U/L), and HER-2 overexpression. Karnofsky performance status was determined by reviewing each patient’s medical record. Karnofsky performance status was either documented numerically on the chart or was determined according to the Eastern Cooperative Oncology Group (ECOG) performance status, with patients with an ECOG score of 0–1 being considered as having a KPS $70. All biologic determinations at diagnosis were performed in the same laboratory. The upper limit of normal for LDH determination was 500 U/L. Lymphopenia was defined as <0.7 109 lymphocytes per liter at BM diagnosis, because this threshold had previously been shown to be important for predicting survival in patients with BM from breast carcinoma (15). The status of HER-2 in the tumor was assessed for 83 patients in the same laboratory, using a standardized testing procedure. Patients were considered to have HER-2 overexpression if they were 3+ on immunohistochemistry or 2+ on immunohistochemistry with gene amplification identified by fluorescence in situ hybridization testing.
0.2
Abbreviations as in Table 1. Values are number (percentage).
0.6
792 (66) 338 (28) 70 (6)
0.4
2 (1.7) 58 (50.9) 54 (47.4)
overall survival
RTOG database
0.6
I II III
Centre Rene´ Huguenin brain metastases patients
0.4
RTOG RPA class (n = 114)
1.0
Table 2. Comparison between the Centre Rene´ Huguenin brain metastases patients and the RTOG brain metastases database by RPA class
841
Patients outcomes Ninety-five patients were dead at last contact. Five-year and 10-year overall survival rates from initial breast cancer were 50.2% (95% CI, 41–59.3%) and 26% (95% CI, 18.2– 35.6%), respectively (Fig. 1). Median survival and 1- year and
0.0
RESULTS
0
10
20
30
40
time (months)
Fig. 2. Overall survival with brain metastases.
50
1.0 0.6
0.8
Log rank test, p=0.02
0.2
0.4
RTOG-RPA class 1, 2
RTOG-RPA class 3
0.0
histology of primary breast carcinoma, Scarff-BloomRichardson grade, and HER-2 overexpression (Table 3). The prognostic value of the presence of extracranial systemic metastases could not be investigated because only 7 patients had BM only in our series. Patients who had a KPS $70 at time of BM (n = 60) had a median survival of 8 months, compared with 3 months for those with KPS <70 (n = 54) (p = 0.02). Similarly, the first subset, consisting of 60 patients with RPA Class I–II at the time of BM diagnosis, had a median survival of 8 months, compared with 3 months for those with Class III (p = 0.02) (Fig. 3). The second subset, consisting of 65 patients with positive HR tumor status, had a median survival of 8 months, compared with 4 months for those with negative HR tumor status (n = 47) (p = 0.013) (Fig. 4). The third subset, consisting of 64 patients without lymphopenia at BM diagnosis, had a median survival of 7 months, compared with 2 months for those with lymphopenia (n = 43) (p = 0.0011) (Fig. 5). There was no significant correlation between lymphopenia at BM diagnosis and administration of prior chemotherapy. Multivariate analyses. Using multivariate analysis, the independent prognostic predictors for poor survival were RTOG RPA Class III or KPS <70 (p = 0.05), lymphopenia (p = 0.05), and negative HR status (p = 0.026) (Table 4). Our analysis of prognostic factors suggests that the prognosis of a patient at time of BM can be estimated according to several influential factors and that groups of patients with different prognoses can be defined, possibly requiring treatments of different intensities. Three prognostic groups were defined on the basis of factors identified in the multivariate analysis.
Volume 69, Number 3, 2007
overall survival
I. J. Radiation Oncology d Biology d Physics
842
0
10
20
30
40
50
time (months)
Fig. 3. Survival curves according to Radiation Therapy Oncology Group Recursive Partitioning Analysis (RTOG-RPA) class.
The good-prognosis group included patients with no adverse prognostic factors and had a median survival of 15 months (n = 23). The intermediate-prognosis group included patients with only one adverse prognostic factor and had a median survival of 5 months (n = 33). The poor-prognosis group included patients with more than one adverse prognostic factor and had a median survival of 3 months (n = 44). Median, 6month, and 1-year overall survivals according to prognostic score are presented in Table 5. Survival curves according to the three-factor prognostic score are presented in Fig. 6.
Table 3. Prognostic factors for poor survival in patients with brain metastases from breast cancer (univariate analyses)
0.02 0.45 (ns) 0.21 (ns) 0.26 (ns)
6.105
0.013
2.36 0.013
0.12 (ns) 0.91 (ns)
4.8 10.73 0.35 3.18 9.68
0.028 0.0011 0.55 (ns) 0.20 (ns) 0.0019
0.09
0.75 (ns)
5.42
0.02
1.0
5.42 0.559 1.58 2.67
Log rank test, p=0.013 0.8
0.68 (ns)
0.6
0.17
Response to WBRT Of the 117 patients, only 40 had at least one follow-up brain CT scan or MRI by which the radiographic response
0.4
p
overall survival
Alone vs. presence of systemic metastases KPS <70 vs. $70 Age at BM (y) <50 vs. $51 Age at BM (y) <65 vs. $65 Histology Ductal vs. lobular and other Tumor HR status Negative vs. positive SBR grade 1–2 vs. 3 HER-2 overexpression Negative vs. positive LDH (U/L) #500 vs. >500 Lymphocyte count #700 vs. >700 Time interval (y) <2 vs. $2 No. of lesions Single vs. multiple Neurologic function 0–2 vs. 3–4 scale Total radiation #30 vs. >30 dose (Gy) RTOG RPA class I–II vs. III
Log–rank
RH +
0.2
Brain metastases
Comparison
RH -
0.0
Covariate
0
Abbreviations: ns = not significant; KPS = Karnofsky performance status; LDH = lactate dehydrogenase. Other abbreviations as in Table 1.
10
20
30
40
50
time (months)
Fig. 4. Survival curves according to hormone receptor (RH) status.
Brain metastases from breast carcinoma: A prognostic score d R. LE SCODAN et al.
843
1.0
Table 5. Rates of overall survival at 6 months and 1 year according to the three-factor prognostic score Log rank test, p=0.0011
0.6
No adverse prognostic factors One prognostic factor More than one prognostic factor
0.4
Lymphocyte count > 0.7 109/ L
No. Median of survival patients (mo)
Overall survival at 6 mo (%)
Overall survival at 1 y (%)
23
15
76.1
60.9
33
5
38.6
20.7
44
3
29.5
15.9
Chi-square = Chi-square = 16.59 13.85 p = 0.00033 p = 0.0011
0.0
0.2
overall survival
0.8
Prognostic score
Lymphocyte count < 0.7 109/ L 0
10
20
30
40
50
time (months)
Fig. 5. Survival curves according to lymphopenia.
could be assessed. Seventeen patients (42.5%) with followup scans had brain failure, either local (original site) or distant (new brain lesions), and 1 was treated with stereotactic radiosurgery.
DISCUSSION Because several subgroups of patients with metastatic breast cancer are at high risk of developing BM (1, 4) and because systemic therapy, particularly trastuzumab, has limited efficacy to prevent or control intracranial metastases, BM from breast carcinoma are becoming a major problem associated with a detrimental impact on survival and quality of life (2, 18, 19). Identification of subgroups of patients with substantially different outcomes is, thus, mandatory to enable tailoring of therapy and to influence the design, stratification, and interpretation of future clinical trials. The following findings emerged from this study: (1) this study confirms the
prognostic value of the RTOG RPA score, (2) lymphopenia and HR tumor status are additional prognostic factors that predict overall survival at the time of BM diagnosis, and (3) using KPS, lymphopenia, and negative HR status, a prognostic score was developed classifying patients into three subgroups with significantly different overall survival. Whole-brain RT palliates symptoms and prolongs median survival time by approximately 4 months, but poor median survival rates continue to be reported in the range of 5 to 6 months, compared with 4–6 weeks for untreated patients (8–10). Gaspar et al. (3) reviewed the results of three RTOG studies, conducted to evaluate the effectiveness of different time–dose fractionation schemes or radiation sensitizers in the treatment of patients with BM from all primary tumors and reported on the use of RPA to determine prognostic classes in those patients. They proposed three prognostic classes defined by age, KPS, and disease status (3). The best survival (RPA Class I), with a median of 7.1 months, was
Table 4. Multivariate analyses of prognostic factors for poor overall survival from BM in 85 patients Covariate Tumor HR status Positive Negative Lymphocyte count >700 #700 RTOG RPA class I–II III
Coefficient (Cox)
Hazard ratio
95% CI
p
1 1.75
1.07–2.86
0.026
1 1.64
1.0–2.71
0.05
1 1.63
1.0–2.67
0.05
0.5589 0.4974 0.488
Abbreviation: CI = confidence interval. Other abbreviations as in Table 1.
Fig. 6. Survival curves according to the three-factor prognostic score.
844
I. J. Radiation Oncology d Biology d Physics
observed in patients aged <65 years, a KPS of $70, and a controlled primary tumor with the brain as the only site of metastases. The worst survival (RPA Class III), with a median of 2.3 months, was seen in patients with a KPS <70. Remaining patients (RPA Class II) had a median survival of 4.2 months. In this analysis, no general biologic factors or specific parameters for breast carcinoma, such as tumor HR status or HER-2 overexpression, were analyzed. In our study, prognostic independent predictors for poor survival were RTOG RPA Class III or KPS <70 (p = 0.05), lymphopenia (p = 0.05), and negative HR status (p = 0.026). Karnofsky performance status is the most commonly reported prognostic factor for survival for patients with BM from breast cancer (10, 15, 20, 21). Our data corroborate the RTOG RPA classification system in approximating the survival for breast cancer patients with BM. However, only 2 patients in our series were RPA Class I, compared with 66% in the RTOG database, because only 7 patients (6%) had BM only in our series. Therefore, only two groups (RPA Class I–II vs. RPA Class III) with different prognosis could be identified. In the Lyon experience, the analysis regarding the prognostic value of RTOG RPA classes was also performed with Classes I and II considered together, owing to the small number of patients in Class I (15). The majority of patients (50.9%) in our series were RPA Class II, and 47.4% of patients were Class III, compared with 28% and 6%, respectively, in the RTOG database, reflecting differences in patient selection. In the series reported by Claude et al. (15), only 20.3% of patients had BM only, and 57.6 % of patients with BM from breast cancer were RTOG RPA Class III. We believe that our population closely resembles those patients treated in daily practice and represents the natural history of BM from breast carcinoma. Relatively little information is available on the importance of biologic parameters in patients with BM from breast carcinoma. Claude et al. previously reported that lymphopenia was an independent prognostic factor for poor survival in cancer patients with BM, both for chemo-naı¨ve and previously treated patients (15, 22). Negative HR status is predictive for the occurrence of BM in patients with metastatic breast cancer (4, 23, 24) and also seems to be an independent prognostic factor in our series. In 1997 Wronski et al. (25) reported that negative HR status was an independent prognostic factor for patients with BM from breast carcinoma treated with surgery with or without WBRT, with a median survival of 12.5 months for 20 patients with negative HR status, compared with 21.9 months for 22 patients with positive HR status (p < 0.05). Estrogen receptor negative status was also an independent pejorative factor for 45 patients with BM from breast cancer treated at the University of California, San Francisco (26). Those two biologic parameters remain independent prognostic factors even when tested against RTOGRPA class or KPS in multivariate analysis. Although response rates after WBRT vary, partial or complete responses have been documented in more than 60% of patients in randomized controlled studies conducted by RTOG (11). Only 40 patients in our series had at least one
Volume 69, Number 3, 2007
follow-up brain CT scan or MRI, and because of the retrospective nature of our study and the potential selection of patients who were radiographically controlled, the failure rate (42.5%) reported here is possibly overestimated. Several trials tested the dose, timing, and fractionation of WBRT, without clear superiority of a particular treatment schedule (11). Thus, the most widely used WBRT regimen uses 30 Gy in 10 3-Gy fractions. However, this schedule is probably inadequate for long-term tumor control. In a recent analysis, Li et al. (27) studied 208 patients who received 10 daily fractions of 3-Gy WBRT in the control arm of the Phase III trial PCI-P120-9801 evaluating motexafin gadolinium. Neurocognitive function and survival were compared in 135 patients assessable at 2 months with tumor shrinkage below (poor responders) and above (good responders) the population median (45%) on MRI scans. In this report, good responders survived significantly longer than did poor responders (median survival, 300 26 vs. 240 19 days; p = 0.03), with a longer median time to neurocognitive function deterioration, supporting the use of techniques to maximize intracranial control. Our study confirms that RPA classes are valid for stratifying patients with BM from breast carcinoma but suggests that the addition of two biologic parameters may better define prognosis. We constructed a three-factor prognostic scoring system that predicts 6-month rates of overall survival in the range of 76.1–29.5% (p = 0.00033) and 1-year rates of overall survival in the range of 60.9–15.9% (p = 0.0011). Three groups of patients with different prognoses can be defined, and further studies of more aggressive therapy could be proposed for selected patients with no adverse prognostic factors, who could expect a median survival time of 15 months and a 1-year survival rate of 60.9%. We have not tested the value of this three-factor prognostic scoring system on an independent sample of patients with BM from breast carcinoma. A prospective validation of this prognostic score is ongoing in our institution, and we encourage other investigators to validate this score externally.
CONCLUSIONS This retrospective study of a relatively large number of patients confirms the value of established prognostic factors, such as the RTOG RPA score, and some less-wellrecognized factors, such as lymphopenia and tumor HR status. Our three-factor prognostic scoring system uses clinical characteristics that can be easily collected at the time of BM diagnosis and identifies three groups of patients with substantially different outcomes, particularly patients with no adverse prognostic factors, who can expect a median survival time of 15 months and a 1-year survival rate of 60%. The prognostic factors identified may be useful to tailor the therapy for subgroups of patients, to define homogeneous cohorts for prospective randomized trials, and to identify more precisely patients with relative good prognosis who could be treated with innovative approaches.
Brain metastases from breast carcinoma: A prognostic score d R. LE SCODAN et al.
845
REFERENCES 1. Gabos Z, Sinha R, Hanson J, et al. Prognostic significance of human epidermal growth factor receptor positivity for the development of brain metastasis after newly diagnosed breast cancer. J Clin Oncol 2006;24:5658–5663. 2. Miller KD, Weathers T, Haney LG, et al. Occult central nervous system involvement in patients with metastatic breast cancer: Prevalence, predictive factors and impact on overall survival. Ann Oncol 2003;14:1072–1077. 3. Gaspar L, Scott C, Rotman M, et al. Recursive partitioning analysis (RPA) of prognostic factors in three Radiation Therapy Oncology Group (RTOG) brain metastases trials. Int J Radiat Oncol Biol Phys 1997;37:745–751. 4. Slimane K, Andre F, Delaloge S, et al. Risk factors for brain relapse in patients with metastatic breast cancer. Ann Oncol 2004; 15:1640–1644. 5. Clayton AJ, Danson S, Jolly S, et al. Incidence of cerebral metastases in patients treated with trastuzumab for metastatic breast cancer. Br J Cancer 2004;91:639–643. 6. Bendell JC, Domchek SM, Burstein HJ, et al. Central nervous system metastases in women who receive trastuzumab-based therapy for metastatic breast carcinoma. Cancer 2003;97: 2972–2977. 7. Lin NU, Bellon JR, Winer EP. CNS metastases in breast cancer. J Clin Oncol 2004;22:3608–3617. 8. Fokstuen T, Wilking N, Rutqvist LE, et al. Radiation therapy in the management of brain metastases from breast cancer. Breast Cancer Res Treat 2000;62:211–216. 9. Mintz AH, Kestle J, Rathbone MP, et al. A randomized trial to assess the efficacy of surgery in addition to radiotherapy in patients with a single cerebral metastasis. Cancer 1996;78: 1470–1476. 10. Lentzsch S, Reichardt P, Weber F, et al. Brain metastases in breast cancer: Prognostic factors and management. Eur J Cancer 1999;35:580–585. 11. Khuntia D, Brown P, Li J, et al. Whole-brain radiotherapy in the management of brain metastasis. J Clin Oncol 2006;24: 1295–1304. 12. Sundstrom JT, Minn H, Lertola KK, et al. Prognosis of patients treated for intracranial metastases with whole-brain irradiation. Ann Med 1998;30:296–299. 13. Murray KJ, Scott C, Greenberg HM, et al. A randomized phase III study of accelerated hyperfractionation versus standard in patients with unresected brain metastases: A report of the Radiation Therapy Oncology Group (RTOG) 9104. Int J Radiat Oncol Biol Phys 1997;39:571–574.
14. Mehta MP, Rodrigus P, Terhaard CH, et al. Survival and neurologic outcomes in a randomized trial of motexafin gadolinium and whole-brain radiation therapy in brain metastases. J Clin Oncol 2003;21:2529–2536. 15. Claude L, Perol D, Ray-Coquard I, et al. Lymphopenia: A new independent prognostic factor for survival in patients treated with whole brain radiotherapy for brain metastases from breast carcinoma. Radiother Oncol 2005;76:334–339. 16. Kaplan EL, Meier P. Nonparametric estimation from incomplete observations. J Am Stat Assoc 1958;53:457–481. 17. Cox DR. Regression models and life-tables. J R Stat Soc B 1972;34:187–202. 18. Engel J, Eckel R, Aydemir U, et al. Determinants and prognoses of locoregional and distant progression in breast cancer. Int J Radiat Oncol Biol Phys 2003;55:1186–1195. 19. Chang J, Clark GM, Allred DC, et al. Survival of patients with metastatic breast carcinoma: Importance of prognostic markers of the primary tumor. Cancer 2003;97:545–553. 20. Lederman G, Wronski M, Fine M. Fractionated radiosurgery for brain metastases in 43 patients with breast carcinoma. Breast Cancer Res Treat 2001;65:145–154. 21. Mahmoud-Ahmed AS, Suh JH, Lee SY, et al. Results of whole brain radiotherapy in patients with brain metastases from breast cancer: A retrospective study. Int J Radiat Oncol Biol Phys 2002;54:810–817. 22. Ray-Coquard I, Van Glabbeke M, Negrier S. Lymphopenia as an independent prognostic factor for survival in advanced carcinoma, soft tissue carcinoma and lymphoma [Abstract]. Proc Am Soc Clin Oncol 2000;19:2591. 23. Samaan NA, Buzdar AU, Aldinger KA, et al. Estrogen receptor: A prognostic factor in breast cancer. Cancer 1981;47:554–560. 24. Stewart JF, King RJ, Sexton SA, et al. Oestrogen receptors, sites of metastatic disease and survival in recurrent breast cancer. Eur J Cancer 1981;17:449–453. 25. Wronski M, Arbit E, McCormick B. Surgical treatment of 70 patients with brain metastases from breast carcinoma. Cancer 1997;80:1746–1754. 26. Sneed P, Melisko M, Moore D, et al. Predicting survival in breast cancer patients with brain metastases [Abstract]. Proc Am Soc Clin Oncol 2003;22:168. 27. Li J, Bentzen SM, Renschler M, et al. Regression after wholebrain radiation therapy for brain metastases correlates with survival and improved neurocognitive function. J Clin Oncol 2007;25:1260–1266.