Occult central nervous system involvement in patients with metastatic breast cancer: prevalence, predictive factors and impact on overall survival

K. D. Miller1,+, T. Weathers1, L. G. Haney1, R. Timmerman2, M. Dickler3, J. Shen4 and G. W. Sledge Jr1

1 Division of Hematology and Oncology, 2 Department of Radiation Oncology, and 4 Division of Biostatistics, Indiana University, Indianapolis, IN; 3 Breast Medicine Service, Memorial Sloan Kettering Cancer Center, New York, NY, USA

Received 18 July 2002; revised 20 January 2003; accepted 14 March 2003


    Abstract
 Top
 Abstract
 Introduction
 Patients and methods
 Results
 Discussion
 References
 
Background:

As screening central nervous system (CNS) imaging is not routinely performed, the incidence and clinical relevance of occult CNS metastases in advanced breast cancer is unknown.

Patients and methods:

All patients screened for participation in one of four clinical trials were included; each of the trials excluded patients with known CNS involvement and required screening CNS imaging. A cohort of breast cancer patients with symptomatic CNS metastases was identified from the IU Cancer Center Tumor Registry for comparison.

Results:

From November 1998 to August 2001, 155 screening imaging studies were performed. Twenty-three patients (14.8%) had occult CNS metastases. HER-2 overexpression (P = 0.02) and number of metastatic sites (P = 0.03) were predictive of CNS involvement by multivariate analysis. Median survival from time of metastasis (1.78 versus 2.76 years; P <0.0001) and from screening (4.67 versus 10.4 months; P = 0.0013) was shorter in patients with than without occult CNS metastasis. Survival among patients with occult CNS metastasis was similar to patients with symptomatic CNS disease.

Conclusions:

Patients with CNS involvement, whether occult or symptomatic, have an impaired survival. Occult CNS metastasis is relatively common, but impact on survival of treating occult CNS disease in patients with progressive systemic metastases is questionable.

Key words: breast cancer, central nervous system, metastasis, natural history


    Introduction
 Top
 Abstract
 Introduction
 Patients and methods
 Results
 Discussion
 References
 
Clinically evident central nervous system (CNS) metastases are expected in about 10–15% of patients with metastatic breast cancer. Survival after the identification of symptomatic CNS metastases is generally short with a median survival of ~4 months; 2-year survival is <2% [1, 2]. Younger age, single metastasis, surgical resection, whole brain radiation therapy and chemotherapy have been associated with prolonged survival in multivariate analyses. Despite the poor prognosis associated with symptomatic CNS metastasis, most patients ultimately die of systemic disease progression [3].

Autopsy series, generally completed before the advent of effective systemic therapies, report a much higher incidence of CNS metastasis (Table 1). As routine CNS imaging is not recommended for asymptomatic patients and autopsy rates have declined dramatically [4], the incidence of clinically occult CNS disease in patients with metastatic breast cancer receiving modern systemic therapy is unknown. Similarly, the clinical relevance of occult CNS metastasis, survival after identification of occult CNS disease and the importance of CNS therapy are unknown.


View this table:
[in this window]
[in a new window]
 
Table 1. Incidence of central nervous system (CNS) metastasis in breast cancer based on autopsy series
 
As angiogenesis is required for local growth and dissemination of breast cancer, inhibition of angiogenesis offers an attractive therapeutic target. Although expected to have minimal toxicity, early clinical trials identified bleeding at previously unrecognized CNS metastases [5] as a potential serious complication of antiangiogenic therapy. To ensure patient safety, most subsequent trials excluded patients with any CNS involvement. The requirement for CNS imaging as a criterion for enrollment in trials of novel antiangiogenic agents provides a unique opportunity to investigate the prevalence and clinical relevance of occult CNS metastasis in patients with advanced breast cancer.


    Patients and methods
 Top
 Abstract
 Introduction
 Patients and methods
 Results
 Discussion
 References
 
All patients screened for participation in one of four clinical trials evaluating novel antiangiogenic agents at Indiana University or Memorial Sloan Kettering Cancer Center were included in the analysis (Table 2). Each of the trials excluded patients with known CNS involvement and required either head computed tomography (CT) scan or brain magnetic resonance imaging (MRI) prior to enrollment. Local institutional review boards approved all protocols; patients provided informed consent prior to CNS imaging. Demographics, disease characteristics, treatment history and survival were abstracted from the medical record; the referring oncologist confirmed survival when needed. A consecutive cohort of breast cancer patients with symptomatic CNS metastasis treated at Indiana University from 1996 to 2001 was identified for survival comparisons. As details of prior systemic therapy, steroid receptor status and HER-2 overexpression were not routinely available, symptomatic patients were not included in the analysis of predictive factors.


View this table:
[in this window]
[in a new window]
 
Table 2. Trials contributing patients to analysis
 
Logistic regression was used to identify predictive factors for occult CNS involvement. CNS metastasis (present versus not present) was considered a binary response variable. An initial univariate analysis was conducted to screen for potential predictive factors; all factors reaching significance at the 0.15 level were included in the multivariate model. Wald chi-square tests and odds ratios along with corresponding 95% confidence intervals were calculated. Three different survival intervals were analyzed: survival from initial diagnosis of breast cancer, survival from first distant metastasis and survival from on-study screening CNS imaging. Survival curves were constructed according to the method of Kaplan and Meier. Log-rank tests were employed to compare survival among the three patient cohorts (screened, no CNS metastasis; screened, occult CNS metastasis; symptomatic CNS metastasis).


    Results
 Top
 Abstract
 Introduction
 Patients and methods
 Results
 Discussion
 References
 
From November 1998 to August 2001, 155 screening imaging studies were performed; 11 patients were screened twice. Median age at diagnosis was 42.7 (range 26–76) years. Median time from first metastasis to screening was 12 months. Patients had received a median of three prior chemotherapies (range 0–13). Twenty-three patients (14.8%) had occult CNS metastases (Table 3). Identification of occult CNS metastases by CT (12 of 86 patients) and MRI (11 of 69 patients) was similar. Screening identified single parenchymal lesions in nine patients; 12 patients had multiple lesions. Isolated leptomeningeal or dural involvement was found in two patients. Seventy-three consecutive patients with symptomatic CNS metastasis referred for radiation therapy from 1996 to 2001 comprised the symptomatic cohort. Screened patients were slightly younger, likely representing bias in referral for clinical trials.


View this table:
[in this window]
[in a new window]
 
Table 3. Patient characteristics
 
Univariate analysis identified the number of sites of metastatic disease, performance status and HER-2 overexpression (HER-2-positive includes 3+ by immunohistochemistry or gene amplification by fluorescence in situ hybridization) as predictors of CNS metastasis (Table 4). HER-2 overexpression (P = 0.02) and number of metastatic sites (P = 0.03) remained predictive of CNS involvement by multivariate analysis; age (P = 0.052) and performance status (P = 0.054) were of borderline predictive value. Among asymptomatic patients undergoing screening CNS imaging, survival, whether from time of initial diagnosis, first metastasis or on-study CNS screening, was shorter for patients with compared to without occult CNS involvement (Figure 1 and Table 5). Lead- or length-time bias did not explain these differences as the initial disease-free interval (18.5 versus 22.2 months) and time from first distant metastasis to CNS screening (11.4 versus 12.1 months) was similar in patients with and without occult CNS involvement. Survival for patients with CNS involvement was similar, whether diagnosed clinically (symptomatic) or on screening images (occult).


View this table:
[in this window]
[in a new window]
 
Table 4. Predictors of occult central nervous system metastasis
 


View larger version (20K):
[in this window]
[in a new window]
 
Figure 1. Survival comparisons. (A, B) Impaired survival in patients with central nervous system (CNS) metastasis, whether occult or symptomatic. (C) Survival from time of CNS screening is shorter in patients with occult CNS metastasis (4.67 versus 10.43 months; P = 0.0013).

 

View this table:
[in this window]
[in a new window]
 
Table 5. Survival based on central nervous system (CNS) involvement
 
Twenty-one patients with occult CNS metastases received whole-brain radiation therapy; two patients with small solitary CNS lesions and rapidly progressive systemic disease received no CNS therapy. Twenty-one patients with occult CNS metastasis have died, all from apparent systemic disease progression. No patient with occult CNS disease developed neurological symptoms (focal deficits, seizures or need for chronic corticosteroids). Seven patients without apparent CNS involvement at screening subsequently developed symptomatic CNS disease requiring radiotherapy. Radiotherapy provided effective palliation of neurological symptoms in all six patients for whom details are available; all six succumbed to systemic disease. We were not able to determine how many patients presenting with symptomatic CNS metastasis died of systemic disease versus CNS progression.


    Discussion
 Top
 Abstract
 Introduction
 Patients and methods
 Results
 Discussion
 References
 
We found occult CNS metastasis in ~15% of patients with disseminated breast cancer screened for participation in antiangiogenic clinical trials. The patients we screened were young, heavily pretreated and highly selected, limiting the ability to generalize our results to the wider population of patients with metastatic disease. Nonetheless, the incidence in our series is consistent with previously reported, unselected autopsy series. Tsukada et al. found clinically unrecognized CNS involvement in 18% of 1044 breast cancer patients at autopsy [6]. Similar rates of occult CNS disease were reported by Amer [7] and Hagemeister et al. [8]. Given the prevalence of occult CNS involvement, clinical trials employing therapy known or suspected to exacerbate problems associated with CNS metastasis (such as an increased risk of bleeding or seizure) should consider mandatory screening.

Amer first suggested that systemic therapy may alter the natural history of breast cancer, leading to an increase in CNS metastasis [7]. Although adjuvant therapy increased overall survival, patients who relapsed after adjuvant therapy had a higher incidence of CNS metastasis in one series [9]. More recently, Carey et al. reported an excellent disease-free survival but disproportionate CNS relapse after aggressive neoadjuvant therapy [10]. An increase in CNS relapse after initial chemotherapy for metastatic disease has also been suggested. Thirty per cent of patients treated with epirubicin and docetaxel combination therapy subsequently developed symptomatic CNS metastasis [11]. A similar frequency of CNS metastasis was found after initial response to paclitaxel; 12% of patients had isolated CNS relapse with continued control of systemic disease [12].

Two explanations have been offered for the apparent increase in CNS metastasis after chemotherapy: first, the effectiveness of the blood–brain barrier and secondly, prolonged survival of patients after initial recurrence allows microscopic CNS metastasis to become clinically evident. Our data cannot fully address this question as most patients in our study were heavily pretreated before CNS imaging. Muss et al. used early CT technology to image 116 patients with disseminated breast cancer. CNS lesions were found in 11 of 37 (30%) patients with CNS symptoms but only 1 of 79 (2%) patients without CNS symptoms [13]. The increased prevalence of occult CNS metastasis in our patients likely represents both improvement in imaging technology and patient selection; the impact of differential exposure to systemic therapy is not clear. However, the consistency between the frequency of occult CNS metastasis in our patients and historical autopsy series suggests the impact of chemotherapy may be modest rather than large.

We were able to identify several factors associated with an increased risk of occult CNS metastasis. Performance status and number of extracranial sites of disease likely reflect bulk of systemic disease rather than tumor biology. The association of HER-2 overexpression and CNS metastasis deserves further investigation. Crivellari et al. retrospectively measured HER-2 expression in patients developing symptomatic CNS disease after initial chemotherapy [11]. Although overexpression (3+ by immunohistochemistry) was found in 10 of 16 (62%) patients, HER-2 was not assessed in patients without CNS metastasis, thereby limiting conclusions. An increase in CNS involvement has also been reported in patients receiving trastuzumab [14]. Prior autopsy studies found an inverse correlation between patient age and CNS metastasis [6, 15, 16]. Age was not predictive of occult CNS disease in our analysis, likely due to the young age of patients attracted to early-phase clinical trials. Although other authors have debated the relative sensitivity of contrasted CT and brain MRI for detection of CNS metastasis [17, 18], both appeared equally effective as screening modalities in our patients.

Initial disease-free interval was similar for patients with and without CNS involvement but patients with CNS metastasis had an impaired survival. Somewhat surprisingly, the survival of patients with occult CNS disease was identical to that of patients with symptomatic CNS lesions, whether measured from initial diagnosis, first recurrence or diagnosis of CNS metastasis. The median survival after diagnosis of occult CNS metastasis in our patients is similar to that reported for patients with symptomatic lesions in other series as well [1, 10]. These results imply a unique and aggressive biology of tumors trophic for the CNS with symptoms dependent on the anatomic location of the lesions. Given the disparity in survival, stratification for CNS involvement in phase III trials is warranted if patients with CNS metastasis are enrolled.

The impact of identifying and immediately treating occult CNS metastasis in our patients was questionable at best. End-of-life details are known for 21 of 23 patients with occult CNS lesions; all died of systemic disease progression without intervening CNS symptoms. Occult CNS metastasis appeared to be a marker for, rather than a cause of, limited survival. We cannot exclude the possibility that some patients were spared development of CNS symptoms as nearly all received whole-brain radiation therapy after diagnosis of occult CNS disease. Although some authors have raised the question of prophylactic CNS radiation in high-risk patients as a strategy to improve survival [11], we cannot recommend either routine CNS screening or prophylactic therapy in patients with uncontrolled systemic disease. However, our results cannot be generalized to patients with less advanced disease. Might patients with HER-2-positive breast cancer with prolonged disease control with trastuzumab benefit from CNS screening? This increasingly important clinical question remains unanswered and deserves further study.


    Acknowledgements
 
K. D. Miller was supported in part by a Clinical Research Training Grant for Junior Faculty CRTG-00-199-01-CCE from the American Cancer Society. G. W. Sledge Jr was supported in part by a grant from the Breast Cancer Research Foundation and the Walther Medical Foundation.


    Footnotes
 
+ Correspondence to: Dr K. D. Miller, 535 Barnhill Drive, RT-473, Indianapolis, IN 46202, USA. Tel: +1-317-274-0920; Fax: +1-317-274-3646; E-mail: kathmill{at}iupui.edu Back


    References
 Top
 Abstract
 Introduction
 Patients and methods
 Results
 Discussion
 References
 
1. DiStefano A, Yong Yap Y, Hortobagyi GN et al. The natural history of breast cancer patients with brain metastases. Cancer 1979; 44: 1913–1918.[ISI][Medline]

2. Zimm S, Wampler GL, Stablein D et al. Intracerebral metastases in solid-tumor patients: natural history and results of treatment. Cancer 1981; 48: 384–394.[ISI][Medline]

3. Hall WA, Djalilian HR, Nussbaum ES et al. Long-term survival with metastatic cancer to the brain [citation]. Med Oncol 2000; 17: 279–286.[ISI][Medline]

4. Sinard JH. Factors affecting autopsy rates, autopsy request rates, and autopsy findings at a large academic medical center. Exp Mol Pathol 2001; 70: 333–343.[CrossRef][ISI][Medline]

5. Gordon MS, Margolin K, Talpaz M et al. Phase I safety and pharmacokinetic study of recombinant human anti-vascular endothelial growth factor in patients with advanced cancer. J Clin Oncol 2001; 19: 843–850.[Abstract/Free Full Text]

6. Tsukada Y, Fouad A, Pickren JW et al. Central nervous system metastasis from breast carcinoma. Autopsy study. Cancer 1983; 52: 2349–2354.[ISI][Medline]

7. Amer M. Chemotherapy and pattern of metastases in breast cancer patients. J Surg Oncol 1982; 19: 101–105.[ISI][Medline]

8. Hagemeister F, Buzdar A, Luna M et al. Causes of death in breast cancer. Cancer 1980; 46: 162–167.[ISI][Medline]

9. Ahmann FR, Jones SE, Moon TE. The effect of prior adjuvant chemotherapy on survival in metastatic breast cancer. J Surg Oncol 1988; 37: 116–122.[ISI][Medline]

10. Carey L, Ewend M, Sawyer L et al. Disproportionate CNS relapse after aggressive neoadjuvant chemotherapy. Breast Cancer Res Treat 2001; 69: 299.

11. Crivellari D, Pagani O, Veronesi A et al. High incidence of central nervous system involvement in patients with metastatic or locally advanced breast cancer treated with epirubicin and docetaxel. Ann Oncol 2001; 12: 353–356.[Abstract]

12. Freilich RJ, Seidman AD, DeAngelis LM. Central nervous system progression of metastatic breast cancer in patients treated with paclitaxel. Cancer 1995; 76: 232–236.[ISI][Medline]

13. Muss HB, White DR, Cowan RJ. Brain scanning in patients with recurrent breast cancer. Cancer 1976; 38: 1574–1576.[ISI][Medline]

14. Lower E, Blau R, Bismayer J et al. Increased brain metastasis detected in metastatic breast cancer patients receiving Herceptin. Breast Cancer Res Treat 2001; 69: 271.

15. Aronson S, Garcia J, Aronson B. Metastatic neoplasms of the brain: their frequency in relation to age. Cancer 1964; 5: 558–563.

16. de la Monte S, Hutchins G, Moore G. Influence of age on the metastatic behavior of breast carcinoma. Hum Pathol 1988; 19: 529–534.[ISI][Medline]

17. Taphoorn MJ, Heimans JJ, Kaiser MC et al. Imaging of brain metastases. Comparison of computerized tomography (CT) and magnetic resonance imaging (MRI). Neuroradiology 1989; 31: 391–395.[CrossRef][ISI][Medline]

18. Davis PC, Hudgins PA, Peterman SB et al. Diagnosis of cerebral metastases: double-dose delayed CT vs contrast-enhanced MR imaging. Am J Neuroradiol 1991; 12: 293–300.[Abstract]

19. Lesse S, Netsky M. Metastasis of neoplasms to the central nervous system and meninges. Arch Neurol Psychiat 1954; 72: 133–153.

20. Abrams H, Spiro R, Goldstein N. Metastasis in carcinoma: analysis of 1000 autopsied cases. Cancer 1950; 3: 74–85.[ISI]

21. Viadana E, Cotter R, Pickren JW et al. An autopsy study of metastatic sites of breast cancer. Cancer Res 1973; 33: 179–181.[ISI][Medline]

22. Cifuentes N, Pickren JW. Metastases from carcinoma of mammary gland: an autopsy study. J Surg Oncol 1979; 11: 193–205.[ISI][Medline]

23. Cho SY, Choi HY. Causes of death and metastatic patterns in patients with mammary cancer. Ten-year autopsy study. Am J Clin Pathol 1980; 73: 232–234.[ISI][Medline]

24. Posner J, Chernik N. Intracranial metastases from systemic cancer. Adv Neurol 1978; 19: 579–591.

25. Lamvec J, Zidar A. Association of leptomeningeal carcinomatosis in carcinoma of the breast with infiltrating lobular carcinoma. Arch Pathol Lab Med 1991; 115: 507–510.[ISI][Medline]