A series of retrospective studies have reported that patients with human epidermal growth factor receptor 2(HER2)-positive breast cancer are at a greater risk of central nervous system (CNS) metastases. Trastuzumab, which does not cross the blood-brain barrier, has been associated with this increased risk.
The authors evaluated incidence, survival, and risk factors for CNS metastases in the incident breast cancer population systematically collected by the Parma Province Cancer Registry over the 4-year period between 2004 and 2007.
A total of 1458 patients with a diagnosis of stage I to III invasive breast cancer were analyzed for study purposes. At a median follow-up of 4.1 years, CNS events were observed in 1.3% and 5% of HER2-negative patients and HER2-positive patients, respectively (P < .0001). The administration of trastuzumab either as adjuvant therapy or for metastatic disease was associated with a significantly increased risk of CNS involvement at first disease recurrence and after first extracranial recurrence, respectively. According to multivariate analysis, HER2-positive status and trastuzumab treatment, high Ki-67 index, and hormone receptor negativity remained independent risk factors for the development of CNS metastasis.
The incidence of central nervous system (CNS) metastases varies by tumor site, with primary lung and breast carcinoma accounting for the bulk of cases.1 Clinically evident CNS metastases are diagnosed in 10% to 16% of women with metastatic breast cancer, whereas in autopsy studies, CNS involvement is found in 20% to 30% of patients.1, 2 Recent reports have raised the possibility of a growing incidence of CNS metastasis in patients with breast cancer.3, 4
On the basis of retrospective observations, risk factors for the development of CNS metastases from breast cancer include patient characteristics such as young age and African-American ethnicity, and biological features of the tumor, including estrogen receptor (ER) negativity, human epidermal growth factor receptor 2 (HER2) positivity, high tumor grade, and BRCA1 phenotype.1, 5–16 HER2 (Her-2/neu, c-erbB-2) is a 185-kilodalton transmembrane tyrosine kinase with extensive homology to the epidermal growth factor receptor. In humans, amplification of the HER2 oncogene occurs in 15% to 20% of primary breast cancer, and is associated with diminished disease-free and overall survival.17, 18 Several retrospective analyses have indicated that HER2-overexpressing patients are at a greater risk of developing CNS metastases.3, 13-16 However, other studies have obtained opposite results, demonstrating no association between HER2 status and CNS recurrence.4, 19, 20
The humanized anti-HER2 monoclonal antibody trastuzumab is an effective treatment of patients with HER2-positive breast cancer.21 However, soon after the introduction of trastuzumab in the late 1990s, clinicians began to observe an apparent increase in the incidence of CNS metastases over historical controls. Furthermore, a series of retrospective studies has documented a CNS recurrence incidence of 25% to 40% with trastuzumab-based therapy for patients with HER2-positive metastatic breast cancer.22-30
To our knowledge to date, there have been no large-scale observational studies performed analyzing factors associated with CNS recurrence in a general population of newly diagnosed breast cancer patients with known HER2 status. The objective of the current study was to assess incidence, survival, and risk factors for CNS metastases in the incident breast cancer population systematically collected by the Parma Province Cancer Registry (PCR) over the 4-year period between 2004 and 2007.
MATERIALS AND METHODS
Since 1978, the PCR has been the cancer surveillance system for the Parma district area. Information on newly diagnosed tumor malignancies among the resident population of Parma is systematically and prospectively collected in the PCR from routine medical chart review. Reports of all deaths are received from the municipal population registries.
After obtaining regional review board approval for the study, we used the PCR database to identify incident breast cancer cases diagnosed between January 1, 2004 and December 31, 2007. Only invasive breast cancers (International Classification of Diseases for Oncology-10 codes T-174) were accounted for. Patients who were male and patients who had metastatic disease at the time of diagnosis, >1 primary tumor, or bilateral disease were excluded from the analysis. Date of diagnosis, tumor immunohistochemistry (IHC), and vital status of the selected cases were collected into an electronic registration form. The clinical charts of these patients were then obtained and all other relevant information (personal and familial data and details regarding surgery, accurate staging, disease recurrence status, and treatments performed) was transferred and registered.
Because vital status was updated until April 30, 2010, this date was defined as the end of follow-up. All patients for whom no notification of death was available were assumed to be alive at the end of follow-up.
Clinical and Pathological Data
Date of birth, diagnosis date, pathological stage, lymph node involvement, histological type and grade, hormone receptor status, proliferation rate, and HER2 status of all study patients were reviewed to confirm the accuracy of variables recorded within the PCR database.
Hormone receptor status was labeled as “positive” when the fraction of cells stained by the corresponding immunoreaction was ≥10% and the proliferation rate was considered as “high” when ≥15% of cells were stained by the Ki-67 antibody. Patients were considered to have HER2-positive disease if the primary or metastatic lesion had strong overexpression (3+) on IHC or had gene amplification (ratio of HER2 to chromosome 17 copy number of >2) by fluorescence in situ hybridization (FISH). Patients were considered to have HER2-negative disease if they had either negative expression by IHC (0 or 1&plus) or did not have gene amplification by FISH. Tumor samples were centrally evaluated de novo for HER2 amplification when this marker was not available from pathology records or previous tests gave inconclusive results (eg, IHC expression of 2+).
The site of metastases was categorized as locoregional, soft tissue-lymph nodes, visceral, and CNS. Diagnosis of CNS recurrence was based on either computed tomography and/or magnetic resonance imaging scans of the brain. At the time of disease recurrence, CNS imaging was not performed unless there was a clinical suspicion of CNS recurrence. Because of the wide variability of the timing of trastuzumab treatment relative to the date of CNS metastases, 4 groups of patients with HER2-positive disease were evaluated: patients who had 1) never received trastuzumab; 2) had received trastuzumab as adjuvant treatment; 3) had not received trastuzumab as adjuvant treatment; and 4) received trastuzumab only after the diagnosis of any site of metastatic disease, but before the diagnosis of CNS metastases (Fig. 1).
Outcome Measures and Statistical Analysis
A retrospective analysis of prospectively collected cancer registry data was performed. The primary objectives of the current study were to evaluate the relation between HER2 status and risk of CNS metastases in patients with breast cancer, and to define the role of trastuzumab in increasing that risk. The cumulative number of CNS events occurring in study patients were tabulated by HER2 status and trastuzumab treatment, and compared between groups using the chi-square test31 and the 2-tailed Fisher exact test32 as appropriate. Similar comparisons were applied to the incidence of CNS metastases as the first site of recurrence.
CNS metastases-free survival was defined as the time from breast cancer diagnosis to the date of diagnosis of CNS metastases. CNS metastases-free survival after first extracranial event was defined as the time from the date of first extracranial metastases to the date of diagnosis of CNS metastases. Survival estimates were calculated using the Kaplan-Meier product limit method33 and compared between groups with the log-rank statistic.34 A stepwise Cox regression model was used to determine the association between HER2 status, trastuzumab treatment, and outcomes after adjustment for other relevant features.35 All potential prognostic variables were included in the model, and variables were removed from this model 1 at a time in a backward selection process using the likelihood ratio and a significance level of.05. Competing risks methodology was applied to analyze time to CNS recurrence. The 2 competing risks were CNS recurrence and disease recurrence at all other sites. For the competing risk analyses, Cox proportional hazard models were fitted using a data duplication method.36
The statistical data were obtained using STATA statistical software (version 9.1; StataCorp, College Station, Tex).
Patient and Tumor Characteristics
Between 2004 and 2007, 1458 women with invasive breast cancer were identified, met the eligibility criteria, and were included in the analyses. A total of 1210 (83%) patients had HER2-negative disease and 219 (15%) patients had HER2-positive disease, whereas the HER2 status was missing in 29 (2%) patients. Patients with HER2-positive breast cancer tended to have a higher initial disease stage and more aggressive phenotype than patients with HER2-negative disease (Table 1).
Table 1. Patient and Tumor Characteristics
HER2+ (n = 219) (%)
HER2– (n = 1210) (%)
HER2 indicates human epidermal growth factor receptor 2; +, positive; -, negative; NS, not significant.
Hormone receptor positivity: ≥10% staining cells.
Proliferation rate was considered as “high” when ≥15% cells were stained by the Ki-67 antibody.
Because of the wide variability of timing of trastuzumab treatment relative to the date of CNS metastases, 4 groups of patients with HER2-positive disease were evaluated (Fig. 1): patients who had never received trastuzumab (group 1); those who had received trastuzumab as adjuvant treatment (group 2); those who had not received trastuzumab as adjuvant treatment (group 3); and those who received trastuzumab only after diagnosis of metastatic disease, but before a diagnosis of CNS metastases (group 4).
The sites of metastatic events are shown in Table 2. At a median follow-up of 4.1 years (range, 2.4 years–6.4 years), 190 (13%) patients experienced a metastatic event. Overall, the incidence of CNS metastasis was 1.8% (26 of 1458 cases).
Table 2. Sites of Metastasis and CNS Disease Outcome
Prognostic Significance of HER2 and Trastuzumab Outcome
At the time of last follow-up, 178 (12%) patients had died. The adjuvant administration of trastuzumab appeared to overcome the adverse prognostic significance of HER2 positivity (Fig. 2). The overall survival of HER2-positive patients treated with adjuvant trastuzumab (group 2) was significantly better than that of HER2-positive patients who did not receive the drug (group 3) in the adjuvant setting (P = .006).
Cumulative Incidence of CNS Metastases by HER2 Status and Trastuzumab Treatment
HER2-positive status was found to be significantly associated with an increased risk of CNS metastases at any time (either as first or subsequent disease recurrence). Eleven (5%) of 219 HER2-positive patients developed CNS metastases in comparison with 15 (1.2%) of 1210 HER2-negative patients. This difference in incidence rates was statistically significant (P< .0001) and resulted in a significantly worse CNS metastases-free survival for HER2-positive patients (P< .00001) (Fig. 3A). When subdividing the entire study cohort between HER2-negative patients, HER2-positive patients never treated with trastuzumab (group 1), and HER2-positive patients who received trastuzumab at any time before the diagnosis of CNS metastases (groups 2 and 4), CNS events were observed in 1.3%, 1.6%, and 10.5% of patients, respectively. These incidence rates resulted in a significantly worse CNS metastases-free survival for patients treated with trastuzumab (P< .00001) (Fig. 3B).
Incidence of CNS Metastases After Trastuzumab Administration in the Adjuvant Setting
The administration of adjuvant trastuzumab was associated with an increased risk of CNS involvement at first recurrence. Seven (0.6%) of 1186 patients who underwent breast surgery for HER2-negative breast cancer developed CNS metastases as components of first recurrences compared with 2 (1.2%) of 161 HER2-positive patients who did not receive adjuvant trastuzumab (group 3) and 2 (4%) of 53 HER2-positive patients who received trastuzumab as adjuvant therapy (group 2). This difference in incidence rates was statistically significant (P = .035). It is interesting to note that trastuzumab administration was also associated with a higher incidence of isolated CNS recurrence. All CNS events observed at the time of first recurrence in HER2-positive patients who had received adjuvant trastuzumab treatment occurred with no evidence of extracranial recurrence by standard restaging procedures (Table 3). The median time to a CNS event for all patients with CNS metastases at the time of first recurrence was 18 months (range, 10 months-29 months). The median time to a CNS event for HER2-negative patients, HER2-positive patients who did not receive adjuvant trastuzumab, and HER2-positive patients who received trastuzumab as adjuvant therapy was 19.8 months, 10.3 months, and 20.3 months, respectively. The shorter time interval observed in HER2-positive patients who were not treated with adjuvant trastuzumab reached statistical significance (P = .018 by the log-rank test).
Table 3. CNS Metastases as Isolated Site of First Recurrence
Incidence of CNS Metastases After Trastuzumab Administration in the Metastatic Setting
Overall, 161 patients (11%) experienced extracranial metastases as the first site of tumor progression. Among these patients, there was a significant association noted between trastuzumab-based treatment and further metastatic spread to the CNS. Eight (7%) of 115 HER2-negative patients with metastatic disease developed CNS recurrences after the first extracranial event in comparison with none of the HER2-positive patients with metastatic disease who were never treated with trastuzumab and 7 (21%) of 33 HER2-positive patients who received trastuzumab-based therapy only after a diagnosis of metastatic disease (group 4). These incidence rates resulted in a significantly worse CNS metastases-free survival after the first extracranial event for patients treated with trastuzumab (P = .0004) (Fig. 4). The median time to a CNS event for all patients with CNS metastases after first extracranial disease recurrence was 10.8 months (range, 4 months-32 months). The median time to a CNS event after first extracranial recurrence for HER2-negative patients and HER2-positive patients treated with trastuzumab for metastatic disease (group 4) was 9.8 months and 11.3 months, respectively.
Multivariate Model for CNS Metastasis Risk
The following potential risk factors for CNS metastases at any time (either as a first or subsequent recurrence) were evaluated: age at diagnosis (<40 years vs ≥40 years); stage at diagnosis (I/II vs III); HER2 status and trastuzumab treatment (HER2-negative vs HER2-positive/not treated with trastuzumab vs HER2-positive/treated with trastuzumab); Ki-67 index (<15% [low] vs ≥15% [high]); tumor grade (1/2 vs 3); hormone receptor status (<5% staining cells [negative] vs ≥5% staining cells [positive]); and the occurrence of visceral metastases (visceral metastases before CNS event vs no occurrence). Factors found to be associated with a significant risk of CNS metastases on univariate analysis included stage III disease at diagnosis (hazard ratio [HR], 3.4; 95% confidence interval [95% CI], 1.5-7.2 [P = .01]), high Ki-67 index (HR, 5; 95% CI, 1.8-13.5 [P = .001]), hormone receptor negativity (HR, 8; 95% CI, 3.3-15.4 [P = .001]), and HER2-positive status and trastuzumab treatment (HR, 15; 95% CI, 6.6-20.5 [P<0.001]). Independent predictive factors for CNS event on multivariate analysis were HER2-positive status and trastuzumab treatment (HR, 4.3; 95% CI, 1.5-11.8 [P = .005]), high Ki-67 index (HR, 3; 95% CI, 1.1-8.9 [P = .042]), and hormone receptor negativity (HR, 5.4; 95% CI, 1.5-19.2 [P = .008]).
An additional multivariate model was performed estimating the CNS metastasis risk of first disease recurrence. Similar to the above-described findings, resulting independent predictive factors for CNS event were HER2-positive status and adjuvant trastuzumab treatment (HR, 3.1; 95% CI, 1.1-8.2 [P = .005]) and hormone receptor negativity (HR, 9.5; 95% CI, 2.6-18.2 [P< .001]).
To our knowledge, the current study is the first population-based cancer registry study analyzing factors associated with CNS recurrence in a general population of newly diagnosed breast cancer patients with known HER2 status. The data provide evidence that patients with HER2-positive breast cancer have a significantly higher incidence of CNS metastasis after treatment with trastuzumab.
Cancer registries may reveal the importance of specific trends in cancer incidence, prevalence, survival, and mortality; moreover, they can detect even small changes over time in the main biological indicators of cancer aggressiveness (such as morphology, immunophenotype, and staging). Strengths of the current study include the analysis of all incident stage I-III breast cancers diagnosed in the resident population of Parma, Italy between January 1, 2004 and December 31, 2007. They were systematically and prospectively collected in the PCR database, which should reduce selection bias and thereby prevent falsely elevated incidence rates.
In the current study, patients with HER2-positive breast cancer tended to have a higher initial disease stage and more aggressive phenotype than patients with HER2-negative disease. According to this observation, the possibility that the biological behavior of HER2-positive tumors may result, per se, in an increased rate of CNS metastasis is a rational and plausible hypothesis. A retrospective analysis of 9524 women with early stage breast cancer who were treated in the pre-trastuzumab era documented that the 10-year cumulative incidence of CNS metastasis as either a first or subsequent event was 6.8% in patients with HER2-positive primary tumors, compared with 3.5% in patients with HER2-negative tumors (P< .01).3 However, other studies obtained opposite results, and did not demonstrate an association between HER2 status and CNS recurrence.4, 19, 20, 37 Our analyses confirmed that HER2 status was significantly associated with an increased risk of CNS metastases, but that association was dependent on trastuzumab treatment. Therefore, the putative increase in CNS disease observed in women with HER2-positive breast cancer appears to be at least multifactorial, and could include inherent biological factors and treatment-related factors. However, it is important to note that the median follow-up of the current study was relatively short: only 4.1 years. Especially with regard to making comparisons between patients with HER2-negative tumors (most of whom were also ER positive) versus patients with HER2-positive tumors, many of the anticipated CNS events will not yet have occurred in the HER2-negative cases. The observed association between CNS metastasis risk and HER2-positive status might therefore be related to the time point at which the CNS event occurs (early vs late recurrence).
In the current study, according to data from the literature,38 the adjuvant administration of trastuzumab overcame the adverse prognostic significance of HER2 positivity. The overall survival of HER2-positive patients who were treated with adjuvant trastuzumab was significantly better than that of HER2-positive patients who did not receive the drug in the adjuvant setting. Each of the 3 major, phase 3, clinical trials of HER2-positive early breast cancer (the National Surgical Adjuvant Breast and Bowel Project [NSABP] B-31 trial, North Central Cancer Treatment Group [NCCTG] N9831 trial, and HERceptin Adjuvant [HERA] trial) indicated no statistically significant association between the risk of developing metastatic CNS disease and adjuvant trastuzumab therapy.38 However, a literature-based meta-analysis of these studies found that the incidence of CNS metastases was significantly increased in the trastuzumab-containing treatment arms compared with the non-trastuzumab–containing arms (absolute difference, 0.62%; P = .032).38 Trastuzumab was approved in Italy for the treatment of patients with HER2-positive early breast cancer in the second half of 2006.39 Using this date, the 4-year period between 2004 and 2007 was chosen for study analyses to obtain an adequate sample size of HER2-positive breast cancer patients not receiving adjuvant trastuzumab for comparison with trastuzumab-treated patients in the adjuvant setting. We observed that 0.6% of HER2-negative patients with early breast cancer developed CNS metastases as components of first recurrences compared with 1.2% and 4%, respectively, of HER2-positive patients with early stage disease who did not receive adjuvant trastuzumab and HER2-positive patients who received trastuzumab as adjuvant therapy (P = .035). Results from the current study further documented that, among patients with CNS metastases at the time of first recurrence, the median time to a CNS event was significantly shorter in patients with HER2-positive early disease who were not treated with adjuvant trastuzumab compared with those with HER2-negative disease. This finding can be explained by the observed aggressive nature of HER2-positive tumors. What is more, the results of the current study also indicated that the median time to a CNS event in HER2-positive patients with CNS metastases as the site of first recurrence significantly increased when prior adjuvant trastuzumab was administered. These results are in keeping with previously reported data in the metastatic setting. In series of 598 patients with invasive breast cancer, CNS metastases, and known HER2 status, patients with HER2-negative disease and patients with HER2-positive disease who did not receive trastuzumab were reported to have a shorter median time to CNS metastases compared with patients with HER2-positive disease who had received trastuzumab as first-line therapy for metastases.40 It is presumable that the level of control of the extracranial recurrence rate, which also was confirmed to be improved with adjuvant trastuzumab in the current study, is a contributing factor to the timing of the development of CNS metastases and may be an important factor for consideration when building predictive models.
CNS metastases in most instances develop late in the natural course of metastatic breast cancer, typically preceded by the involvement of the lungs, liver, and bone.37 More recent series have raised the possibility that with more systemic treatments prolonging survival, such as first-line trastuzumab-based therapy for HER2-positive metastatic breast cancer,41, 42 CNS events may be observed more frequently during breast cancer progression.5 Three retrospective studies inconclusively evaluated the risk and outcome of CNS metastases in HER2-positive patients with metastatic breast cancer who were receiving trastuzumab compared with those receiving non-trastuzumab–based treatment.28, 29, 43 In the current study, CNS metastases-free survival after the first extracranial event was found to be significantly worse for HER2-positive patients receiving first-line trastuzumab than for HER2-negative patients and HER2-positive patients not treated with the drug, with an observed CNS recurrence rate of 21% in patients treated with trastuzumab. The available evidence suggests that trastuzumab does not penetrate the blood-brain barrier well, even in the presence of brain metastases.44, 45 Consequently, the CNS is a potential sanctuary site in patients with HER2-positive disease treated with trastuzumab. This hypothesis is further strengthened by our data demonstrating that all CNS events observed at the time of first recurrence in HER2-positive patients who had received adjuvant trastuzumab treatment occurred with no evidence of extracranial disease recurrence. We can therefore define that the improvements in systemic control and overall survival associated with trastuzumab-based therapy lead to an “unmasking” of CNS metastases that would otherwise remain clinically silent before a patient's death.
It is interesting to note that our multivariate model indicated that, together with the combination of HER2-positive status and trastuzumab treatment, hormone receptor negativity was also an independent predictive factor for the occurrence of a CNS event. In accordance with this finding, recent reports have demonstrated that molecular subtypes of breast cancer other than HER2-positive disease (eg, triple-negative tumors) may be related to an increased risk of CNS metastases independently of treatment-related factors.46, 47
The current study is inherently limited by its observational nature. Because of this, it is supposed that the treatment with trastuzumab was not randomly allocated among HER2-positive patients but was determined depending on the patient status. The resulting patient prognosis may thus not be the result of the treatment decision but may be the cause of the decision. Therefore, some aspects of study methodology were considered to differentiate confounding relationships: 1) in addition to the cumulative incidence of CNS metastases at any time (either as a first or subsequent metastatic event), we calculated the occurrence of CNS metastases as the first site of disease recurrence with or without recurrence at any other site; and 2) in stepwise Cox regression analysis, CNS metastasis was considered the endpoint and other non-CNS events were treated as censored (competing risk model).
In conclusion, our population-based cancer registry study demonstrated that trastuzumab treatment-related factors confer an increased risk of CNS metastases in HER2-positive patients with breast cancer. Given the increasing efficacy of systemic therapy for extracranial disease, CNS metastases constitute a growing clinical problem and a therapeutic challenge. Screening studies are needed to investigate whether the early detection of CNS metastases in HER2-positive patients treated with trastuzumab improves outcomes, either by improving survival or by delaying or preventing the appearance of neurologic symptoms. At the same time, there is a greater need to conduct carefully designed trials of both cytotoxic chemotherapeutic agents and targeted agents,48, 49 either alone or in combination, with specific CNS endpoints.
We thank Dr. Vincenzo De Lisi, Dr. Paolo Sgargi, and Dr. Francesco Bozzani (Medical Oncology Unit and Cancer Registry of Parma Province) for their helpful comments and contributions regarding this work.
CONFLICT OF INTEREST DISCLOSURES
Supported by a grant from the Regione Emilia-Romagna.