Carmustine and methotrexate in combination after whole brain radiation therapy in breast cancer patients presenting with brain metastases: a retrospective study

Patients treated in our institution and affected by BC BM, diagnosed using either CT-scan or MRI of the brain, who received the combination of BCNU and MTX were considered in this retrospective study. Patients were identified from the Breast Cancer database of the Val d'Aurelle Medical Cancer Center, and records were reviewed for all patients treated with at least one cycle of BCNU-MTX for BC BM between 1999 and 2007. Identified patients were then followed until death or until February 2008. To be considered suitable for this treatment, patients had to have an Eastern Cooperative Oncology Group (ECOG) performance status (PS) [15] between 0 and 3. Patients were also required to have adequate bone marrow reserve (neutrophils ≥ 1.5.10⁹/L, platelets > 100.10⁹/L). This study was reviewed and approved by our Institutional Review Board.

Patients were treated by the combination of BCNU and MTX according to the proposal of our in-site Breast Cancer Multidisciplinary Committee. Treatment was continued until progression, unacceptable toxicity, inter-current disease or patient's refusal to continue. The 4-week-based chemotherapy regimen consisted of a combination of BCNU 100 mg/m² and MTX 600 mg/m² on day 1 and MTX 600 mg/m² on day 15. A cycle restarted at day 28 pending hematological recovery (absolute neutrophils count ≥1.5.10⁹/L and platelets > 100.10⁹/L) and return to grade 0 or 1 for non hematological toxicity. If one or more of these conditions was not met, then a 1- to 2-week delay was allowed for recovery. A 50% dose reduction was applied to BCNU in case of severe toxicity (grade 4 neutropenia ≥ 7 days during the previous cycle; grade 3 or 4 febrile neutropenia; grade 3 or 4 infection with neutropenia; grade 4 anemia or thrombocytopenia or bleeding requiring transfusion, or any grade ≥ 3 non-hematological toxicity). The drug was discontinued in case of a second occurrence of severe toxicity. Methotrexate was administered on day 15 depending on hematological recovery and return to grade 0 or 1 for non-hematological toxicity. Urinary alkalization using sodium bicarbonate (in order to obtain a urinary pH > 7.5) was based on a 4 h 500 ml infusion of 4.2% sodium bicarbonate before MTX infusion, followed by a 12 h 1.5 l infusion of 1.4% sodium bicarbonate solution. Folinic acid supplementation (25 mg every 6 hours day 2 to 5 and day 16 to 19) was systematically used to minimize methotrexate hematological toxicity. Patients presenting with HER-2 over expressing tumors received in addition (when commercially available) trastuzumab injections at the dose of 4 mg/kg on days 1 and 15 of the same cycle. Granulocyte colony-stimulating factors were used as secondary neutropenia prophylaxis in case of delays in chemotherapy due to long lasting neutropenia. Concerning patients receiving endocrine treatment, the treatment was discontinued at the initiation of the chemotherapy combination.

Pretreatment evaluation included a complete medical history and clinical examination with tumor measurements (imaging studies and physical examination when appropriate), appropriate radiological tests, concomitant treatments, PS, and hematological and biochemical profiles. Tumor measurements were performed every other cycle during the treatment course and every 3 months subsequently until progression. During the treatment duration, complete blood counts including a platelet and leukocyte differential count were performed weekly.

CT-scans of the brain, chest, abdomen and pelvis were performed before the initiation of chemotherapy, then every 8 weeks (2 cycles) until disease progression or chemotherapy stop. For this retrospective analysis, patients were evaluated for response and progression according to the RECIST criteria [16]. A minimal duration of 4 weeks was required to document a response, and the best response was recorded for each patient. A patient with cerebral and extra-cerebral disease was considered as a responder if there was a significant (as per RECIST criterion) decrease of both cerebral and extra-cerebral lesions. Progressive disease was defined as cerebral and/or extra-cerebral progression of the disease. We considered clinical benefit as a disease stabilization lasting at least 6 months or an objective response. Progression-free survival was defined as the time from the date of BCNU-MTX treatment initiation until the date of progression or death. Survival was defined as the time from the date of treatment initiation to the date of death. Toxicity, graded according to the National Cancer Institute Common Toxicity Criteria (NCI-CTC, version 3.0), was assessed by clinical examination and biological tests before each cycle of chemotherapy.

Estimates of median progression-free survival, and overall survival with their 95% confidence intervals (CI) were calculated using the Kaplan-Meier product-limit method [17]. For toxicity analyses, the worst grade for each cycle of chemotherapy was used. Treatment compliance was evaluated for each drug by the Relative Dose Intensity (RDI), defined as the ratio of actual dose intensity for each drug (total administered dose per unit of time, expressed in mg/m²/week) to the planned dose per unit of time [18]. The relative dose intensity of BCNU and methotrexate by patient was calculated as the average RDI for all drugs. In addition, we performed an explanatory analysis to identify the prognostic factors affecting this population. Patients without clinical events at the date of February 29th, 2008 were censored at this date, excepted in cases of patients previously lost to follow-up. The statistical analyses were performed using STATA v9.0.

From 1999 to 2007, 50 patients were treated with this combination. Patients' characteristics at the time of BM diagnosis are summarized in Table 1. Median age at the time of BM diagnosis was 54 years. Median time between diagnosis of BC and BM was 37 months (0-180). Only 1 patient presented with synchronous BM. Roughly two-thirds of the patient population was ECOG PS 0 or 1. No tumor was recorded to be Scarf, Bloom and Richardson (SBR) grade I, 40% were hormonal receptor negative and 50% of the 36 tumors analyzed for HER-2 amplification were positive. Ninety four percent of the patients presented with concomitant extra-cerebral disease, mainly bone (76%), liver (46%), lung (42%) and lymph node (26%) metastases. The median number of previous chemotherapy regimens in the metastatic setting was 2 (0-5). Twenty four patients (92% of the HR positive population) received previous hormonal therapy, either in the adjuvant or metastatic setting. The BM-associated clinico-radiological characteristics are summarized in Table 2. Six patients were clinically asymptomatic at the time of BM diagnosis. Only 5 patients presented with seizures and were treated with various anticonvulsants, precluding the analysis of an eventual interaction between concomitant medications and the chemotherapy efficacy and toxicity. No patients received prophylactic anticonvulsants. Thirty nine patients (78%) were treated by WBRT before the initiation of the BCNU-MTX chemotherapy (median dose 30 Gy, range 18 to 40). Due to the presence of cerebral and extra-cerebral disease, in order to treat extra-cerebral disease, the chemotherapeutic treatment was started shortly after the completion of the WBRT. Thirty three patients received planned chemotherapy immediately (less than one month after radiation therapy completion), and six patients received the combination chemotherapy as a salvage treatment after WBRT failure. Three patients (6%) had surgery and 2 patients were treated by radiosurgery (4%). Six patients were treated by the chemotherapeutic combination without previous radiotherapeutic or surgical treatment. Median time from BM diagnosis to initiation of chemotherapy was 1.3 months (range 0.1 to 8 months). Only 4 patients initiated chemotherapy more than 4 months (8 months) after the BM diagnosis.

A total of 256 chemotherapy cycles were administered to 50 patients, with a median number of 3 (1-20) cycles and a 12.9 week median treatment duration (range 2-91). Thirty percent of the patients received up to 6 chemotherapy cycles, with 20% patients continuing the treatment after 6 cycles. Thirty four percent of the patients discontinued their treatment before the third cycle. The characterization of this population is explored on the prognostic factors section. Concerning the HER-2 over-expressed population, 8 patients received 48 cycles of the combination BCNU-MTX and trastuzumab.

Chemotherapy dose modifications and delays of BCNU and MTX are summarized in table 3. The median RDI for BCNU and MTX were 0.98 (0.31-1.1) and 0.96 (0.57-1.66) respectively. Two patients received a MTX dose of 1 g/m², thus explaining the upper MTX RDI limit of 1.66.

There were 11 objective responses (23% [95%CI 12-37]) among 48 evaluable patients, with an additional 10% patients achieving disease stabilization lasting at least 6 months (5 patients), for a total clinical benefit rate of 33%. Two patients died of disease progression on the first 2 treatment weeks and were considered as non responders. Two patients were completely resected before the chemotherapy initiation and thus are not evaluable for response. Median response duration was 6.5 months (0.8 - 19.5). Median disease stabilization duration was 11.4 months (8.6 - 22.2+). Considering the HER-2 positive population treated by the BCNU-MTX trastuzumab combination (8 patients), 3 patients achieved a partial response and 2 additional patients achieved SD lasting more than 6 months. At the time of the current statistical analysis, 44 patients had a disease progression and 43 patients died. Thirty nine patients died from disease progression. In addition, there were 2 presumed treatment-related deaths (one case of acute renal failure at D6 of the first chemotherapy cycle, one patient developed a lethal febrile neutropenia 30 days after the fourth chemotherapeutic cycle). One patient was lost to follow-up, one patient died of an unrelated cause (myocardial infarction). Seven patients are still alive, 2 of them without evidence of residual disease. Median PFS was 4.2 months (95%CI: 2.8-5.3), with a one year PFS rate of 20.5% (10.3-33.2)(Fig. 1). First progression site was cerebral in 44% of the cases, outside the brain in 36%, and both intra- and extra-cerebral in 19% of the cases. Median overall survival of the whole population is 6.9 months (4.2-10.7), with a one-year OS rate of 32% (20-46) (Fig. 2). These survival data were 7.7 (0.4-NR) and 14.1 (0.4-NR) months respectively for PFS and OS in the population of 8 patients treated with trastuzumab and 3.5 (2.5-4.7) and 5.9 (3.9-8.2) months respectively for PFS and OS in the 42 patients population without trastuzumab treatment.

Worst NCI-CTC toxicity grades experienced by patients during the chemotherapy cycles are given in Table 4. Myelosuppression was the most common toxicity. Neutropenia was observed in 56% of cycles; however grade 3-4 neutropenia was observed in only 16% of the cycles. One case of febrile neutropenia was recorded, leading one month later to the patient's death. However, the clinical presentation was unusual, with a long lasting neutropenia without complete recovery one month later. Grade 2-4 anemia was a rare event (16.5% of cycles). Severe (grade 3-4) thrombocytopenia was observed in only 4.8% of the cycles. No hemorrhagic episode was reported and no patient required platelets transfusion. Besides myelosuppression, adverse events (AEs) possibly or probably related to study drugs were usually mild in severity and manageable. The most common grade 2-4 extra-hematological AE was fatigue. Emesis was an uncommon event, with nearly 80% of the cycles with grade 0 emesis. Considering alopecia, the determination of this toxicity frequency and grade is difficult considering the high proportion of patients previously treated by whole brain radiation therapy. None of the patients treated by this chemotherapeutic combination without previous radiation therapy to the brain suffered from alopecia. There were no reports of chemotherapy-induced neurological side effects.

After treatment failure, 48% (24 patients) of the patients received additional chemotherapeutic or hormonal treatments (1 regimen for 11 patients, 2 regimens for 6 patients, 3 regimens for 3 patients, 4 regimens for 4 patients and 7 regimens for 1 patient) and one patient was lost to follow-up.

Considering the relatively modest number of patients analyzed in this study, the exploratory analyses of prognostic factors affecting this population must be considered with caution.

In univariate analysis, PS 0-1 versus 2-3, BS-BM (Basic Score for Brain Metastases) score 0-1 versus 2-3 and presence of liver metastases were the only factors significantly associated with prognosis. PS appears to be a strong prognostic factor, with a significantly higher 1-year survival rate (47.8% [30% - 60%] versus 5.5% [0% - 20%], p = 0.0002) and a longer median overall survival time (11.9 months [6.9 - 20.4] versus 2.9 months [0,7 - 4.7]) in the population of patients with PS 0 or 1 as compared to 2-3 (Fig. 3). Similarly, patients with a BS-BM score 2 or 3 had a significantly better prognosis than those with scores 0 or 1 (1-year survival 47% versus 14%, median overall survival 10.7 months [6.9 - 23.8] versus 3.9 months [0.75 - 5.49], respectively, p = 0.0007). Patients affected by liver metastases had a significantly worse prognosis than those without liver metastases (1-year survival 19.6% [6.3 - 38.2] versus 43% [24 - 60.7], median overall survival 3.9 months [1.1 - 8.2] versus 10.7 months [5.5 - 23.8], respectively, p = 0.0007). Neither hormone receptor status (negative versus one or both positive receptors) nor HER-2 status were found to be significantly associated with progression-free survival or overall survival (HER-2 status 0-1+ versus 2-3+: 1-year survival 28% versus 49%, median overall survival 6.4 months [2.9 - 31.6] versus 8.4 months [2.1 - 22.5], respectively, p = 0.77; 1-year progression-free survival 32% versus 15%, median progression-free survival 3.0 months [2.4 - 18.9] versus 5.1 months [2.2 - 8.1], respectively, p = 0.78). No statistical difference was observed for overall survival or progression-free survival between patients previously treated with 0-2 chemotherapeutic lines versus patients with 3+ previous chemotherapeutic lines. However, due to the small sample size of this study, significant and non significant results of the effects in particular subgroups needs to be interpretaed with extreme care. Considering this, a multivariate analysis was not performed.

At the same time, we tried to characterize the population of patients who were not able to complete the first 2 chemotherapy cycles due to toxicity, disease progression or performance status alteration. This population of 17 patients was affected with a really poor prognosis, with a median overall survival of one month [0.5 - 2.9]. Two-thirds of this population had a PS 2 or 3 and had BS-BM scores of 0 or 1. Also, liver metastases were present in 70% of the patients, a far more elevated percentage than in the whole study population.