Fax: (919) 681-1697
Overall survival of newly diagnosed glioblastoma patients receiving carmustine wafers followed by radiation and concurrent temozolomide plus rotational multiagent chemotherapy
Version of Record online: 9 JUN 2009
Copyright © 2009 American Cancer Society
Volume 115, Issue 15, pages 3501–3511, 1 August 2009
How to Cite
Affronti, M. L., Heery, C. R., Herndon, J. E., Rich, J. N., Reardon, D. A., Desjardins, A., Vredenburgh, J. J., Friedman, A. H., Bigner, D. D. and Friedman, H. S. (2009), Overall survival of newly diagnosed glioblastoma patients receiving carmustine wafers followed by radiation and concurrent temozolomide plus rotational multiagent chemotherapy. Cancer, 115: 3501–3511. doi: 10.1002/cncr.24398
- Issue online: 20 JUL 2009
- Version of Record online: 9 JUN 2009
- Manuscript Accepted: 7 JAN 2009
- Manuscript Revised: 15 DEC 2008
- Manuscript Received: 23 OCT 2008
- BCNU wafer;
Glioblastoma multiforme (GBM), the most lethal type of brain tumor, has a 1-year median survival. The effect of carmustine wafers on the survival of newly diagnosed GBM patients treated with radiotherapy (RT) and concurrent temozolomide (TMZ) plus RT plus rotational chemotherapy was investigated.
An institutional review board-approved retrospective study was conducted in 85 newly diagnosed GBM patients who received surgical resection with and without carmustine wafers followed by RT and concurrent TMZ plus rotational chemotherapy. Treatment group comparisons were conducted using the log-rank test. Survival experience of the Duke cohort was examined within specific patient subgroups defined by the original Radiation Therapy Oncology Group (RTOG) recursive partition analysis (RPA) class and compared with the European Organization for Research and Treatment of Cancer (Stupp) and RTOG trial.
Overall 1- and 2-year survival for the noncarmustine wafer cohort were 69% and 29%, respectively, with a median survival of 72.7 weeks. One- and 2-year survival for the carmustine wafer cohort were 81% and 47%, with median survival of 89.5 weeks. Carmustine wafer was not an independent predictor (P = .110) of survival after adjustment for RPA class. The proportion of patients in the carmustine wafer cohort who lived longer than predicted based upon Stupp regimen results was significantly greater than 0.5 (P<.006); similar results based upon the RTOG trial data were observed (P < .001).
Carmustine wafer with concurrent TMZ and radiation followed by rotational chemotherapy is a well tolerated, effective therapy, and has a survival benefit compared with radiation alone. Prospective randomized trials are needed to rigorously compare the carmustine wafer regimen to the Stupp and postradiation multimodality regimens. Cancer 2009. © 2009 American Cancer Society.
Newly diagnosed glioblastoma multiforme (GBM) is the most lethal type of brain tumor, with a median survival rate of approximately 12-15 months with standard treatment. Unfortunately, the median survival rate for GBM patients has not significantly changed, despite >25 years of research involving a variety of agents and delivery systems. Standard approach to therapy had consisted of surgical resection followed by radiotherapy. The role of adjuvant chemotherapy in primary glioma treatment has until recently been more controversial. A meta-analysis of 12 randomized trials comparing additional adjuvant chemotherapy to radiation alone demonstrated only a 5% increase (from 15% to 20%) in the 2-year survival rate.1 Thus, the general consensus in 2002 was that the use of chemotherapy after surgery and radiotherapy produced a marginal benefit over that seen with surgery and radiotherapy alone.
More recently, 2 large randomized clinical trials have demonstrated the independent efficacy of temozolomide and carmustine wafer in newly diagnosed malignant glioma patients, which led to Food and Drug Administration (FDA) approval of these agents for the treatment of malignant glioma.
Stupp et al reported the results of a multi-institutional phase 3 European Organization for Research and Treatment of Cancer trial in which patients with newly diagnosed GBM were randomized after surgery to receive either radiation alone or radiation with concurrent temozolomide followed by 6 cycles of adjuvant temozolomide. The 2-year survival was significantly higher (26.4%) for patients receiving radiation and adjuvant temozolomide than for patients receiving radiation therapy alone (10.4%).2 Westphal et al randomized patients with newly diagnosed malignant glioma to surgery with placement of carmustine wafers (polifeprosan 20 with carmustine implant) or a placebo wafer.3 All patients subsequently received radiotherapy. The results demonstrated a survival advantage favoring the use of carmustine wafers, and the FDA extended a prior indication for recurrent malignant glioma to include newly diagnosed malignant glioma. However, after surgical resection and radiation therapy, neither carmustine wafers alone nor single-agent temozolomide can effectively treat malignant gliomas.
Recently, the National Comprehensive Cancer Network 2008 Practice Guidelines in Oncology recommended the combination of carmustine wafers followed by radiation with concurrent adjuvant temozolomide for the treatment of primary malignant gliomas. However, there are no published randomized trials demonstrating its effectiveness or toxicity as a combination regimen. Unfortunately, increases in survival produced by either of these agents when used alone or with radiotherapy are modest, and the overall outcome for GBM patients remains dismal, with a medial survival of 12 weeks after recurrence. Although drug delivery may limit the intracerebral delivery of chemotherapies, the major cause of treatment failure is primary tumor resistance to chemotherapy. Resistance is predictable, given the marked heterogeneity of cancer cells within and between individual tumors, making it highly unlikely that single-agent chemotherapy alone will produce dramatic results in most settings. With only rare exceptions, such as imatinib for chronic myelogenous leukemia,4 monotherapy has shown little success in the treatment of human malignancy as compared with multiagent therapy. Thus, effective adjuvant treatment for malignant gliomas must include a multimodality approach, involving surgery and local treatment (with or without carmustine wafers) followed by radiation therapy and a combination of antineoplastic agents and possibly targeted molecular therapies.
To prevent tumor resistance from developing through multiple adjuvant chemotherapeutic agents with differing antitumor mechanisms, Duke treats primary glioblastoma patients with a standard regimen consisting of radiation therapy and concurrent temozolomide followed by adjuvant multiagent rotational chemotherapy (consisting of temozolomide, lomustine, and irinotecan) for a full year (Fig. 1). We have previously demonstrated that adjuvant rotational multiagent chemotherapy was an effective regimen when compared with patients receiving single-agent adjuvant regimens after radiation and concurrent temozolomide.5
We are interested in determining if there is a survival advantage for patients with newly diagnosed malignant glioma who are treated with the addition of carmustine wafer implantation followed by concurrent radiotherapy and temozolomide plus 1 year of multiagent adjuvant rotational chemotherapy. We hypothesize that the addition of carmustine wafers followed by radiation and concurrent temozolomide plus rotational multiagent chemotherapy (temozolomide, carmustine, irinotecan) would increase survival compared with: 1) patients who did not receive carmustine wafers, 2) patients who received the standard Stupp regimen, and 3) patients who received surgery followed by radiation alone (Radiation Therapy Oncology Group [RTOG] trial). Thus, we conducted a retrospective data analysis of 85 patients to determine the overall survival rate of newly diagnosed glioblastoma patients who were diagnosed between the years 2000 and 2005 by surgical resection (with or without carmustine wafer) and who also received radiation therapy and concurrent temozolomide followed by a full year of adjuvant multiagent rotational chemotherapy. Outcomes for both analyses included 1) overall survival and 2) toxicity.
MATERIALS AND METHODS
A retrospective chart review was performed on 176 patients' charts to determine which patients were eligible for the retrospective study covering the years 2000-2005 (Fig. 2). Retrospective analyses have many limitations, because of the lack of control over the sampling of the populations and the quality of the predictor and outcome variables. Existing data (especially tissue diagnoses, which were not all verified by an independent and secondary pathology review) may be inaccurate, incomplete, or measured in ways not ideal to answer the question. However, most prognostic variables chosen for this project are collected at a rate of 90% in the Duke Brain Tumor Center. Furthermore, the hard outcome of survival is easily obtained through our survival and social security databases. Follow-up may be difficult in some cases and may be a limitation. Selection biases may also be a concern. Most newly diagnosed patients between 2000 and 2004 were not receiving concurrent temozolomide and radiation as standard of care; thus, the patients who received this regimen may not be representative of the true population. Despite primary limitations of this historical data analysis, the recursive partition analysis (RPA) model has been incorporated into this study to address limitation of the potential biases (overt and hidden) and differences between the Duke and Stupp cohorts in their case mix. As described in the statistical section, the RPA model has recently and frequently been used to address this issue in other brain tumor trials. In addition, statistical methods to adjust for the overt prognostic differences not addressed by the RPA (eg, complete resection) between cohorts analysis were performed.
Inclusion criteria were: 1) histologically confirmed diagnosis of primary glioblastoma multiforme, 2) chemotherapy naive before resection or radiation, 3) underwent a gross total or partial resection with or without carmustine wafer placement, and 4) scheduled to receive radiotherapy (for a planned total of 60 grays [Gy]) and concomitant temozolomide therapy followed by rotational chemotherapy (Fig. 1).
Exclusion criteria were: 1) patient underwent biopsy alone, and 2) received monoclonal therapy or 3) upfront chemotherapy. During the study period evaluation (2000-2005), temozolomide and concurrent temozolomide were not standard of care until the seminal Stupp et al article was published in March 2005. Thus, of the 176 patients who had newly diagnosed GBM, 97 received radiation therapy and concurrent temozolomide followed by rotational chemotherapy with temozolomide, carmustine, and irinotecan. Eight patients were enrolled from 2005, when the Stupp regimen became standard of care. Eligible patients for the carmustine wafer cohort included patients who underwent a near complete resection. Only patients who underwent resection could receive carmustine wafer, and thus patients who underwent biopsy alone were not eligible for the retrospective chart review. Of the 97 eligible patients, 12 were excluded from the analysis; 6 patients who received liquid brachytherapy and 6 patients for other reasons (received upfront chemotherapy, uncertain if they received radiation). A final cohort of 85 patients who had a newly diagnosed GBM diagnosed only by surgical resection was included in the analysis. At the time of primary surgery, 49 did not receive carmustine wafer implantation, compared with 36 who did receive carmustine wafers. After radiation and concurrent temozolomide, all patients received the Duke standard newly diagnosed GBM treatment as outlined below (Fig. 1). All patients for whom there was intent to treat with rotational chemotherapy had data collected as outlined below.
Retrospective Chart Review
The following items were abstracted retrospectively from patient medical charts: hospital number, date of birth, race, sex, diagnosis, date of diagnosis, resection (with or without nitrosourea wafers) versus biopsy, surgical pathology number, tumor, Karnofsky performance status (KPS), radiation type, radiation dose, radiation with/without concurrent temozolomide, type of chemotherapy, number of treatment cycles, and date of death or last follow-up. Charts were further reviewed for toxicity. Decadron was administered according to standard of care practice in the perioperative setting and then tapered as tolerated postoperatively.
The objective of this retrospective cohort study was to compare the survival and toxicity experience of 2 patient cohorts treated at Duke: patients who received adjuvant rotational chemotherapy after surgical implantation of carmustine wafers and patients who received similar rotational therapy after surgery without implantation. The product limit estimator was used to describe the survival experience of both Duke cohorts (with vs without carmustine wafer). Estimates of 1-, 2-, and 3-year survival and median survival were generated within each cohort. Within each cohort, patients were classified into subgroups with homogeneous prognosis using the RTOG RPA class.6, 7 Performance status, age, mental status, neurological function, radiation dose, and extent of surgery define these RPA classes. Descriptive Kaplan-Meier estimates were generated within patient subgroups defined by the usage of carmustine wafers and RPA class. A log-rank test was used to assess the individual effect of carmustine usage, RPA class, race, and sex on survival. Given the retrospective nature of this study, a comparison of cohort survival requires adjustment for presurgical differences in cohort patient characteristics. The Cox proportional hazards model was used to compare the survival of patients treated with and without carmustine wafer adjusted for RPA class. The small number of patient events necessitates the need to minimize the number of predictors in the Cox model to retain validity. For that reason, RPA class was included in this model to adjust for performance status, age, extent of surgery, and mental status. An interaction between RPA class and wafer usage was also included in the Cox model.
An attempt was made to compare the survival experience observed within the Duke carmustine cohort to that reported by Stupp and the RTOG. By using estimates of median survival provided by the Stupp and RTOG trials within RPA class (3-4), the proportion of patients living longer than expected within the Duke cohorts was computed. A single-sample binomial test was used to determine whether that proportion was >0.5. The Cochran Armitage test for trend was use to examine the relationship between RPA class (3-4) and this proportion. Comparison with the Stupp regimen is difficult because of unbalanced baseline characteristics, a limitation of this retrospective analysis. Specifically, there was a higher fraction of patients achieving a complete resection in the carmustine wafer cohort (69%) compared with the Stupp cohort (39%). In an attempt to address this limitation, we conducted an analysis to determine whether complete resection had an impact on the proportion of patients who lived longer than expected based on the results of the Stupp cohort given the disparity in the percentage of complete resection between the carmustine wafer cohort and the Stupp cohort. A single-sample binomial test was conducted to determine whether this proportion was different from 0.5.
The maximum grade of each type of toxicity experienced by patients was summarized within each treatment group using the Common Toxicity Criteria version 3.0.
Patient characteristics of the 85 GBM patients who had undergone only surgical resection with or without carmustine-impregnated wafer implantation and subsequent radiotherapy with concomitant daily temozolomide (200 mg/m2 × 5 days vs 75 mg/m2/d for 42 days) plus multiagent rotational chemotherapy are presented in Table 1. The number of carmustine wafers implanted is determined by the size of the cavity and ranged from 1 to 8 wafers in the carmustine wafer cohort. The radiotherapy consisted of conventional doses of 180-200 cGy daily for 6 weeks. No patients were prophylactically treated for Pneumocystis carinii. Within the noncarmustine wafer cohort, 39% of patients were younger than 50 years, and 89% had KPS ≥70%. Within the carmustine cohort, 31% of patients were younger than 50, and 94% had KPS ≥70.
|Characteristics||No BCNU Wafer, n=49, No. (%)||BCNU Wafer, n=36, No. (%)||Stupp, n=287, No. (%)|
|<50||19 (39)||11 (31)||90 (31)|
|>50||30 (61)||25 (69)||197 (69)|
|Mean [SD]||52.7 [13.0]||53.8 [9.1]||NR|
|Men||34 (69)||21 (58)||185 (64)|
|Women||15 (31)||15 (42)||102 (36)|
|Caucasian||38 (78)||34 (94)||NR|
|African American||1 (2)||0 (0)||NR|
|Other||8 (16)||2 (6)||NR|
|Missing||2 (4)||0 (0)||NR|
|WHO performance status|
|0||28 (57)||17 (47)||113 (39)|
|1||14 (29)||17 (47)||136 (47)|
|2||5 (10)||2 (6)||38 (13)|
|3||2 (4)||0 (0)||0 (0)|
|Biopsy||0 (0)||0 (0)||48 (17)|
|Debulking||49 (100)||36 (100)||239 (83)|
|Complete resection||32 (65)||25 (69)||113 (39)|
|Partial resection||11 (23)||6 (17)||126 (44)|
|Missing (no biopsy)||6 (12)||5 (14)||0 (0)|
|BCNU wafer||0 (0)||36 (100)||0 (0)|
|Glioblastoma||49 (100)||36 (100)||221 (92)|
|Anaplastic astrocytoma||0 (0)||0 (0)||7 (3)|
|Inconclusive||0 (0)||0 (0)||3 (1)|
|Other||0 (0)||0 (0)||8 (3)|
Overall survival of the 2 cohorts is described in Table 2 and Figure 3. Although not statistically different, the 1- and 2-year survival in the carmustine wafer cohort was greater than that observed in the noncarmustine cohort (carmustine: 81% and 49%; noncarmustine: 69% and 29%). Similarly, the median survival in the carmustine group (89.5 weeks) was greater than that observed in the noncarmustine group (72.7 weeks). There was no difference in the point estimates in 3-year survival for the noncarmustine wafer and the carmustine wafer cohorts, which were 20% and 21%, respectively.
|No.||No. Dead||1-Year Survival Estimates (95% CI)||2-Year Survival Estimate (95% CI)||3-Year Survival Estimate (95% CI)||Median Survival (95% CI)|
|Overall||85||67||74% (65-84)||37% (28-48)||21% (14-32)||75.7 (66.4-94.7)|
|No BCNU Wafer||49||40||69% (58-84)||29% (18-45)||20% (12-40)||72.7 (62.7-84.3)|
|BCNU Wafer||36||29||81% (69-95)||47% (33-67)||21% (11-40)||89.4 (65.9-136.4)|
Descriptive statistics for the survival within patient subgroups defined by carmustine usage and RPA class are provided in Table 3, along with comparable statistics reported within the Stupp and RTOG studies. Figure 4 displays the survival experience of patient subgroups.
|RPA Class||Duke Rotational Chemotherapy||Stupp Phase 3||RTOG|
|No BCNU Wafers||BCNU Wafers|
|Median, wk||2-Year%||3-Year %||Median, wk||2-Year %||3-Year %||Median, wk||2-Year %||Median, wk||2-Year %|
The log-rank test was used to assess the relationship between survival and each of the following predictors: carmustine wafer usage, RPA class, race, and sex (Table 4). These analyses demonstrated RPA to be a significant predictor of survival (P < .001). The joint effect of RPA class, an indicator of carmustine wafer use, and its interaction on survival was examined within the context of the Cox proportional hazards model. The likelihood ratio test for interaction was not statistically significant (P = .086). A main effect model with factors describing RPA class and carmustine usage was fit to the data. Only RPA effect was highly significant. As described in Table 4, the likelihood ratio test for carmustine impregnated after adjustment for RPA class was not statistically significant (P = .110). As stated previously, we were also concerned about the effect of complete resection on the survival outcome. Although RPA analysis classifies patients based on extent of surgical procedure, it only differentiates between biopsy and resection, not degree of resection (complete vs partial resection). Because there was such a disparity in the percentage of complete resection between the carmustine wafer cohort (69%) and the Stupp cohort (39%), we deemed it necessary to determine the impact of this specific prognostic factor (complete resection) on survival (Table 4). The median survival within the Duke complete resection subgroup was 76.3 weeks, as compared with the reported 81.5 weeks in the Stupp Cohort. Among patients with complete resection, the proportion of patients who lived significantly longer than expected is not significantly different from 0.5 (P = .346). Thus, there was no difference in median survival between the Stupp and Duke cohorts in patients who had a complete resection.
|Hazard Ratio (95% CI)||Univariate P|
|BCNU wafer||0.75 (0.46-1.73)||.252|
|4 vs 3||1.30 (0.91)||—|
|5 vs 3||1.09 (0.92)||—|
|6 vs 3||1.99 (1.55)||—|
|White race||0.64 (0.32-1.31)||.248|
The median survival reported for the Stupp and RTOG trials within RPA classes was used to assess the survival experience of patients treated with carmustine and rotational chemotherapy (Table 5). In both cases, >70% of patients treated within the Duke carmustine cohort lived longer than expected according to statistics provided from the Stupp and RTOG trials (P = .001 and P = .006).
|Groups||No.||% >Expected||95% CI||P|
|Duke vs Stupp||34||74||56-87||.006|
|Duke vs RTOG||34||79||62-91||.001|
In comparing the Duke carmustine wafer cohort experience to RTOG and Stupp, we limited the analysis to RPA Class 3-4. By using estimates of median survival provided by the Stupp and RTOG trials within each RPA class, the proportion of patients living longer than expected within the Duke cohort was assessed, tested, and compared with the Stupp and RTOG trial. If the Duke carmustine wafer cohort regimen had a comparable outcome to Stupp or RTOG, then on average half should live longer than the reported median, and half should live a shorter period of time. The Cochran-Armitage Trend Test demonstrated that the proportion of patients within the Duke carmustine wafer cohort (RPA Class 4-5) lived longer compared with the Stupp and RTOG (radiation alone) regimen (P value in both situations is .018).
Table 6 provides a summary of the toxicity experienced by patients receiving radiation therapy and concurrent temozolomide plus rotational multiagent chemotherapy with and without carmustine wafer implantation. Both treatment regimens are well tolerated in comparison to the recently published “gold” standard adjuvant regimen. Comparison of the different regimen toxicities is difficult due to case mix; however, those toxicities reported for carmustine wafers or Stupp regimen reveal a similar toxicity profile. La Rocca did report more toxicity related to thrombosis, fatigue, and toxicity; however, the data are preliminary and ongoing.8 The incidence of the toxicities demonstrated in the La Rocca trial might decrease as the sample size and study power increase. Adverse events in GBM patients who did not receive carmustine-impregnated wafer implantation but received radiation with concurrent temozolomide followed by multiagent chemotherapy included (16%) grade 3 or 4 hematological toxicity consisting of 8% (4 of 49) grade 3-4 thrombocytopenia and 8% (4 of 49) grade 3-4 leukopenia. One (2%) patient demonstrated grade 4 pulmonary emboli. Four (8%) patients experienced grade 3-4 thromboses. Thirteen (27%) patients experienced grade 2-4 fatigue. Four (8%) patients experienced grade 2-4 infections. Four (8%) patients demonstrated grade 3-4 seizures during the entire study period. Four (8%) patients experienced grade 2 hemorrhages. No patient demonstrated symptomatic cerebral edema.
|Toxicity||PRTBTC, No BCNU Wafer, Temozolomide During Radiotherapy With Rotation Chemotherapy, n=49, No. (%)||PRTBTC, BCNU Wafer, Temozolomide During Radiotherapy With Rotation Chemotherapy, n=36, No. (%)||BCNU Wafer, Temozolomide During Radiotherapy With Adjuvant Temozolomide, * n=35, No. (%)||BCNU-impregnated Wafers,† No Adjuvant, n=120, No. (%)||Daily Temozolomide During Radiotherapy With Adjuvant Temozolomide, ‡ n=287, No. (%)|
|Thrombocytopenia, grade 3-4||4 (8)||6 (17)||NR||NR||33 (12)|
|Leukopenia, grade 3-4||4 (8)||5 (14)||NR||NR||20 (7)|
|Anemia, grade 2-4||0 (0)||3 (8)||NR||NR||4 (1)|
|Infection, grade 2-4||4 (8)||4 (11)||2 (5)||5 (4.2)||27 (9)|
|Headache, grade 3-4||2 (4)||2 (5)||NR||NR||NR|
|Pseudomeningocele||0 (0)||1 (3)||NR||NR||NR|
|Hemorrhage, grade 2||4 (8)||0 (0)||NR||NR||NR|
|Seizures, grade 3-4||4 (8)||4 (11)||NR||40 (33.3)||17 (5.9)|
|Fatigue, grade 2-4||13 (27)||5 (14)||11 (31)||NR||146 (50.8)|
|Pulmonary embolus, grade 4||1 (2)||1 (3)||4 (11)||NR||NR|
|Thrombosis, grade 3-4||4 (8)||3 (8)||12 (34)||NR||56 (19.5)|
|Gastrointestinal, grade 2-4||17 (35)||8 (22)||28 (80)||NR||85 (29.6)|
Toxicities in GBM patients who received carmustine wafers and subsequent radiation with concurrent temozolomide followed by multiagent chemotherapy had higher (31%) grade 3 or 4 hematological toxicity, consisting of 17% (6 of 36) grade 3-4 thrombocytopenia and 14% (5 of 36) grade 3-4 leukopenia. Three (8%) patients in this cohort experienced grade 2-4 anemia. One (3%) patient demonstrated grade 4 pulmonary emboli. Three (8%) patients experienced grade 3-4 thromboses. Five (14%) patients experienced grade 2-4 fatigue. Four (11%) patients experienced grade 2-4 infections. Four (11%) patients demonstrated grade 3-4 seizures during the entire study period. No patient experienced a grade 2 hemorrhage. One (3%) patient experienced a pseudomeningocele.
The evaluation of the role of chemotherapy in the treatment of malignant glioma has not produced impressive results to date. The FDA has approved only 3 agents for treatment of newly diagnosed malignant glioma over the past 30 years: carmustine wafers,2 temozolomide, and most recently, bevacizumab.3 Unfortunately, the vast majority of patients treated with available agent die of progressive tumor. New approaches will have to recognize the fallacy that single-agent chemotherapy will produce a significant increase in survival except perhaps in subsets of patients with very specific molecular profiles. Examples of this may include low O6-alkylguanine-DNA alkyltransferase9-15 or epidermal growth factor receptor VIII with wild-type PTEN.16 Accordingly, new strategies for patients with newly diagnosed malignant glioma will need to address identification of new agents, and targeting novel tumor molecular pathways,17 as well as combinations of agents with proven activity. This latter approach is the hallmark of strategies used where chemotherapy is particularly successful, including childhood acute lymphocytic leukemia, Wilms tumor, Hodgkin disease, and germ cell tumors.
The combination of carmustine wafers and temozolomide is especially intriguing, because temozolomide has been demonstrated to deplete methylguanine methyltransferase (MGMT), the DNA repair enzyme responsible for carmustine resistance. Preclinical studies have suggested that the synergistic effect of the combination of both carmustine and temozolomide may be in part related to depletion of MGMT. However, a study combining intravenous carmustine and temozolomide for malignant solid tumors has been associated with significant hematological toxicity, with a maximum tolerated dose of 110 mg/m2/d. Raizer et al concluded that the treatment of 80 malignant gliomas with intravenous carmustine and low-dose temozolomide (80 mg/m2/d) was not feasible for prolonged therapy because of myelosupression.18 Gururangan et al demonstrated in a phase 1 study that dual therapy with carmustine wafers and temozolomide (dose range: 100, 150, 200 mg/m2 × 5 days) was well tolerated, with no new or unexpected toxicities.19 Although this study demonstrated that temozolomide could be safely administered with carmustine wafers, and 2 patients in this trial completed 12 months of temozolomide treatment without progressive disease, it was a very small, 10-patient, noncomparative study and did not evaluate overall survival as a study endpoint. Preliminary data from a phase 2 multicentered trial consisting of 34 primary malignant glioma patients who received carmustine wafer implantation followed by radiation and concurrent temozolomide (75 mg/m2/d, 7 days/wk) followed by 6 monthly standard temozolomide (200 mg/m2/d × 5 days every 28 days ≤18 cycles) had a 1-year overall survival and progression-free survival of 64% and 29%, respectively. The most frequently reported adverse advents were nausea, constipation, and fatigue. Severe grade 3 and 4 toxicities include venous thrombosis, deep vein thrombosis, and pulmonary embolism. One patient experienced Staphylococcus aureus bacteremia, and 1 patient experienced a brain abscess.8
We have subsequently conducted a retrospective analysis to examine the efficacy of treatment with surgery, with or without carmustine wafer placement, followed by radiotherapy and concomitant temozolomide and rotational chemotherapy in patients with newly diagnosed malignant glioma. Although not statistically significant, these data suggest that the inclusion of carmustine wafers in a treatment regimen involving rotational chemotherapy may be of benefit. The small sample size and hence low power may have contributed to the lack of statistical significance. Given the limitations of this retrospective analysis, we have also shown that carmustine wafers followed by this multimodal regimen are tolerable, as there was no increase in toxicity in excess of that reported with either carmustine wafers or temozolomide alone.
Furthermore, carmustine wafers followed by this multimodal regimen improve the number of patients who survive longer than radiation alone, and are an effective and tolerable regimen. Although this single-center retrospective analysis has multiple limitations, we have made an effort to address the limitations by adjusting for RPA class. These results support the feasibility, tolerability, and activity of carmustine wafers followed by radiation and concurrent temozolomide plus rotational multiagent chemotherapy in patients with newly diagnosed malignant glioma. Future multicenter, prospective, phase 3 randomized trials should be conducted to confirm the efficacy and effectiveness of combining temozolomide with carmustine wafers.
Conflict of Interest Disclosures
The authors made no disclosures.
- 5Radiation and concurrent temozolomide followed by rotational multi-agent chemotherapy for glioblastoma (GBM) patients [abstract]. J Clin Oncol. 2007; 25( suppl): 92. Abstract 33458., , , et al.
- 8A phase II study of radiation with concomitant and then sequential temozolomide (temozolomide) in patients with newly diagnosed supratentorial high-grade malignant glioma who have undergone surgery with carmustine (BCNU) wafer insertion [abstract]. Neuro Oncol. 2006; 8: 445. Abstract TA28:445., , , et al.