|
Advertisement |
|
|
||
 |
|
|||||
|
|||||
|
 |
||||||
|
||||||
|
||||||
|
© 1999 American Society for Clinical Oncology
Procarbazine, Lomustine, and Vincristine (PCV) Chemotherapy for Anaplastic Astrocytoma: A Retrospective Review of Radiation Therapy Oncology Group Protocols Comparing Survival With Carmustine or PCV Adjuvant ChemotherapyFrom the Department of Neurosurgery, University of California–San Francisco, San Francisco, CA. Address reprint requests to Michael Prados, MD, University of California–San Francisco, Box 0372, San Francisco, CA 94143-0372; email pradosm{at}neuro.ucsf.edu ABSTRACT PURPOSE: To determine any differences in outcome for patients with anaplastic astrocytoma (AA) treated with adjuvant carmustine (BCNU) versus procarbazine, lomustine, and vincristine (PCV) chemotherapy. MATERIALS AND METHODS: The Radiation Therapy Oncology Group (RTOG) database was reviewed for patients with newly diagnosed AA treated according to protocols that included either BCNU or PCV adjuvant chemotherapy. All patients were treated with radiation therapy. The outcome analysis included overall survival, taking into account patient age, extent of resection, Karnofsky performance status (KPS), and treatment group (BCNU v PCV). Stratified and nonstratified Cox proportional hazards models were used, as well as an analysis using matched cases between the groups. RESULTS: A total of 257 patients were treated with BCNU according to RTOG protocols 70-18, 83-02, and 90-06; 175 patients were treated with PCV according to RTOG protocol 94-04. All pretreatment characteristics except KPS were well balanced by treatment group; 61% of the BCNU group had a KPS of 90 to 100 compared with 73% of the PCV group (P = .0075). No statistically significant difference in survival was observed in any age group or by KPS or extent of surgery. The stratified analysis also showed no trends for improved survival by treatment group (P = .40). The Cox model identified only age, KPS, and extent of surgery as important variables influencing survival, not treatment group. Matching cases between groups using age, KPS, and surgery resulted in 133 matched pairs. No difference in survival was observed (P = .41). In a Cox model in which each matched pair is a strata, there was no difference between groups (P = .20). CONCLUSION: Using this retrospective analysis, there does not seem to be any survival benefit to PCV chemotherapy. Future phase III studies for patients with AA may need to consider whether BCNU or PCV is used in the control arm. PROCARBAZINE, lomustine, and vincristine (PCV) is one of the most commonly used combination chemotherapy regimens in neuro-oncology. This is due, in part, to a report published in 1990 that analyzed data from a Northern California Oncology Group (NCOG) randomized trial that had closed in 1983.1 NCOG protocol 6G61 was a randomized trial that compared the effects of either carmustine (BCNU) alone or PCV combination chemotherapy after whole-brain radiation therapy. Oral hydroxyurea was used during radiation therapy in both arms of the study, which included patients with anaplastic astrocytomas (AA) and glioblastoma multiforme (GBM). The conclusion from this reanalysis was that PCV chemotherapy produced longer survival and time to progression than did BCNU, but only for patients with AA. Median survival duration was 157.1 weeks for AA patients treated with PCV compared with 82.1 weeks for AA patients treated with BCNU (P = .021). There was no survival benefit for patients with GBM (median survival duration, 50.4 weeks with PCV and 57.4 weeks with BCNU). After this study, the NCOG initiated a phase II trial using bromodeoxyuridine (BUdR) given during radiation therapy, followed by PCV chemotherapy. The results of the trial were promising, with 50% of AA patients alive at 4 years.2 A phase III trial then commenced, comparing radiation therapy with or without BUdR, with all patients receiving adjuvant PCV chemotherapy. This last study originated as an NCOG trial but became an intergroup trial coordinated by the Radiation Therapy Oncology Group (RTOG). The study, RTOG 94-04, was the first trial conducted within the RTOG that used PCV chemotherapy as adjuvant treatment to radiation and excluded patients with GBM. Previously, BCNU was the most commonly used adjuvant chemotherapy used by the RTOG, and patient enrollment could include patients with GBM or AA. RTOG protocol 94-04 was closed before full enrollment because of a statistically strong likelihood of a lack of benefit using BUdR. A subsequent analysis of NCOG and RTOG studies was then conducted to study the influence of BUdR radiosensitization for patients with AA and GBM.3 As part of that retrospective analysis, RTOG studies that used BCNU adjuvant chemotherapy were compared with NCOG studies that used PCV chemotherapy. Although the primary goal of the review was to study BUdR, no apparent survival differences were noted in the subset of AA patients included in the analysis, at least suggesting that PCV was possibly no more effective than BCNU. RTOG protocol 94-04 provided an opportunity to review specifically survival outcome for AA patients treated with either BCNU or PCV within the RTOG, with much larger patient numbers than NCOG protocol 6G61. That earlier study included only 36 and 37 patients in the PCV and BCNU arms, respectively. The central goal of this current retrospective analysis is to determine if there is any difference in survival outcome for AA patients treated with adjuvant BCNU or PCV. MATERIALS AND METHODS The RTOG database was used to identify newly diagnosed patients with AA treated according to protocols 79-18, 83-02, 90-06, and 94-04. The first three protocols included the use of BCNU chemotherapy as an adjuvant to radiation therapy. RTOG 79-18 was a phase III randomized trial of radiation therapy with or without misonidazole used as a radiation sensitizer during radiation treatment.4 Patients were treated with whole-brain single-fraction radiation therapy to 60 Gy and received BCNU on days 3, 4, and 5 of treatment, and then every 8 weeks. RTOG 83-02 was a phase I/II randomized dose-escalation trial of hyperfractionated (1.2 Gy two times per day) partial brain radiation and accelerated hyperfractionated (1.6 Gy two times per day) partial brain radiation.5 Four dose levels of hyperfractionation were used: 64.8 Gy, 72.0 Gy, 76.8 Gy, and 81.6 Gy. Two dose levels of accelerated hyperfractionation were used: 48.0 Gy and 54.4 Gy. All patients were treated with BCNU using the same dose and schedule as in RTOG 79-18. RTOG 90-06 was a phase III randomized trial of partial-brain hyperfractionated radiation therapy to 72 Gy versus standard fractionated radiation treatment to 60 Gy.6 All patients were treated with adjuvant BCNU chemotherapy in a fashion similar to the other protocols. BCNU was administered at a dose of 80 mg/m2 on days 3, 4, and 5 of week 1 of radiation, and then for 3 days every 8 weeks beginning on day 64. Alternatively, BCNU could start on days 1, 2, and 3. RTOG 94-04 was a phase III randomized trial of standard, fractionated partial-brain radiation with or without BUdR used as a radiation sensitizer. All patients in this study received adjuvant PCV chemotherapy, which was started after the completion of radiation therapy. RTOG protocol 79-18, 83-02, and 90-06 included patients with both GBM and AA. RTOG protocol 94-04 only included patients with AA. The database was searched for patient variables, including age, initial Karnofsky performance status (KPS), extent of surgery (biopsy v any other extent of resection), and pathology. Only patients classified as AA based on central pathology review were included in this analysis. Re-review of original pathology material was not performed for this study. The RTOG criteria for "astrocytoma with anaplastic features" required a tumor with multifocal or diffuse cellular and/or nuclear pleomorphism, increased cell density, increased mitotic features, and increased vascular prominence.7 Tumor necrosis was not allowed. These pathologic features were used to classify patients for RTOG protocols 79-18, 83-02, and 90-06. Patients enrolled onto RTOG protocol 94-04 were classified as AA if they had focally moderate or high cellularity and at least two of the following features: high nuclear/cytoplasmic ratio, coarse nuclear chromatin, increased mitotic activity, and nuclear and/or cytoplasmic pleomorphism.8 For the purposes of this review, all patients are called AA. If the original pathology report identified any component of oligodendroglioma in the specimen, that was also noted. Patients were considered assessable for this analysis if they had central review of pathology, conformed to eligibility criteria for the study they were on, and had available survival data. Survival was the primary outcome variable of interest.
Statistical Methods RESULTS
From the four protocols, a total of 257 assessable patients were treated with BCNU (50 from RTOG 79-18, 133 from RTOG 83-02, and 74 from RTOG 90-06), and 175 assessable patients were treated with PCV. The pretreatment characteristics were well balanced by treatment group (Table 1). Fourteen percent of the BCNU group was
The BCNU group had a longer follow-up duration than the PCV group. Sixty-five percent of the patients in the BCNU group underwent complete follow-up evaluation compared with 35% in the PCV group. The censoring pattern is uniform over 13 years in the BCNU group and over 7 years in the PCV group. The median survival duration has not been reached for patients treated with PCV who are less than 50 years of age. Tables 2 to 6 show survival by patient age, grouped by decade. The 1-year survival is equal in the 18- to 29-year-old patients (Table 2), but only two deaths have occurred in the PCV group. There is no statistically significant difference in survival (P = .19), but it may be too early to determine. Table 3 indicates equal survival between the groups for the 30- to 39-year-old patients (P = .93). Patients aged 40 to 49 (Table 4) had only slightly higher survival with PCV than with BCNU, but it was not statistically significant (P = .29). Patients older than 50 years of age (Tables 5 and 6) fared slightly better with BCNU, but this also was not statistically significant. There does not seem to be any age relationship between treatment group and outcome.
There was no difference in survival by KPS according to treatment group when patients were grouped by KPS 70, 80, 90, or 100 (data not shown). Patients who received PCV after a resection had higher 3-year survival, but this may be a result of lack of follow-up evaluation (Table 7). There was no difference in the biopsy-only group (Table 8).
A stratified analysis using age, KPS, and extent of surgery to define strata was performed. Table 9 defines the strata. As can be seen in Table 10, there is no trend for improved survival for one treatment group over the other (P = .40). A stratified Cox proportional hazards model provided equivalent results (Table 11; P = .35). The nonstratified Cox regression model indicates that age, KPS, and extent of resection are significant factors, but not treatment group (Table 12).
Finally, cases were matched between the two groups using age, KPS, and extent of surgery. There were 133 matched patients in both groups. Patients without a match were not included in this analysis. Table 13 presents the overall survival, indicating no difference in survival between the two groups (P = .20). Interestingly, the stratified analysis gave a risk ratio of 1.17 favoring of BCNU, but the matched analysis had a risk ratio of 0.74 favoring PCV (Table 14). This would indicate no difference and some variability. Figure 1 shows overall survival of all patients treated with either PCV or BCNU (P = .546).
Similar analyses were performed to compare the outcome of patients with a mixed high-grade oligoastrocytoma (data not shown). Of the 257 patients treated with BCNU, 29 had a reported oligodendroglial component. Of the 175 patients treated with PCV, 41 had a reported oligodendroglial component. In a stratified analysis comparing survival for patients with a mixed tumor, no difference was noted (P = .63). Similarly, a regression analysis was performed, again showing no difference in outcome (P = .59). A nonstratified Cox regression model showed no impact of treatment on survival for these patients. Finally, overall survival was similar in the PCV- and BCNU-treated patients with mixed tumors. Median survival for BCNU-treated patients was 8.5 years and was not yet reached for the PCV group (P = .26). In conclusion, there does not seem to be any survival benefit to the use of PCV. DISCUSSION The goal of this study was to determine whether there was any difference in survival for AA patients treated with either adjuvant BCNU or PCV chemotherapy. The rationale for the study was, in part, a result of concern that a definitive study has yet to be conducted to show superiority of PCV over BCNU in this subset of patients. The only comparative study conducted to address this question was the initial NCOG trial that completed enrollment in 1983. Patients with malignant glioma that included both GBM and AA were eligible for randomization to either PCV or BCNU adjuvant chemotherapy. The only stratification variable used was the initial KPS, and randomization took place before radiation therapy. Patients were enrolled onto the study between December 1977 and February 1983. The analysis of the trial in 1985 did not result in a statistically significant advantage for the use of PCV chemotherapy.9 A subsequent reanalysis of a subset of patients resulted in the conclusion that PCV was superior to BCNU, but only in the group of patients with AA other than GBM. This reanalysis only included patients with a KPS of 70 to 100 who received radiation therapy with hydroxyurea and had at least one course of chemotherapy with either BCNU or PCV. Patients with lower KPS or those who did not complete at least one full cycle of chemotherapy were not included. Thus, the reanalysis was not performed as an intent-to-treat analysis. In addition, survival in the subset of patients with AA who were treated with BCNU (control arm) was shorter than one would expect based on current standards. Furthermore, the numbers of patients in each group was small. A sample size of less than 40 patients per group could not be expected to confidently detect a survival advantage with high power. Despite these issues of sample size and a retrospective analysis of a subset of patients, the use of PCV chemotherapy has become commonplace in the adjuvant treatment of AA patients. The current review is also retrospective and is subject to the limitations of any such analysis. For instance, the type of irradiation and the pathologic classification of AA may have influenced the outcome. The treatment given to patients differed by protocol and included various forms of radiotherapy. Some protocols used single fractionated irradiation, whereas others used accelerated or hyperfractionated irradiation. Some protocols used whole-brain irradiation, whereas others used focal or partial-brain irradiation fields. The dose varied by protocol as well. One could argue that the most important treatment used in these protocols was radiation therapy, and any potential differences in survival outcome would be a result of radiation dose or fractionation scheme rather than the type of adjuvant chemotherapy. However, despite long years of research related to dose and fractionation, there does not seem to be a significant benefit to any scheme other than single-dose treatment to 60 Gy to a focal or partial field. It would seem unlikely that the various dose/fractionation schemes used in the protocols in this analysis would have selectively impacted survival either positively or negatively. If there were any impact, it would seem probable that the use of whole-brain irradiation would negatively affect survival. BCNU chemotherapy was used in the few protocols in which whole-brain irradiation was given, and all patients who were treated with PCV were given partial-brain irradiation. One could theoretically assume that the use of whole-brain irradiation with BCNU would negatively impact the survival outcome of that patient group. However, that fact did not seem to change the outcome of the current analysis. Finally, patients in the BCNU group had longer follow-up, but trends in the first 3 years provide no indication of future differences. The precise pathologic diagnosis of AA is subject to some debate among neuropathologists, and other than the requirement that central review was performed by protocol design to be included in this analysis, a re-review of all pathology was not performed. A small number of patients were noted to have mixed high-grade tumors. It is possible that some patients declared as AA in earlier protocols may be considered mixed tumors today. In cases in which the tumor was considered to have an oligodendroglial component, however, no difference in survival outcome could be observed between the two treatment groups. The neuropathologist who performed central review of patients treated with PCV used pathologic criteria that differed slightly from those treated with BCNU. Thus, it is conceivable that some pathologic discordance exists in this group of patients, which may account for some differences in outcome between the PCV group and the BCNU group. In a previous study of patients with both GBM and AA treated according to NCOG and RTOG protocols, the two neuropathologists who performed the central review compared cases to assess the potential degree of discordance.3 There was disagreement in approximately 10% of sampled cases. These same two neuropathologists reviewed cases for the current studies. Our hope is that with sufficient numbers of patients, significant differences in pathology would not be present to influence the results in one direction or the other. Other patient factors that influence outcome were well balanced between the groups, and the analysis was specifically performed to account for these factors. There has not been a prospective phase III study conducted in the United States that has randomized patients with AA to surgery plus irradiation versus surgery plus irradiation and adjuvant chemotherapy. Previous RTOG studies included patients with both GBM and AA and suggested a benefit of chemotherapy for younger patients with tumors other than GBM. Again, sample size and protocol design did not specifically address the question of adjuvant chemotherapy for AA patients. Recent preliminary data from a Medical Research Council phase III trial conducted in the United Kingdom, which randomized patients after surgery to irradiation alone versus irradiation plus adjuvant PCV, suggest that PCV chemotherapy offers no survival benefit for patients with either GBM or AA.10 The final analysis of that study is still awaiting further follow-up evaluation. Thus, one cannot assume that either BCNU or PCV is the standard of care or, indeed, if any chemotherapy is of proven benefit given immediately after irradiation for this patient group. If the final analysis of the Medical Research Council phase III study is statistically compelling enough to confidently show that adjuvant PCV has no benefit for patients with AA, then future studies may want to consider irradiation alone as the control arm. Again, however, if the AA survival data from that study is substantially lower than median survival expectations reported from other large studies, the "final" answer concerning adjuvant chemotherapy may still await further trials. Pertinent to this question is whether BCNU or PCV should be considered the standard form of chemotherapy to be used in the adjuvant setting, if one considers adjuvant chemotherapy to be important in improving survival. The current retrospective analysis suggests that either BCNU or PCV result in similar survival outcomes. Thus, either therapy could be considered a reasonable option after irradiation. The choice could be made based on the potential toxicity of either the single agent or the combination, or by patient or physician preference. Ideally, it would be desirable to make decisions about adjuvant chemotherapy based on well-designed, appropriately sized, prospective phase III studies. Study designs are now being considered for patients with AA. One could argue that a phase III trial of BCNU versus PCV should now be considered for such a study. However, it may be difficult to accrue patients onto such a study because of physician or patient bias that a trial of BCNU versus PCV would not likely yield a positive result (ie, superiority of one over the other), would consume valuable patient and clinical trials resources, and is not an "exciting" enough study question to evaluate in 1999. Alternatively, one could consider the use of either BCNU or PCV as the control arm of a new phase III study, randomizing that therapy to a promising new chemotherapy agent. Another option would be to consider the use of BCNU alone compared with BCNU plus another agent, assuming sufficient phase II experience was available for the potential combination. Alternatively, one may consider that there is not a good phase III question to study at this point for this patient group, and further phase II trials need to be completed to identify promising new strategies. Careful consideration needs to be given to these and other ideas before a large phase III trial is started that likely will not yield results for many years. Hopefully, as we learn from past studies and consider various options, results will be forthcoming that translate into a "true" standard of care. ACKNOWLEDGMENTS Supported by grants no. RTOG U01 CA21661, CCOP U10 CA37422, Stat U10 CA32115, and UCSF P01 CA13525 from the National Cancer Institute, Bethesda, MD. REFERENCES 1. Levin VA, Silva P, Hannigan J, et al: Superiority of postradiotherapy adjuvant chemotherapy with CCNU, procarbazine and vincristine (PCV) over BCNU for anaplastic glioma: NCOG 6G61 final report. Int J Radiat Oncol Biol Phys18:321-324, 1990[Medline] 2. Levin VA, Prados MD, Wara WM, et al: Radiation therapy and bromodeoxyuridine chemotherapy followed by procarbazine, lomustine, and vincristine for the treatment of anaplastic glioma. Int J Radiat Oncol Biol Phys32:75-83, 1995[Medline] 3. Prados MD, Scott CB, Rotman M, et al: Influence of bromodeoxyuridine radiosensitization on malignant glioma patient survival: A retrospective comparison of survival data from the Northern California Oncology Group (NCOG) and Radiation Therapy Oncology Group (RTOG) trials for glioblastoma multiforme and anaplastic astrocytoma. Int J Radiat Oncol Biol Phys40:653-659, 1998[Medline] 4. Nelson DF, Diener-West M, Weinstein AS, et al: A randomized comparison of misonidazole sensitized radiotherapy plus BCNU and radiotherapy plus BCNU for treatment of malignant glioma after surgery: Final report of an RTOG study. Int J Radiat Oncol Biol Phys12:1793-1800, 1986[Medline] 5. Weiner-Wasik M, Scott CB, Nelson DF, et al: Final report of a phase I/II trial of hyperfractionated and accelerated hyperfractionated radiation therapy with carmustine for adults with supratentorial malignant glioma: RTOG 83-02. Cancer77:1535-1543, 1996[Medline] 6. Scott CB, Curran WJ, Yung WKA, et al: Long term results of RTOG 9006: A randomized trial of hyperfractionated radiotherapy (RT) to 72.0 Gy and carmustine vs standard radiation and carmustine for malignant glioma patients with emphasis on anaplastic astrocytoma patients. Proc Am Soc Clin Oncol 17:401a, 1998 (abstr 1546) 7. Nelson DF, Nelson JS, Davis DR, et al: Survival and prognosis of patients with astrocytoma with atypical or anaplastic features. J Neurooncol3:99-103, 1985[Medline] 8. Davis RL: Grading of gliomas, Fields WS (ed):Primary Brain Tumors: A Review of Histologic Classification150-158New York, NY, Springer, 1989 9. Levin VA, Wara WM, Davis RL, et al: Phase III comparison of chemotherapy with BCNU and the combination of procarbazine, CCNU and vincristine administered after radiation therapy with hydroxyurea to patients with malignant gliomas. J Neurosurg63:218-223, 1985[Medline] 10. Brada M, Thomas DGT, Bleehen NM, et al: Medical Research Council (MRC) randomised trial of adjuvant chemotherapy in high grade glioma (HGG): BR05. Proc Am Soc Clin Oncol 17:400a, 1998 (abstr 1543) Submitted March 8, 1999; accepted July 14, 1999.
This article has been cited by other articles:
|
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
|
|||||||||||
|
|||||||||||
|
|
Copyright © 1999 by the American Society of Clinical Oncology, Online ISSN: 1527-7755. Print ISSN: 0732-183X
|
60 years of age compared with 10% of the PCV group. The mean age was 42 years in both groups. Approximately 38% of each treatment group were diagnosed by biopsy. KPS was not well balanced between the groups, with 61% of BCNU patients having a KPS of 90 to 100 compared with 73% of the PCV patients (P = .0075). A total of 29 patients treated with BCNU had an oligodendroglial component, and 41 patients treated with PCV had an oligodendroglioma component. 
