Fax: (813) 745-3713
Salvage chemotherapy with cyclophosphamide for recurrent temozolomide-refractory anaplastic astrocytoma
Article first published online: 1 DEC 2005
Copyright © 2005 American Cancer Society
Volume 106, Issue 1, pages 172–179, 1 January 2006
How to Cite
Chamberlain, M. C., Tsao-Wei, D. D. and Groshen, S. (2006), Salvage chemotherapy with cyclophosphamide for recurrent temozolomide-refractory anaplastic astrocytoma. Cancer, 106: 172–179. doi: 10.1002/cncr.21582
- Issue published online: 23 DEC 2005
- Article first published online: 1 DEC 2005
- Manuscript Accepted: 27 JUL 2005
- Manuscript Revised: 23 JUN 2005
- Manuscript Received: 12 MAY 2005
- recurrent anaplastic astrocytoma;
A prospective Phase II study of cyclophosphamide (CYC) was conducted in adult patients with recurrent temozolomide-refractory anaplastic astrocytoma (AA) with a primary objective of evaluating 6-month progression-free survival (PFS).
Forty patients (28 men, 12 women) ages 26–57 years (median, 43 yrs) with neuroradiographically recurrent AA were treated. All patients had previously been treated with surgery and involved-field radiotherapy. Additionally, all patients were treated with temozolomide (TMZ) chemotherapy after radiotherapy. All patients were treated at recurrence with CYC administered intravenously on 2 consecutive days (750 mg/m2/day) every 4 weeks (operationally defined as a single cycle). Neurologic and neuroradiographic evaluation were performed every 8 weeks.
All patients were evaluable. A total of 215 cycles of CYC (median, 4 cycles; range 2–12 cycles) was administered. CYC-related toxicity included alopecia (all patients, 100%), anemia (5, 12.5%), thrombocytopenia (6, 15%), and neutropenia (8, 20%). Four (10%) patients required transfusion. Nine patients (22.5%) (95% confidence interval [95% CI], 11%–39%) demonstrated a neuroradiographic partial response, 16 patients (40.0%) (95% CI, 25%–57%) demonstrated stable disease, and 15 patients (37.5%) (95% CI, 23%–54%) had progressive disease after 2 cycles of CYC. Time to tumor progression ranged from 2–19 months (median, 4 mos; 95% CI, 2–6 mos). Survival ranged from 2–26 months (median, 8 mos; 95% CI, 6–10 mos). The 6-month and 12-month PFS was 30% and 8%, respectively.
CYC demonstrated modest efficacy with acceptable toxicity in this cohort of adult patients with recurrent anaplastic astrocytoma, all of whom had failed prior TMZ chemotherapy. Cancer 2006. © 2005 American Cancer Society.
The treatment of recurrent high-grade gliomas is problematic because only partially effective therapeutic modalities are available. These therapies include chemotherapy, radioactive implants, stereotactic radiotherapies, immunotherapy, and reoperation. 1–12 Chemotherapy for recurrent malignant primary brain tumors is of modest benefit, primarily because response to chemotherapy is of limited duration. In an analysis of eight institutional Phase II studies of chemotherapy for recurrent high-grade gliomas, Wong et al. 13 reported that the response rate in recurrent anaplastic astrocytoma (AA) was 14% and progression-free survival (PFS) at 6 months was 31%. The drugs that are the most active are nitrosoureas such as carmustine and lomustine, in addition to temozolomide (TMZ), procarbazine, cis-retinoic acid, and platinum compounds. 1, 2, 4, 5, 10–12, 14–17 Another chemotherapy with purported activity in recurrent high-grade gliomas is cyclophosphamide (CYC). 18–20
The primary objective of this single-institution prospective Phase II trial was to observe whether CYC given at a dose of 750 mg/m2/day for 2 consecutive days every 4 weeks could significantly delay progression in patients with neuroradiographically recurrent AA. Forty adult patients with recurrent supratentorial AA previously treated with surgery, radiotherapy, and TMZ chemotherapy regimen were entered into the study.
MATERIALS AND METHODS
The study was performed at the University of Southern California, Norris Comprehensive Cancer Center and Hospital. The study was begun in October 1999 and closed in January 2004. Approval of the protocol and informed consent by the university human investigation committee was obtained. Informed consent was obtained from each subject.
Objectives and Endpoints
The two primary objectives of this study included determination of efficacy and toxicity of CYC in the treatment of patients with TMZ-refractory recurrent or progressive AA. The primary endpoint was PFS at 6 months (6-mo PFS). Secondary endpoints included overall survival, time to progression, and response. Toxicity was evaluated in all eligible patients receiving at least one cycle of CYC.
Patients were required to have a histologically proven AA that was recurrent neuroradiographically. Patients must have progressed after definitive radiotherapy and TMZ chemotherapy. At least 4 weeks must have elapsed since the last dose of chemotherapy and patients must have recovered from the adverse effects of prior therapy. Patients could not have received prior CYC therapy. Patients were required to have radiographically measurable intracranial disease wherein recurrent tumor was bidimensionally measurable (at least 1 × 1 cm) by cranial contrast-enhanced magnetic resonance imaging (MRI). Histological confirmation of tumor recurrence was not required for entry into the study. Pregnant or lactating women were not permitted to participate. Patients of child-bearing potential were required to implement adequate contraceptive measures during participation in this study. Patients must have had an Eastern Cooperative Oncology Group performance status of 0–2 (Karnofsky score ≥60) and a life expectancy greater than 3 months.
Adequate hematologic, renal, and hepatic functions were required and were defined by the following: absolute granulocyte count >1500/dL or leukocyte count >4000/dL, platelet count >100,000/dL, total bilirubin level <1.8 mg/dL, transaminase level <4 times the upper limit of normal, and creatinine concentration <1.8 mg/dL (or creatinine clearance ≥60 mL/m2/1.73).
All patients were aware of the neoplastic nature of their disease and willingly consented to participate after being informed of the procedures to be used, experimental nature of the therapy, alternatives, potential benefits, side effects, risk, and discomforts. Patients with carcinomatous meningitis were not eligible. No serious concurrent medical illnesses or active infection could be present that would jeopardize the ability of the patient to receive CYC therapy. Patients could not have an active concomitant malignancy except skin carcinoma (squamous cell or basal cell). Patients could range in age from 18–80 years.
CYC (Cytoxan; Meade Johnson Pharmaceuticals, Princeton, NJ) was administered to all patients at a dose of 750 mg/m2 given intravenously (i.v.) over 30 minutes on 2 consecutive days. 20 Concurrent dexamethasone was permitted for control of neurologic signs and symptoms. Premedication included ondansetron at a dose of 0.15 mg/kg and dexamethasone at a dose of 4.0 mg, both administered i.v. Prechemotherapy hydration utilized 1 L of normal saline given i.v. over 2 hours. Neither mesna nor posthydration was administered.
Postchemotherapy medication included prochlorperazine for nausea or emesis. CYC administration (at a dose of 1500 mg/m2) was repeated 4 weeks after the initial dose. A cycle of therapy was operationally defined as 28 days during which CYC was administered on Days 1 and 2. Treatment with CYC was repeated every 28 days from Day 1 provided that all hematologic toxicity from the previous cycle had resolved to National Cancer Institute (NCI) Grade 2 or less, and all nonhematologic toxicity had recovered to either Grade 1 or less. If recovery had not occurred by Day 28, the subsequent cycle of CYC was delayed until these criteria were met. All toxicities, including hematologic, resulting from CYC therapy were rated according to the NCI Common Toxicity Criteria (version 2.0).
No dose escalations were permitted. Dose reduction for toxicity was by 25% in patients with Grade ≥3 toxicity and only 2 dose reductions were allowed. Patients having Grade 3 toxicity of any type after two dose reductions were removed from the study.
Oral dexamethasone was used concurrently in 26 patients and was increased in 8 patients with clinical disease progression. Dexamethasone dose was decreased in six patients as patient clinical status permitted.
Method of Evaluation
Blood counts were obtained weekly, neurologic examination was performed every 4 weeks, and contrast-enhanced cranial MRI was performed every 8 weeks after even-numbered cycles of CYC.
Neuroradiographic response criteria as defined by Macdonald et al. 21 were used. Complete response (CR) was defined as the disappearance of all enhancing or nonenhancing tumor on consecutive computed tomography (CT) or MRI scans at least 1 month apart, with the patient not receiving corticosteroids and neurologically stable or improved. Partial response (PR) was defined as a >50% reduction in the size of tumor on consecutive CT or MRI scans at least 1 month apart, with the corticosteroid dose stable or decreased and the patient neurologically stable or improved. Progressive disease (PD) was defined as a greater than 25% increase in the size of tumor or any new tumor on CT or MRI scans, or the patient being neurologically worse with a stable or increased corticosteroid dose. Stable disease (SD) was defined as all other situations. SD response as with CR and PR required a confirmation MRI scan 1 month after documenting the best response.22–31
In patients with SD, PR, or CR, two additional cycles of CYC were to be administered, after which patients were assessed again as described. Patients were continued to receive CYC therapy until documentation of PD, at which time patients were removed from the study and were either monitored or offered alternative therapy.
PFS and overall survival (OS) were defined as the time from the first day of treatment with CYC until PD or death (PFS) or death (OS). Patients were removed from the study if there was PD, the development of unacceptable toxicity, patient refusal, or noncompliance with protocol requirements.
Experimental Design and Statistical Methods
The primary objective was to determine whether CYC could significantly delay progression in patients with recurrent AA. Historical values were obtained from analysis of a database of 350 patients with recurrent high-grade glioma (125 AA cases) who were treated on consecutive prospective Phase II trials in which the 6-month PFS was 31% for AA. 13 The hypothesis tested were H0: P ≤ p0 versus H1: P ≥ p1, in which P is the probability of remaining alive and free of disease progression at 6 months, with a Type I error, α, ≤ 0.05 and a Type II error β ≤ 0.20. For AA, p0 was set at 0.25 and p1 at 0.45, looking for an improvement of 0.20. The current study was designed to accrue 40 AA patients. For AA patients, success was defined as observing more than 18 of 40 patients alive and free of disease progression at 6 months (yielding α = 0.03 and β = 0.21). The associations of OS and PFS with patient baseline characteristics were tested using the log-rank test. 32 The Pike 33 estimate of relative risk based on the log-rank test was used to provide a quantitative summary of the data, with 95% confidence intervals (95% CIs). 34, 35 Initially, univariate survival analyses were used to evaluate the association of all the prognostic factors in with survival and PFS. All factors with P < 0.15 (i.e., age, location of tumor, adjuvant chemotherapy, and surgery type, and whether the patient underwent a reoperation after the first recurrence, and patient performance status) were included in the Cox proportional hazards model for multivariate analyses. Age and location of the tumor remained the statistically significantly associated OS and PFS at the 0.05 level, based on the likelihood chi-square ratio test. The median survival, time to disease progression, and the associated 95% CIs were computed. Kaplan–Meier plots 36 were constructed to display the estimated probabilities of overall survival and time to progression.
Forty patients (28 men and 12 women) ages 26–57 years (median, 43 yrs), with recurrent AA (original pathology reviewed and confirmed in all cases) (Table 1) were treated with CYC. Recurrent AA was defined by objective neuroradiographic disease progression (>25% increase in tumor size) as compared with prior baseline neuroradiographic images using the criteria reported by Macdonald et al. 21 All neuroradiography was reviewed at the study institution both by neuroradiologists blinded to treatment (two radiologists) and by the first author (M.C.C.). All patients underwent cranial MRI scans demonstrating PD within 2 weeks of CYC administration. Eleven patients (27.5%) underwent a reoperation (complete resection in 7 patients and a partial resection in 4 patients) in which repeat tumor histology was consistent with AA.
|Factors||No. of patients|
|Age in yrs|
|Location of tumor|
|Type of surgery|
|Radiotherapy in Gy|
|Adjuvant chemotherapy (TMZ)|
|> 8 cycles||18||45%|
|Best response to TMZ chemotherapy|
|Resurgery after first recurrence|
|≤ 4 cycles||21||53%|
|> 4 cycles||19||48%|
|Best response to CYC treatment|
Patients presented at the time of tumor recurrence with the following signs and symptoms: increased intracranial pressure as manifested by increasing headache (n = 26), worsening seizures (n = 12), altered mental status (n = 6), progressive hemiparesis (n = 6), new onset of homonymous hemianopsia (n = 2), and gait ataxia (n = 2). Of the 12 patients with worsening seizures, seizure semiology was as follows. Patient performance status using the Karnofsky scale ranged from 70 to 100 (median, 80) at the time of documented tumor recurrence and initiation of CYC therapy. Tumor locations, including multilobar tumors, were as follows: frontal lobe (n = 16), temporal lobe (n = 9), parietal lobe (n = 5), occipital lobe (n = 4), insula (n = 3), and thalamus (n = 3). Thirty-seven patients had lobar tumors and 3 patients had multilobar tumors. Pathology (reviewed by two neuropathologists) showed all tumors to be AA based on World Health Organization (WHO) criteria.
All patients had been treated previously with surgery in which a complete resection was accomplished in 15 patients (resection of all visible contrast-enhancing tumor confirmed in the immediate postoperative period), a partial resection in 16 patients, and biopsy only in 9 patients (Table 1). Eleven patients (27.5%) underwent a second surgery before study entry.
All patients had previously been treated with adjuvant limited-field radiotherapy (Table 1), and in all conventional fractionated radiotherapy was used in which 1.8–2.0 grays (Gy) were administered daily, with a median tumor dose of 60 Gy (range, 59–61 Gy). No patient was treated with stereotactic radiotherapy. Three patients appeared to progress during radiotherapy as demonstrated by preradiotherapy and postradiotherapy cranial MRI comparisons, although they subsequently had disease stabilization with adjuvant TMZ.
All patients were treated with TMZ chemotherapy after radiotherapy (Table 1) administered in the standard 5-day schedule. Patients received a median of 8 cycles of TMZ therapy (range, 2–12 cycles). All patients began CYC therapy immediately after documentation of tumor progression after treatment with TMZ as demonstrated by neuroradiographic progression (in all patients) or clinical disease progression (60% of patients). The median time to the initiation of CYC after initial surgery was 14.5 months (range, 12–28 mos). A total of 215 cycles of CYC were administered. A minimum of 2 cycles of CYC was administered to each patient with a median of 4 cycles (range, 2–12 cycles). CYC was administered at the proscribed dose in all patients. No other antiglioma agents aside from dexamethasone were utilized during the study. All patients tolerated the 1-L of normal saline prehydration without difficulty.
Toxicity was recorded for all grades for all patients by type using the NCI Common Toxicity Criteria. Table 2 lists all Grade 3–4 toxicities observed, with each figure representing the sum of the highest grade of toxicity attained, per toxicity, per cycle for all patients. A total of 215 treatment cycles were administered, 51 of which (23.7%) were Grade 3 adverse events (AEs) and 4 of which (1.9%) were Grade 4 AEs. No Grade 5 toxicity was observed. The most common Grade 3–4 AEs were leukopenia (7.9%), fatigue (4.2%), granulocytopenia (3.7%), thrombocytopenia (2.8), and anemia (2.3%). All patients developed CYC-related alopecia. Four patients required transfusion, two with packed red blood cells and two with platelets. One patient developed febrile neutropenia; however, body fluid cultures were negative. No treatment-related deaths occurred.
|Toxicitya||Grade 3||Grade 4||Total|
|Infection, without neutropenia||1||0||1|
All patients were assessable for survival and 33 patients for response (Fig. 11). After 2 cycles of CYC (at a dose of 1500 mg/m2 followed in 4 weeks by 1500 mg/m2), 15 patients (37.5%) demonstrated PD. Sixteen patients (40%) received 6 or more cycles of therapy. At the conclusion of CYC, Karnofsky performance status ranged from 40–70, with a median of 60 in the entire study group. Patients who failed to respond to CYC were offered alternative or supportive therapy. Survival in the entire cohort ranged from 3–26 months, with a median of 8 months (95% CI, 6–10 months). The probability of survival at 12 months was 8% ± 7%. At the time of last follow-up, all patients had died, and all deaths were directly attributable to the effects of progressive intracranial tumor.
Nine patients (22.5%) demonstrated a neuroradiographic PR (95% CI, 11%–39%) and 16 patients (40%) demonstrated SD (95% CI, 25%–57%). In patients with either a neuroradiographic response or SD (25 patients; 62.5%), the median time to tumor progression was 6 months (range, 3–19 mos; 95% CI, 5–10 mos) and the median survival was 10 months (range, 6–26 mos; 95% CI, 8–15 mos).
There was a trend to suggest that patients who had progressed on TMZ were more likely to progress with CYC (P = 0.038, Mantel–Haenszel chi-square test), but no other clear association was observed between response to CYC and response to the prior regimen of TMZ. Of the 5 patients who experienced a PR to TMZ, only 1 (20%) achieved a PR to CYC, and 2 patients (40%) experienced SD with CYC. Of the 30 patients who experienced SD with TMZ, 7 (23%) achieved a PR and 14 (47%) experienced an SD. In the 5 patients who developed PD while receiving TMZ, 1 (20%) achieved a PR with CYC and none experienced SD. No difference was observed in the pretreatment tumor volume in patients with either a CYC PR or SD compared with patients with PD. 37, 38
The overall probability of PFS at 6 months was 30% (95% CI, 16–44%) and that at 1 year was 8% (95% CI, 0–16%). The overall median time to tumor progression was 4 months (range, 2–19 mos; 95% CI, 2–6 mos). With regard to the primary endpoint of the study (6-mo PFS), the results failed to achieve the specified critical value (i.e., only 12, and not the required 18, patients experienced a PFS of greater than 6 mos [AA patients: expected 45%, observed 30%]). 13
Prados et al. 22 in a seminal article reporting on the outcome of patients with AA concluded that these patients can be expected to have a median survival of over 3 years, that young age and high Karnofsky performance status have a positive influence on survival, and that salvage therapies may extend survival after the onset of tumor progression for nearly a year. In a separate European analysis of anaplastic gliomas (predominantly AA), the median overall survival was 29 months with a 5-year probability of survival of 38%. 23 In another analysis by Prados et al., 2 the Radiation Therapy Oncology Group (RTOG) database was reviewed and compared patients with newly diagnosed AA treated according to protocol with either carmustine or procarbazine, lomustine, and vincristine (PCV) adjuvant chemotherapy after surgery and conventional external beam radiotherapy. Using this retrospective analysis, there did not appear to be any survival benefit to PCV adjuvant chemotherapy. The study further posited that the inclusion of chemotherapy in the treatment of newly diagnosed AA patients, although common practice, was not yet defined pending a randomized study. The Medical Research Council Brain Tumor Working Group reported on the largest randomized trial comparing patients treated with radiotherapy alone with those treated with radiotherapy plus PCV chemotherapy. 1 Of 594 eligible patients, 19% (113 patients) had AA histology. There was no difference noted in survival (13-mo median survival for the radiotherapy-only group); however, a 2-month increase in survival was noted for the PCV arm (2-yr survival rate 37% for the radiotherapy-only arm vs. 42% for the radiotherapy plus PCV arm). In the recent meta-analysis by the Glioma Meta-Analysis Trialists Group of 12 randomized trials, adjuvant chemotherapy improved the 2-year survival by 6% in AA patients (31% vs. 37%). 24 These studies suggest a modest benefit for the inclusion of chemotherapy in the adjuvant treatment of AA patients, although the choice of adjuvant chemotherapy agent has evolved. Since the introduction of TMZ in clinical practice in 1999, TMZ has established efficacy in the treatment of recurrent AA and modest TMZ-related toxicity, and increasingly TMZ has been used as an adjuvant chemotherapeutic agent in place of nitrosoureas such as carmustine for the treatment of newly diagnosed AA. 16 RTOG and the Southwest Oncology Group (SWOG) are currently comparing adjuvant TMZ with carmustine in a randomized clinical trial in newly diagnosed patients with AA, in what to our knowledge is the first randomized trial to date to directly compare TMZ with carmustine. The current study utilized TMZ after radiotherapy, as is increasingly common neurooncology practice and is supported in part by the recent European Organization for Research and Treatment of Cancer (EORTC) trial of adjuvant TMZ for glioblastoma multiforme, which demonstrated a survival benefit when compared with historical controls (16 mos vs. 11 mos). 25
How best to manage recurrent AA remains ill-defined, notwithstanding a variety of studies. However, most studies (similar to the current study), are Phase II nonrandomized trials comparing outcome with historical controls. Yung et al., 16 in a randomized trial of TMZ versus procarbazine for recurrent AA, demonstrated an advantage for TMZ with a 6-month PFS of 46%, an objective neuroradiographic response rate of 35%, and an OS of 13.6 months. In the study by Yung et al., 60% of patients had received adjuvant carmustine, 18% underwent reoperation at the time of recurrence, and the median time to tumor recurrence was 15.2 months. These figures are similar to the current study with the exception that 100% of patients were treated with adjuvant chemotherapy. Brem et al. 26 reported on another randomized trial of patients with recurrent high-grade glioma and compared surgery with or without the placement of biodegradable polymers of carmustine (Gliadel; MGI Pharma, Bloomington, MN). Twenty-eight of the 88 patients (32%) enrolled in this study had AA. The study demonstrated a 35% improvement in overall survival (31 wks vs. 23 wks and a 50% increase in 6-mo PFS [64% vs. 44%]). Although 32% of patients entered into the study had AA, a separate analysis of this group was not reported. These robust results suggest that reoperation is of benefit in patients with AA and when a near-complete resection can be performed, carmustine is an effective therapy. Unfortunately, only a minority of patients (28% as seen in the current study) with recurrent AA are candidates for reoperation. Therefore, the majority of patients, if desirous of further therapy, are offered chemotherapy. Building on the strength of TMZ, Jaeckle et al. 27 recently reported on the combination of TMZ and cis-retinoic acid (Accutane; Roche, Nutley, NJ) for the treatment of patients with recurrent high-grade glioma, 22% of whom were chemotherapy naive. A 46% 6-month PFS and 47-week median OS was noted in patients with AA. These are compelling results, although the dataset of patients with AA was small (n = 28). Two other investigational treatments for recurrent high-grade glioma, convection-enhanced intratumoral delivery of radiopharmaceutical toxins and molecularly targeted therapies such as small molecule inhibitors of receptor tyrosine kinases, currently are under active study. 28, 29 Consequently, the issue of how to incorporate these treatments into the care of patients with recurrent AA outside of investigational trials is unclear.
The current study was directed at the AA population that has failed prior chemotherapy (TMZ in 100%) and for whom further treatment appeared warranted. The study did not require histologic proof of recurrent AA and the possibility of radiation necrosis as opposed to recurrent tumor is possible. This appears unlikely for the following reasons. No patient received stereotactic radiotherapy and the risk of radiation necrosis is <5% in patients treated with standard fractionated radiotherapy. Furthermore, 12 patients (30%) underwent 18Ffluorodeoxyglucose-positron emission tomography and 14 patients (35%) underwent MRI spectroscopy, in whom recurrent viable tumor was radiographically confirmed. Lastly, 11 patients (27.5%) underwent reoperation, in whom histopathology was confirmed as AA. The study did not require reexamination of histology and therefore, a proportion of patients assumed to have AA may have progressed to glioblastoma. Therefore, the study may be evaluating both patients with AA and secondary glioblastoma.
CYC appears attractive, as prior single-agent studies have suggested activity against high-grade gliomas and, in addition, CYC is a core agent in the infant chemotherapy regimens in use today. 18, 20, 30 Furthermore, CYC toxicity is manageable (an approximately 25% rate of Grade 3–4 toxicity in the current study) and noncumulative, permitting administration without growth factor support. Lastly, anticonvulsant medication (almost universally used in patients with AA) (in particular, hepatic mixed function cytochrome P450-inducing drugs such as phenytoin and carbamazepine) up-regulate chemotherapy catabolism. This principle, first articulated by Fetell et al., 31 results in chemotherapy underdosing due to enhanced hepatic metabolism induced by heterocyclic anticonvulsant drugs. Using an alkylator-based therapy such as CYC, which does not have pharmacodynamic interactions with enzyme-inducing anticonvulsant drugs, obviates this problem.
CYC used at this dose and in this schedule in patients with previously treated TMZ-refractory recurrent AA appears to be of modest benefit (6-mo PFS of 30%). Outside the availability of a clinical trial, CYC is a reasonable choice for patients with recurrent AA who have previously failed TMZ. Nonetheless, with regard to the primary endpoint of the study (6-mo PFS), the results failed to exceed the 20% threshold for success, assuming a 20% improvement compared with the database reported by Wong et al. 13 (AA: expected 45%, observed 30%).
- 1The Medical Research Council Brain Tumor Working Party. Randomized trial of procarbazine, lomustine, and vincristine in the adjuvant treatment of high-grade astrocytoma: a medical research council trial. J Clin Oncol. 2000; 19: 509–518.
- 4Phase III study comparing three cycles of infusional carmustine and cisplatin followed by radiation therapy with radiation therapy and concurrent carmustine in patients with newly diagnosed supratentorial glioblastoma multiforme: Eastern Cooperative Oncology Group Trial 2394. J Clin Oncol. 2003; 21: 1485–1491., , , et al.
- 15Carboplatin and recurrent childhood brain tumors. J Clin Oncol. 1987; 5: 759–763., , , et al.
- 31Clinical and pharmacologic study of preirradiation Taxol administrated as a 96-hour infusion in adults with newly diagnosed glioblastoma multiforme [abstract]. Proc Am Soc Clin Onco 1994; 13: 179., , .
- 32Survival analysis. New York: John Wiley & Sons, 1981: 114–118.
- 33Contribution to the discussion on the paper by R. Peto and J. Peto, ‘Asymptotically efficient rank invariant procedures.’ J R Stat Soc, Ser A. 1972; 135: 201–203..
- 35Statistical models and methods for lifetime data. New York: John Wiley & Sons, 1982: 345–354..