External beam irradiation and the combination of cisplatin and carmustine followed by carmustine alone for the treatment of high-grade glioma

A phase 2 Southwest Oncology Group trial

Abstract

BACKGROUND.

The poor prognosis reported for patients with high‒grade glial neoplasms indicates a need for the development of multimodality therapeutic approaches. The addition of chemotherapy has contributed variably to increased survival. The objective of the current study (Southwest Oncology Group [SWOG] 9016) was to determine whether concurrent radiotherapy and chemotherapy with the combination of carmustine and cisplatin could be given safely in a cooperative group setting. Additional objectives included the estimation of response rate, the rate of disease stabilization, and the probability of 1-year survival.

METHODS.

SWOG 9016 study included 59 eligible patients with grade III or IV astrocytoma who received radiotherapy concurrently with carmustine/cisplatin chemotherapy. Patients were required to have either measurable or evaluable disease. The therapeutic endpoints were comprised of complete response (CR), partial response (PR), or progressive disease (PD).

RESULTS.

Six patients achieved a CR (CR rate of 10%; 95% confidence interval [95% CI], 4–21%), 4 achieved a PR (PR rate of 7%; 95% CI, 2–16%), and 2 patients (3%) experienced an unconfirmed response. Twenty-four patients (41%; 95% CI, 28–54%) had stable disease and 10 patients (17%) demonstrated PD. The overall disease stabilization rate (CR + PR + stable disease, excluding unconfirmed response) was 58% (95% CI, 44–70%).

CONCLUSIONS.

Despite the presence of a cohort of long‒term survivors, the results of the current study do not appear to support the additional studyor routine use of concurrent cisplatin and carmustine. Cancer 2008. © 2008 American Cancer Society.

Primary brain tumors represent approximately 2% of deaths due to cancer. The majority of tumors are supratentorial and are comprised of high‒grade gliomas, either anaplastic astrocytoma (AA) or glioblastoma (GBM).1, 2 The prognosis remains dismal despite the standard intervention of surgical resection followed by radiotherapy. However, in 2 prospective, randomized trials by the Brain Tumor Study Group (BTSG), the addition of postoperative radiotherapy was found to extend the median survival from 3.5 months to only 9 months.3, 4

Attempts to improve the patient outcome have included the addition of chemotherapy to conventional therapy. Lipid-soluble nitrosoureas, which cross the blood-brain barrier, are the most commonly used agents. The benefit derived from carmustine and other nitrosoureas has been reported to be relatively small. Meta-analyses of chemotherapy trials have demonstrated a small but significant benefit in 1-year survival from adjuvant chemotherapy.5, 6 The estimated increase in survival was 10.1% at 1 year and 8.6% at 2 years, with the greatest benefit occurring in patients with AA.

Chemotherapeutic agents that have demonstrated promise in in vitro studies have been explored in a variety of settings.7 Platinum compounds possess several features that predict efficacy against gliomas. Cisplatin crosses the blood-brain barrier and the ratio of ultrafiltrate platinum in the cerebrospinal fluid (CSF) to plasma is 0.89.8 After a dose of 100 mg/m2, the peak CSF level of ultrafiltrate platinum (0.8 μg/mL) is within the therapeutic range. Furthermore, in vitro capillary cloning assays confirm that cisplatin has the least cross-resistance with nitrosoureas.9 Several pilot studies have validated cisplatin activity in patients with recurrent gliomas. In 1 study, 4 of 21 patients (19%) with nitrosourea-refractory high‒grade glioma achieved a partial response (PR) with acceptable toxicity after a dose of 100 mg/m² of cisplatin administered on Day 1 and Day 8 every 4 weeks.10

Given the lack of overlapping dose-limiting toxicity and minimal cross-resistance, the combination of cisplatin and carmustine is potentially active in gliomas. This combination has been studied in other solid tumors, but with myelosuppression as the dose-limiting toxicity.11

Accordingly, Southwest Oncology Group study 9016 (SWOG 9016) was designed to determine whether concurrent radiotherapy and chemotherapy with the combination of carmustine and cisplatin could be given safely in a cooperative group setting. Additional objectives included the estimation of response rate, the rate of disease stabilization, and the probability of 1-year survival.

MATERIALS AND METHODS

Eligibility

Requirements included histologically confirmed GBM, AA, gemistocytic astrocytoma (grade 3), malignant glioma (not otherwise specified), or grade 3 astrocytoma. Patients with a complete resection or prior chemotherapy or radiotherapy were ineligible. Other requirements included a SWOG performance status of 0 to 2, measurable or evaluable disease on imaging studies, registration within 4 weeks of surgery, adequate renal function (creatinine within institutional normal limits and creatinine clearance ≥60mg/minute), adequate bone marrow reserve (leukocyte count >4000/mm³ × 10⁹/mm³, a granulocyte count >1500/mm³, and a platelet count >150,000/mm³), and normal hepatic function (bilirubin within institutional normal limits, and aspartate aminotransferase and alkaline phosphatase within twice the upper limits of normal).

Treatment Protocol

Within 5 days of patient registration, chemotherapy was initiated with carmustine at a dose of 80 mg/m² given intravenously over 1 hour on Days 1, 2, and 3, and cisplatin at a dose of 33 mg/m² administered over 1 hour on Days 1, 2, and 3 combined with radiotherapy. The maximal cumulative dose of cisplatin has generally been 400 mg/m² due to dose-limiting sensory neuropathy. In addition, cumulative carmustine doses in excess of 1400 mg/m² are associated with an increased risk of pulmonary fibrosis and myelodysplasia.12 Therefore, the upper dose limit of carmustine in this trial was limited to 1440 mg/m². This treatment was repeated during Weeks 7, 13, and 19. Patients received carmustine alone at the same dose on Weeks 25 and 31. Steroids, anticonvulsants, and other standard supportive care agents (including antiemetics and antibiotics for febrile neutropenia) were continued during chemotherapy. Radiotherapy was initiated within 5 days of registration, corresponding to 4 weeks after surgery. Using opposed lateral or wedged pair portals covering the entire enhanced tumor (by preoperative computed tomography [CT] scan) plus a 3‒cm margin, the total dose was 5940 centigrays (cGy) given in 33 increments at 180 cGy per fraction on a schedule of 5 days per week. The eyes, nasopharynx, and cervical regions were shielded, and after 5400 cGy were received, field reduction excluded the optic nerves and brainstem.

Criteria for Removal From SWOG 9016

Patients who requested removal or experienced unacceptable toxicity or disease progression after at least 2 courses of carmustine and cisplatin were removed from the protocol. If necessary, those patients who experienced clinical deterioration before 2 cycles of chemotherapy were treated with corticosteroids.

Measurement of Response

Patients were required to have either measurable disease (a bidimensionally measurable lesion with clearly defined margins and at least 1 dimension measuring >0.5 cm) or evaluable disease (either unidimensionally measurable lesions or masses with margins not clearly defined). A complete response (CR) was defined as the disappearance of all measurable and evaluable disease. For patients with measurable disease, a PR was defined as an at least 50% decrease under baseline in the sum of the products of the perpendicular dimensions of all measurable lesions. Disease progression was defined as either the smaller of a 50% increase in the sum of the products of all measurable lesions or an increase of 10 cm² or worsening of evaluable disease. Patients who did not qualify for a CR, PR,, or progressive disease were defined as having stable disease. Confirmation of disease status stipulated an assessment being performed 3 to 6 weeks after initial documentation.

Statistical Analysis

For statistical purposes, the primary endpoint of the current study was toxicity, with patients classified according to 1) histology, 2) age <40 years versus age 40 to 59 years versus age ≥60 years, and 3) a SWOG performance status of 0 to 1 versus 2. A sample size of 60 patients was selected to capture relatively infrequent toxicities and to provide reasonable accuracy for the estimation of toxicity and disease stabilization rates in this phase 2 setting. With 60 patients, any toxicity that occurred with at least a 5% chance was likely (probability of at least .95) to be observed, and the incidence of a particular toxicity or the disease stabilization rate could be estimated within at least ±13% (within 95% confidence).

The design specified that if at any point 3 deaths due to treatment were observed, the study would be closed. Alternatively, this regimen was likely (probability of .87) unsuitable if the true probability of toxic death was .08 and unlikely (probability of .02) toxic if the probability was .01.

Survival was measured from the day of registration until death from any cause. Time to treatment failure was measured from date of registration to the date of disease progression or removal from study due to toxicity, refusal, or death. The distribution of survival was estimated by the Kaplan and Meier method.13

RESULTS

Patient Characteristics

From July 1991 through May 1992, 74 patients were entered on the SWOG 9016 study. Informed consent was obtained from all patients before registration. Of the 15 ineligible patients, 11 did not have postoperative measurable or evaluable disease, 2 lacked appropriate pathology, 1 was registered >4 weeks after surgery, and 1 failed to submit baseline information. Per SWOG policy, ineligible patients were not included for efficacy analyses or toxicity assessment.

There was 1 major protocol deviation because the patient did not receive radiotherapy within 1 week of registration.

Patient characteristics are summarized in Table 1. The majority of patients were white, with a median age of 50 years (range, 18-71 years); there was a slight majority of males. Forty-three patients (73%) had GBM and 16 patients (27%) had AA.

Table 1. Patient Characteristics of the SWOG 9016 Study (N = 59)

SWOG indicates Southwest Oncology Group; GBM, glioblastoma; AA, anaplastic astrocytoma.
Age, y
Median	50
Range	18–71
<40	15	25%
40–59	29	49%
≥60	15	25%
Gender
Male	33	56%
Female	26	44%
Race
White	58	98%
Hispanic	1	2%
SWOG performance status
0–1	51	86%
2	8	14%
Histology
GBM	43	73%
AA	16	27%

Response

Of the 59 eligible patients, 6 achieved a CR (CR rate of 10%; 95% confidence interval [95%CI], 4–21%), 4 achieved a PR (PR rate of 7%; 95% CI, 2–16%), and 2 patients (3%) experienced an unconfirmed response. Twenty-four patients (41%; 95% CI, 28–54%) had stable disease and 10 patients (18%) demonstrated progressive disease. The overall disease stabilization rate (CR + PR + stable disease, excluding unconfirmed response) was 58% (95% CI, 44–70%). Thirteen patients had inadequate assessments for response determination. The overall median time to treatment failure was 4 months (95% CI, 2–6 months) (Fig. 1) and was 4 months (95% CI, 2–6 months) for patients with GBM and 6 months for patients with grade III astrocytomas (95% CI, 3–74 months) (Fig. 2).

Figure 1. Time to treatment failure. RT indicates radiotherapy; CDDP, cisplatin; BCNU, carmustine.

Figure 2. Time to treatment failure by histology. Glio Multi indicates glioblastoma multiforme.

Survival

The 1-year survival rate for eligible patients with follow-up was 47% (95% CI, 35–60%) (Fig. 3). When survival was evaluated by histology, the 1-year survival rate for patients with GBM was 44% (95% CI, 29–59%) and was 56% (95% CI, 32–81%) for patients with AA (Fig. 4). It is interesting to note that the 5‒year survival rate of patients with AA was nearly 40%, compared with 7% for patients with GBM. At the time of last follow‒up, 3 patients with AA were alive, >6 years after study entry. These patients include a 31‒year‒old woman with a SWOG performance status of 0 to 1 and a survival time at the time of last follow‒up of 80 months, a 36‒year‒old man with a SWOG performance status of 0 to 1 and a survival time of 118 months, and a 59‒year‒old woman with a SWOG performance status of 0 to 1 and a survival time of 129 months.

Figure 3. Overall survival by treatment arm. RT indicates radiotherapy; CDDP, cisplatin; BCNU, carmustine.

Figure 4. Overall survival by histology. Glio Multi indicates glioblastoma multiforme.

Advanced age was also found to adversely affect survival time. One-year survival rates for patients aged <40 years, those ages 40 to 59 years, and patients aged ≥60 years were 60% (95% CI, 35–85%), 52% (95% CI, 34–70%), and 27% (95% CI, 4–49%), respectively (Fig. 5). The 2‒year and 5-year overall survival rates are depicted in Table 2.

Figure 5. Overall survival by patient age.

Table 2. 2-Year and 5-Year Overall Survival Rates

	Overall survival
	2-year rate	5-year rate
All patients	25% (14–37%)	15% (6–24%)
By age, y
<40	60% (35–75%)	47% (21–72%)
40–59	21% (6–35%)	7% (0–16%)
≥60	0%	0%
By histology
Other histologies	50% (26–75%)	38% (14–61%)
Glioblastoma multiforme/grade IV	16% (5–27%)	7% (0–15%)

The median overall survival for all patients was 10 months (95% CI, 9–16 months).

Toxicity

In the 59 eligible patients, toxicity was evaluated according to standard SWOG criteria.14 Approximately 80% of these patients had at least 1 toxicity that was higher than grade 3. Significant toxicities (grades 3–5) are summarized on Table 3. The major toxicity associated with carmustine and cisplatin was myelosuppression with significant granulocytopenia and thrombocytopenia (85%). Of the 20 patients who completed all 6 cycles of chemotherapy, only 1 patient did not require a dose reduction due to myelosuppression. Of the 59 patients studied, 35 (59%) experienced grade 3 or 4 thrombocytopenia. Leukopenia was observed in 81% of patients and granulocytopenia was evident in 59% of patients, with 26 patients (44%) exhibiting grade 3 or 4 granulocytopenia. Although myelosuppression was relatively severe, there were only 2 documented infections, neither of which was life‒threatening. Only 1 patient experienced significant hemorrhage related to thrombocytopenia; however, myelosuppression led to significant dose reductions for patients receiving treatment.

Table 3. Toxicities (N = 59)*

Category	Adverse event	Grade
		1	2	3	4	5
AST indicates aspartate aminotransferase; ALT, alanine aminotransferase. * Shown as the number of patients with a given type and grade of adverse event. Adverse events that were unlikely or not related to treatment were excluded. Adverse events with no entries for grades 3 to 5 were suppressed.
Circulatory	Edema	0	0	1	0	0
	Hypotension	1	1	1	0	0
	Phlebitis/thrombosis/embolism	0	0	1	0	0
Gastrointestinal	Diarrhea	3	1	1	0	0
	Nausea	18	12	6	0	0
	Vomiting	12	9	2	2	0
Hematologic	Anemia	19	14	6	0	0
	Granulocytopenia	5	4	10	16	0
	Leukopenia	7	17	19	5	0
	Thrombocytopenia	8	7	19	16	0
Infection	Infection	0	0	2	0	0
	Respiratory infection	0	0	1	0	0
Liver	Transaminase (AST, ALT) increase	1	1	1	0	0
Lung	Pneumonitis/effusions/infiltration	0	0	1	0	0
Miscellaneous	Weight loss	1	0	2	0	0
Mucosal	Pharynx/esophagitis	0	0	1	0	0
Neurologic	Dizziness/vertigo	7	0	2	0	0
	Malaise/fatigue/lethargy	21	5	8	0	0
	Uncoordination/ataxia	1	0	1	0	0
	Speech impairment	0	0	1	0	0
	Weakness	0	0	4	0	0
	Hearing	3	4	1	0	0
	Vision	0	0	2	1	0
Maximum grade of any adverse event
No.		4	6	23	24	0

There were no acute treatment-related deaths reported. After the completion of treatment, 1 ineligible patient died of respiratory complications associated with pulmonary infiltrates believed to be secondary to treatment. Other significant toxicities included grade 3 fatigue and lethargy (8 patients) and grade 4 vomiting requiring hospitalization (2 patients). Grade II or greater neurologic toxicities reported include visual changes (3 patients), ataxia (1 patient), dizziness (2 patients), hearing (5 patients), and speech (1 patient). There was only 1 eligible patient for whom a thromboembolic complication (phlebitis) was reported.

Treatment Delivery

Myelosuppression often precluded the effective delivery of chemotherapy. Only 1 patient received all 6 cycles of chemotherapy (a 34‒year‒old with AA). By the third cycle (Week 13), only 33 patients were still receiving chemotherapy (therapy was discontinued in the others either due to disease progression, refusal, or toxicity, most commonly prolonged cytopenia). Of these patients, only 12 were able to receive full doses of chemotherapy.

DISCUSSION

Chemotherapy resistance is partially responsible for the poor outcome of patients with high‒grade glioma. Resistance mechanisms are multifactorial, including poor penetration of the blood-brain barrier, decreased drug uptake resulting in low intracellular concentration, high drug efflux, repair of drug-induced damage, and metabolic drug deactivation.15 Despite this well-documented resistance, standard therapy with adjunct chemotherapy has been reported to provide only a modest survival benefit.16–19 In a meta-analysis, Stewart documented a 2-month increase in the median survival when chemotherapy was combined with radiotherapy.6 Other investigators have documented that the combination of procarbazine, lomustine, and vincristine (PCV) offers patients with AA a median survival of 39.3 months versus 20.5 months for those receiving carmustine alone.20 However, recent trials have not confirmed a significant advantage of PCV over carmustine alone for either AA21 or GBM.22 Since the SWOG 9016 study closed, radiotherapy combined with temozolomide has been established as the standard therapy for patients with newly diagnosed GBM. The median survival for patients treated with this regimen is 14.6 months, with a 2‒year survival rate of nearly 50% for patients with MGMT promoter methylation.23, 24

The 3 long-term survivors in the current study were patients with AA. Their survivals were much longer than even the most favorable (recursive partitioning analysis [RPA] class II) expected median survival for a similar RPA class.25 Although this cohort was not statistically significant, we can speculate that certain patients with AA may benefit from current or other adjuvant chemotherapy regimens. In addition, molecular diagnostic testing of this cohort might have revealed a profile that was predictive of longer survival. However, in the current study, tissue was not available for additional studies. Recent and ongoing cooperative trials may answer important questions regarding molecular predictors of treatment response and long-term survival.

Other groups have investigated combined therapy with cisplatin and carmustine.26 The Eastern Cooperative Oncology Group (ECOG) administered infusional carmustine and cisplatin before radiotherapy.27, 28 Of the 50 patients treated, 1 patient achieved a CR, 10 patients demonstrated a PR, and 18 had stable disease after 3 cycles of chemotherapy. Based on these results, an ECOG/SWOG intergroup phase 3 protocol (ECOG 2394/SWOG 9508) was proposed for newly diagnosed GBM patients. A total of 223 patients, 219 of whom were eligible, were randomized to receive infusional cisplatin and carmustine before radiotherapy or standard radiotherapy with adjuvant carmustine. No difference was observed with regard to the median survival. Approximately 56% of patients finished all 3 cycles with serious toxicity, primarily hematologic, and toxicity was reported to occur more often than in patients who received combined cisplatin and carmustine.28 Another phase 3 cooperative group trial, North Central Cancer Treatment Group (NCCTG) 93-72-52/SWOG 9503, compared the addition of carmustine to cisplatin before and concurrent with standard radiotherapy versus accelerated radiotherapy. The results of this trial of 401 eligible patients demonstrated no statistical difference between standard radiotherapy and accelerated radiotherapy. The addition of cisplatin to concurrent carmustine and radiotherapy resulted in increased toxicity without improved survival.29

In the SWOG 9016 study, the majority of patients developed a significant hematologic toxicity, although there were no life-threatening infections or episodes of life-threatening hemorrhage reported. These responses were disappointing compared with the single‒institution and multicenter experiences that preceded this study. One explanation may be that only patients with an incomplete surgical resection were eligible for the current study. Furthermore, hematologic toxicity contributed to decreased dose intensity and the delivery of carmustine when compared with standard treatment.

The results of the current study do not support the additional study or routine use of concurrent cisplatin and carmustine. Our observed median survival of 10 months is inferior to the estimated survival of a cohort with a similar tumor grade, patient age, and performance status as per the RPA classification.25 Nevertheless, efforts continue to evaluate new chemotherapeutic agents and combinations.30, 31 Temozolomide in particular has exhibited significant activity in patients with high‒grade glioma.23 The implantation of carmustine wafers followed by radiotherapy for the treatment of newly diagnosed glioblastoma has also added a small survival advantage over the use of radiation alone.32 Recent advances in understanding the drug resistance properties of glioma cells may hold the greatest promise for increasing the survival of these patients.33, 34

In the current study, the 1‒year survival rate of 44% for patients with GBM and the substantial toxicity reported are simply unacceptable. Compared with the current standard of radiotherapy and temozolomide, the combination of carmustine and cisplatin is not recommended for the initial treatment of patients with high-grade glioma.

Acknowledgements

We wish to acknowledge the participation of institutions that registered patients to this study. We also thank Calleleh Bonugli, Bonnie Granados, Dona Marrah, Patricia O'Kane, Angela Ribble, and Susan Schulman for assistance in conducting the study and preparing the article.