Impact of volume of irradiation on survival and quality of life in glioblastoma: a prospective, phase 2, randomized comparison of RTOG and MDACC protocols

Abstract

Background

Though conformal partial-brain irradiation is the standard adjuvant treatment for glioblastoma, there is no consensus regarding the optimal volume that needs to be irradiated. European Organisation for Research and Treatment of Cancer (EORTC) and The University of Texas MD Anderson Cancer Center (MDACC) guidelines differ from the Radiation Therapy Oncology Group (RTOG) in their approach toward peritumoral edema, whereas RTOG and MDACC guidelines differ from EORTC in the concept of boost phase. A scarcity of randomized comparisons has resulted in remarkable variance in practice among institutions.

Methods

Fifty glioblastoma patients were randomized to receive adjuvant radiotherapy using RTOG or MDACC protocols. Apart from dosimetric and volumetric analysis, acute toxicities, recurrence patterns, progression-free survival (PFS), overall survival (OS), and quality of life (QoL) were compared using appropriate statistical tests.

Results

Both groups were comparable with respect to demographic characteristics. Dosimetric analysis revealed significantly lower boost-phase planning treatment volumes and V60 Gy in the MDACC arm (chi-squared, P = .001 and .013, respectively). No significant differences were observed in doses with respect to organs at risk, acute toxicity, or recurrence patterns (chi-squared, P > .05). On the log-rank test, median PFS (8.8 months vs 6.1 months, P = .043) and OS (17 months vs 12 months, P = .015) were statistically superior in the MDACC group.

Age, extent of resection, and proportion of whole brain receiving prescription dose were associated with improved PFS and OS on regression analysis. QoL of patients was significantly better in the MDACC group in all domains except cognitive, as assessed with the EORTC Quality of Life Questionnaire (QLQ-C30) and Brain Cancer Module (QLQ-BN20) (general linear model, P < .05).

Conclusions

Use of limited-margin MDACC protocol can potentially improve survival outcomes apart from QoL of glioblastoma patients, as compared with the RTOG protocol.

Glioblastomas are highly aggressive malignant neoplasms arising from the glial cells, constituting about 14.9% of all brain tumors diagnosed in the United States between 2004 and 2014.¹ Though the incidence rate of glioblastomas has been relatively stable in developed countries over the past decade, developing countries are experiencing an increasing trend.^1,2

Though maximal surgical resection is the primary treatment of choice for glioblastomas and has prognostic value, the infiltrating nature of these tumors and proximity to vital structures often precludes complete surgical resection, resulting in almost universal recurrences and making adjuvant radiotherapy indispensable for achieving local control and overall survival (OS).³ Unfortunately, even the addition of concurrent and adjuvant chemotherapy has resulted in only modest improvements in survival outcomes for glioblastomas, with poor 2- and 5-year survival rates of 17.2% and 5.5%, respectively.^1,4,5 Most of the failures in these studies were observed to be local and within the irradiated volume, thus reiterating the importance of accurate target delineation in improving local control and survival in these patients.

Landmark randomized studies like the Brain Tumor Cooperative Group 80-01 have unequivocally substantiated the therapeutic benefits of partial-brain irradiation as compared with the historically practiced whole-brain irradiation for glioblastomas, eventually establishing 3-dimensional (3D) conformal radiation (3D-CRT) as the global standard of care in these patients.^6,7 However, there is an unfortunate ambiguity in achieving a uniform consensus regarding the delineation of the clinical target volume (CTV) for planning adjuvant radiotherapy for glioblastomas, as evident from the proposal and practice of different guidelines by different international bodies of repute.⁸

This ambiguity is largely attributed to 2 different schools of thought concerning the etiology of peritumoral edema, with deliberations concerning whether it is a physical response to mass effect and vascular permeability factors secreted by the gross tumor, or a pathological consequence of microscopic infiltration by malignant cells.⁹ Whereas The University of Texas MD Anderson Cancer Center (MDACC) and European Organisation for Research and Treatment of Cancer (EORTC) guidelines disregard the peritumoral edema during treatment planning, the Radiation Therapy Oncology Group (RTOG) recommends its inclusion in generating CTV margins.⁸ Since the volume of brain irradiated is often considered an accurate surrogate for delayed neurotoxicity, multiple retrospective and dosimetric studies have evaluated the feasibility of limited-margin radiotherapy.^8,10 However, there is no prospective, randomized evidence analyzing the impact of these margins on the survival of glioblastoma patients. The purposes of this study are to prospectively compare the recurrence patterns in patients treated with the RTOG protocol and a limited-margin MDACC protocol, and to analyze the impact of treatment volume on their survival and quality of life (QoL).

Material and Methods

This study was an investigator-initiated, partially blinded, phase 2, randomized, controlled trial with 2 arms, comparing the outcomes using 2 different guidelines for target delineation: RTOG (Arm A) and MDACC (Arm B). Sample sizes were estimated based on a superiority design to identify an improvement in the median OS from 14 months (with standard margins) to 16 months (with limited margins). It was calculated that 42 patients would be required to have an 80% power of detecting an improvement of 2 months in OS, and 56 patients would be required to have a 90% power. Assuming 20% attrition due to drop-outs or losses to follow-up, the required sample sizes increased to 52 and 70, for a power of 80% and 90%, respectively. Being a phase 2 study, a limited sample size of 50 was decided on so that subsequent phase 3 studies could be planned according to the results obtained. Approval was obtained from the institute ethics committee before initiation of the trial, and informed consent was obtained from all patients at the time of enrollment. Accrual was performed between July 2009 and December 2011, in accordance with the specified inclusion and exclusion criteria, and randomization was conducted using computer-generated random tables.

Inclusion Criteria

• i) Histopathologically proven primary glioblastoma

• ii) Age ≥18 years and ≤70 years

• iii) KPS scores ≥70

• iv) Willingness to consent to treatment and follow-up as per trial specifications

Exclusion Criteria

• i) Prior history of any other malignancy

• ii) Prior history of chemotherapy or radiotherapy

• iii) Uncontrolled comorbidities such as diabetes or hypertension interfering with the delivery of chemotherapy or radiotherapy or steroid as per trial protocol

Apart from history and physical examination and baseline hematological and biochemical investigations, initial evaluation included preoperative and postoperative MRI with T1 contrast and fluid-attenuated inversion recovery sequences. Contrast-enhanced (CE) treatment planning CT scan with 3 mm slice thickness was obtained using thermoplastic cast for immobilization, and coregistered with appropriate MR sequences using the Eclipse treatment planning system, version 11 (Varian Medical Systems Inc, Palo Alto, CA, USA).

Target Volumes and Dosimetric Analysis

Radiotherapy target volumes in the 2 arms are described in Table 1. Patients in both arms were treated with a total dose of 60 Gy in 30 fractions at 2 Gy per fraction, 5 days per week over 6 weeks, in 2 phases using 3D-CRT. A dose of 40 Gy was delivered in the initial phase, followed by a sequential boost of 20 Gy with volume reduction, according to the departmental protocol at the time of enrollment. All patients received concurrent and adjuvant chemotherapy with temozolomide according to established protocols, unless absolutely contraindicated or tolerated poorly.^4,5

Table 1.

Radiotherapy Target Volumes in Both Treatment Groups

Target Volumes	Arm A (RTOG)	Arm B (MD Anderson)
Initial phase	GTV = gross disease	GTV = gross disease
	CTV = GTV + peritumoral edema + 2 cm isotropic margin	CTV = GTV + 2 cm isotropic margin
Boost phase	CTV = GTV + 2.5 cm margin	CTV = GTV + 0.5 cm

Abbreviations: CTV, clinical target volume; GTV, gross tumor volume; PTV, planning target volume; RTOG, Radiation Therapy Oncology Group.

5 mm PTV margin for setup errors was used for all phases, as per institutional protocol.

Dosimetric analysis was performed by comparing the planning target volume (PTV) in each of the phases along with doses to the organs at risk and proportion of whole brain receiving prescription doses, in both treatment groups, using the t-test. SPSS version 22 was used for all statistical analyses.

Treatment-Related Toxicity

During the course of treatment, all patients were evaluated weekly (during concurrent chemoradiation) and monthly (during adjuvant chemotherapy) with a thorough clinical examination and appropriate laboratory investigations. All acute treatment-related toxicities were graded using the Common Terminology Criteria for Adverse Events, version 3.0.¹¹ The chi-squared test was used to analyze the differences in proportion of low-grade (grades 1 to 2) and high-grade (grades 3 to 4) toxicities in both groups.

Patterns of Failure and Survival Analysis

After treatment completion, response assessment was conducted at 3 months using CE MRI of the brain, and repeated at 6-month intervals or suspicion of clinical progression. MR spectroscopy was used to differentiate pseudoprogression in all cases with suspected radiological progression.¹²

In patients with evaluable recurrences, appropriate MR sequences were coregistered with the initial planning images, and recurrent tumor volumes (RTVs) were delineated for classification in relation to the prescription isodose volumes of the treatment plan as defined in the literature.¹³ Accordingly, recurrences were classified as central (>95% of RTV inside 95% isodose volume), infield (>95% of RTV between 95% and 80% isodose volumes), marginal (>95% of RTV between 80% and 20% isodose volumes), and distant (>95% of RTV beyond 20% isodose volume).

Patterns of failure were compared between the treatment groups using the chi-squared test. Progression-free survival (PFS) and OS were analyzed using Kaplan–Meier methods and the log-rank test. Regression analysis was used to identify various factors correlating with survival, toxicity, and QoL.

QoL Analysis

QoL assessment was performed before the start of treatment by a blinded observer, using the EORTC Quality of Life questionnaire (QLQ C30) and Brain Cancer Module (QLQ-BN20), which have been validated as efficient tools in various international trials.^14,15 This evaluation was repeated 1 month after completion of radiation, and at 3-month intervals thereafter for 9 months. A general linear model was used to estimate the differences in various domains of QoL, such as global health status, functional scores, and symptom scales in the QLQ C30 and QLQ-BN20 questionnaires at various points during the course of treatment and follow-up.

Results

Demographics characteristics of all patients are presented in Table 2, and no significant differences were observed between the arms on 2-way ANOVA. About 36% of patients did not receive concurrent chemotherapy, largely because of poor affordability (30%) and poor tolerance (6%). However, the distribution of these patients was uniform among both groups.

Table 2.

Demographic Characteristics of Patients in Both Treatment Groups

	Arm A (RTOG)	Arm B (MD Anderson)	P Value
Number	25	25	>.05
Sex (M:F)	15:10	16:9
Age, mean (range)	52 (18-70)	48 (20-68)
KPS, median (range)	70 (70-100)	70 (70-100)
Extent of Surgery
GTR	11 (44%)	14 (56%)	.55
NTR	07 (28%)	07 (28%)
STR	07(28%)	04 (16%)
Concurrent chemotherapy	15 (60%)	17 (68%)	.55
Adjuvant chemotherapy	13 (52%)	14 (56%)	.77

Abbreviations: F, female; GTR, gross total resection; KPS, Karnofsky’s Performance Score; M, male; NTR, near total resection; RTOG, Radiation Therapy Oncology Group; STR, subtotal resection.

In Arm A, the temporal lobe was the most common site (n = 8 patients) followed by the frontal lobe (n = 4 patients) and parietal lobe (n = 4 patients); in Arm B, the frontal lobe was the most common site (n = 8 patients) followed by the temporal lobe (n = 5 patients) and parietal lobe (n = 3 patients). Dosimetric and volumetric data are presented in Table 3. There was no significant difference in the mean PTV among the groups in the initial phase (P = .24), whereas a significant reduction in mean PTV was observed in the boost phase with the MDACC protocol compared with the RTOG protocol (P = .001). Though the mean doses to the organs at risk (brainstem, optic apparatus, temporal lobes) were lesser in Arm B, the difference did not reach statistical significance.

Table 3.

Dosimetric and Volumetric Analysis of Patients in Both Treatment Groups

		Arm A (Mean ± 2 SD)	Arm B (Mean ± 2 SD)	P Value
PTV (Initial Phase)		539.20 ± 142.85	593.81 ± 184.97	.249
PTV (Boost Phase)		436.10 ± 126.19	246.92 ± 116.02	.001
Brain	V 60 Gy	356.79 ± 137.57	255.47 ± 141.06	.013
	V 40 Gy	806.06 ± 218.99	764.76 ± 238.31	.526
Ipsilateral temporal lobe	V 60 Gy	47.29 ± 39.00	32.34 ± 30.31	.137
	V 54 Gy	57.43 ± 42.60	43.78 ± 33.70	.215
Contralateral temporal lobe	V 60 Gy	0.69 ± 1.24	0.73 ± 2.41	.936
	V 54 Gy	9.35 ± 8.87	7.74 ± 13.63	.622
Brainstem	V 60 Gy	5.92 ± 8.84	2.47 ± 5.18	.098
	V 54 Gy	13.37 ± 11.29	8.09 ± 8.44	.067
Optic apparatus	D max	55.6 ± 3.2	54.7 ± 2.1	.622

Abbreviation: Dmax, dose maximum; PTV, planning target volume; SD, standard deviation; V, volume.

Statistically significant difference (in bold) in boost phase PTVs and volume of brain receiving 60 Gy between the arms.

The absolute volume of whole brain receiving 60 Gy was significantly less in Arm B as compared with Arm A (P = .013). Since the volume of whole brain is an independent variable, the relative percentage of the whole brain receiving 95% of the prescription dose (57 Gy) was evaluated, and a significant difference was observed between Arm A (mean proportion, 40.21%; SD, 11.67) and Arm B (mean proportion, 30.41%; SD 12.05) with a P value less than .005. However, no statistically significant differences were observed in treatment-related acute toxicities and steroid requirements between both arms (P > .05).

About 36% of patients in each group succumbed to progression of their illness at home, and radiological classification of their recurrence patterns could not be performed. Hence, they were excluded from the patterns of failure analysis, and the results of evaluable patients in both groups are presented in Table 4. Central recurrences were the most common in both groups, and no significant differences were observed in the proportions of recurrences on chi-squared test.

Table 4.

Patterns of Recurrence in Evaluable Patients in Both Treatment Groups

	Arm A	Arm B	P Value
Central	12 (75%)	11 (68.75%)	P = .81
Infield	2 (12.5%)	3 (18.5%)
Marginal	2 (12.5%)	1 (6.25%)
Distant	0	1 (6.25%)

Recurrences were managed with salvage surgery, reirradiation, chemotherapy, or best supportive care, according to the performance status of the patient at the time of recurrence, and no significant differences were observed between the modalities used among both groups on chi-squared test (P > .05).

Kaplan–Meier survival curves for PFS and OS are presented in Fig. 1. A statistically significant difference was observed between both groups in PFS (6.1 months vs 8.7 months) as well as OS (12 months vs 17 months), with P values of .043 and .015, respectively, on log-rank test.

Fig. 1.

Kaplan–Meier Survival Curves for Progression-Free Survival (PFS) and Overall Survival (OS)RTOG indicates Radiation Therapy Oncology Group.

Multivariate regression analysis was performed to identify the factors correlating with PFS and OS. Age of the patient, extent of initial surgery, and percentage of whole brain receiving 57 Gy were found to be significant independent factors correlating with PFS and OS, as presented in Table 5.

Table 5.

Factors Associated With Progression-Free Survival (PFS) and Overall Survival (OS)

	PFS		OS
Factor	Standardized Coefficient	P Value	Standardized coefficient	P Value
Age	–0.423	.0001	–0.462	.0001
Extent of resection	0.295	.008	0.283	.007
Percentage of whole brain receiving 57 Gy	–0.434	.0001	–0.484	.0001

Abbreviation: OS, Overall Survival; PFS, Progression Free Survival.

QoL analysis is presented in Fig. 2, A and B. Statistically significant differences favoring Arm B were observed in various QLQ C30 domains such as global (P = .008), physical (P = .005), role functional (P = .009), emotional (P = .006), and social (P = .003), but not in the cognitive domain (P = .393). However, overall QoL assessed with the QLQ C30 as well as the site-specific QLQ-BN20 module was significantly better in Arm B compared with Arm A, with P values of .007 and .005, respectively.

Fig. 2.

A and B, Quality of Life Domain Analysis Using A General Linear Model

BN20 indicates European Organisation for Research and Treatment of Cancer brain cancer module; M, months; RT, radiation therapy; RTOG, Radiation Therapy Oncology Group.

Discussion

Beginning with the seminal work of Hochberg and Pruitt, multiple studies have demonstrated that local recurrences were the predominant pattern of failure after treatment of glioblastomas, and that most local recurrences occur within 2 cm to 3 cm of the primary tumor.^16–19 Further substantiated by multiple prospective trials, conformal partial-brain irradiation has replaced whole-brain radiotherapy as the standard of care for adjuvant treatment of glioblastomas.^7,20,21

Although treatment protocols for management of glioblastoma have undergone considerable transformation in the past few decades with evolution of surgical and radiation techniques and newer chemotherapeutic agents, the prognosis for these patients still remains dismal, with reported median survival in the range of 9 to 12 months and 2-year survival in the range of 8% to 12%.^4,5

Furthermore, the observation that nearly 90% of all recurrences occurred within the treatment fields in spite of dose escalation to 70 Gy to 90 Gy reiterated the radioresistance of glioblastomas and insinuated the futility of giving larger margins around the primary tumor.^22,23 Simultaneously, the impact of larger volumes of irradiation on performance status, QoL, and delayed toxicity was realized.^8,10,24

Though these studies answered important questions, they simultaneously generated new clinically relevant conundrums regarding the optimal volume to be irradiated for maintaining adequate local control while reducing treatment-related toxicity. A large part of this debate could be attributed to the etiology of peritumoral edema. Halperin and colleagues have analyzed postmortem topography of recurrent glioblastomas and observed that a 3 cm margin around the preoperative tumor and the peritumoral edema would be necessary to encompass all the tumor cells during radiation planning.²⁵ Other studies by Kelly et al and Lu and colleagues have demonstrated that infiltrating tumor cells may have a considerable contribution to peritumoral edema apart from the vasogenic component.^26,27 Whereas some studies reported the prognostic significance of peritumoral edema, its unreliability as a prognostic factor because of steroid-induced variations was highlighted by others.^28–31

This disparity in the interpretation of these studies led to an unfortunate lack of consensus regarding the optimal margins for irradiation. It is aptly reflected in the results of an audit among radiation oncologists of Canada, in which it was observed that 54% of responders follow institute-specific protocols for target delineation, whereas published guidelines by the EORTC and RTOG were followed by a dismal 14% and 32%, respectively.³²

EORTC and MDACC guidelines differ from RTOG guidelines in their approach toward inclusion of peritumoral edema in the treatment volume, whereas RTOG and MDACC guidelines differ from EORTC in the concept of boost phase.⁸ Though there are limited retrospective and prospective studies comparing RTOG and EORTC protocols, there have been no prospective studies comparing MDACC and RTOG protocols, to the best of our knowledge. In their retrospective dosimetric reviews, Chang and colleagues and Minnitti et al observed no significant differences in patterns of recurrence between RTOG and EORTC delineation protocols.^10,33 Similar outcomes were reported in 2 prospective multicentric trials in which both EORTC and RTOG protocols were allowed for radiation planning.^34,35

In our study comparing RTOG and MDACC protocols, the patterns of recurrence among both groups were statistically similar and comparable to those reported in the literature. In their retrospective study of patients treated with the EORTC protocol, Sherriff et al reported that 77% of all recurrences were central, though their definition for classification of recurrences was slightly different from that used in our study.³⁶ Using the same definitions as in our study, Ogura and colleagues observed central, infield, marginal, and distant recurrences in 66.7%, 19%, 0%, and 9.5% of patients, respectively.³⁷ However, it should be noted that the definitions used for classifying recurrences are based on prescription isodose lines rather than their distance from the initial tumor location, and hence could be subject to potential confounding when target volumes for irradiation are different. This might result in a spuriously higher proportion of central recurrences when larger margins are given around the tumor (RTOG) as compared with smaller margins (MDACC) as has been observed in our study, albeit without statistical significance.

Though recurrence patterns were similar in both groups, we observed a statistically significant improvement in the PFS and OS with the MDACC protocol compared with the RTOG protocol. As reported in our regression analysis, this improvement in survival outcomes may largely be attributed to the difference in the percentage of whole brain irradiated at the prescription dose. Our findings are in concordance with the study by Gebhardt et al, in which margins smaller than RTOG and EORTC were used according to Adult Brain Tumor Consortium guidelines. They reported a median PFS of 8 months, which is comparable to that of the MDACC arm (8.7 months) in our study.³⁸ However, similar improvement in survival outcomes was not observed in the studies comparing RTOG and EORTC protocols.^10,33–35 A possible explanation for this could be that the benefit of volume reduction in the EORTC protocol might have been offset by the absence of a boost phase, thereby increasing the proportion of whole brain receiving the prescription dose in those patients.

Apart from improvements in the PFS and OS, a significantly better QoL was observed in the MDACC arm of our study compared with the RTOG arm. This difference was consistent throughout the period of evaluation, and substantiates the importance of treatment volume reduction in reducing treatment-related late toxicity.

Notwithstanding a phase 2 design, our study is limited by a relatively smaller sample size and lack of information regarding molecular and genetic markers that carry important prognostic value. Moreover, the dose schedule followed (40 Gy + 20 Gy) in our study is different from that of current standard practice (46 Gy + 14 Gy) and might have influenced the outcomes in our analysis.

Conclusion

Disregarding peritumoral edema during target delineation does not influence the patterns of failure in glioblastoma. Reducing the volume of irradiation by following the limited-margin MDACC protocol has the potential to improve survival outcomes apart from QoL of glioblastoma patients, as compared with the RTOG protocol.

Funding

This research received no specific grant from any funding agency in the public, commercial, or not-for-profit sectors.

Conflict of interest statement. None declared.