Comparison of radiation regimens in the treatment of Glioblastoma multiforme: results from a single institution

After obtaining GEN-Research Ethics Board ethics board approval, we compiled a list of all adult patients who underwent surgery for histologically-confirmed GBM between January 2005 and December 2012. Patient information, tumor characteristics including MGMT status, treatment details, and outcomes were obtained using computerized hospital charts and patient registry. Of the 475 patients identified, a total of 342 patients received adjuvant RT, with or without concomitant TMZ. From this group, 66 patients were excluded from the final analysis for not having completed a curative RT treatment to a dose of at least 40 Gy, having a history of prior cranial RT, and/or having infra-tentorial GBM lesions.

Patients were immobilized with a thermoplastic mask for simulation and treatment. They underwent CT-based planning with 3 mm slices. Co-registration of the post-operative MRI was performed, when imaging was available. Enhancing residual disease, if present, as well as the surgical cavity were contoured. A margin of 2 cm was added with appropriate cropping to respect anatomical boundaries. For patients who received the hypofractionated regimen, the treatment planning included two planning target volumes (PTV): (i) PTV40 was defined by adding a 1.5 cm margin on the GTV/cavity and it received 40 Gy in 20 fractions; (ii) PTV60 was a concomitant boost volume defined by adding a 0.5 cm margin on the GTV/residual and it received a total dose 60 Gy in 20 fractions. Radiation treatment was planned using either forward or inverse IMRT. In most cases, the PTV was covered by the 95% isodose. Organs at risk including the brainstem, optic chiasm, and optic nerves were contoured. Dose to the brainstem and optic chiasm was limited to less than 54 Gy for conventional fractionation. Dose constraints for the hypofractionated regimen were specified as less than 40 Gy in 20 fractions to the brainstem, chiasm, and at least one retina.

The following fractionation schedules were used: conventional RT with 60 Gy in 30 fractions (ConvRT); hypofractionated RT with 60 Gy in 20 fractions (HF60); hypofractionated RT with 40 Gy in 15 fractions (HF40). All the regimens were delivered 5 days per week. The HF60 regimen excluded patients with T1 gadolinium enhancement present at less than 1.5 cm from the brainstem and/or optic apparatus. Furthermore, patients over the age of 70 or with KPS <70 usually received the HF40 regimen, as per current treatment guidelines [17].

The majority of patients received concomitant TMZ at a dose of 75 mg/m2 daily during radiation treatment. This was followed by adjuvant TMZ with a dose 150 mg/m2 daily 5 days every 28 days, as per the NCIC regimen [1]. At the time of progression, patients were assessed for second line chemotherapy, repeat surgery, and/or re-irradiation.

Patients were assessed weekly during RT with physical examination, and bloods tests including complete blood counts, serum electrolytes as well as renal and liver function. After completion of RT, patients were assessed on follow-up visit every three months with clinical evaluation and MRI. Evaluation of tumor response was determined based on imaging reports and review of the MRI T1 enhanced and/or T2 images.

The date of recurrence was determined from the time at which either progressive residual enhancement was observed or new areas of disease appeared on imaging. Progression free survival (PFS) was defined from the date of diagnosis until the documentation of disease progression clinically and/or radiologically. Overall survival (OS) was measured from date of diagnosis until date of death, or date of last clinical follow-up visit.

Pseudoprogression was defined as radiological progression within three to six months of RT completion, followed by stability or regression of the lesion without the patient receiving treatment for recurrence. Patients who had repeat surgery for presumed recurrence who did not show evidence of persistent malignancy on pathology were also deemed to have pseudoprogression. The pattern of failure was deemed to be central if in continuity with the initial tumor cavity, and distal if more than 2 cm from the cavity.

The primary endpoint of this retrospective study was OS and the secondary endpoint was PFS. Statistical analysis was carried out using SAS version 9.3 (SAS Institute, Cary, NC, USA). Descriptive statistics were used to summarize the study population. Univariate and multivariate Cox proportional hazard regression models were used to assess the impact of different treatment regimens on OS and PFS. The index date was date of death for analyses of OS, and date of first progression for PFS. The effect of treatment regimen on survival was quantified by means of hazard ratios (HRs) with 95% confidence intervals (95% CIs). Analyses were adjusted for important prognostic factors for survival, such as KPS (dichotomous: ≥ 70 versus 70), age (dichotomous: ≥ 65 versus <65), extent of initial surgery performed (biopsy versus subtotal resection or gross total resection), having any chemotherapy before the index date (dichotomous), having repeated surgery before the index date and methylation status of MGMT (methylated, unmethylated and unknown). In order to investigate artificial differences in survival caused by treatment based on age and KPS status, we conducted sensitivity analyses of multivariate models, testing for the multiplicative interaction terms between age and treatment group and between KPS status and treatment group. Analyses were independently repeated for each treatment group. We assessed the assumption of proportional hazard by examining graphs of scaled Schoenfeld residuals. Statistical significance of Kaplan-Meyer curves was assessed by the log-rank test. All analyses were two-sided with p ≤ 0.05 being considered significant.

A total of 276 patients with histologically-proven GBM who received adjuvant RT, with or without concomitant TMZ, were included in this population-based study. Overall median follow-up time was 13.2 months (range 1.4 to 105.7 months). Patient characteristics are summarized in Table 1. One hundred and forty-seven patients received ConvRT, 86 patients received hypofractionated RT as per the HF60, and 43 patients as per the HF40 regimen. Two hundred and two patients were found to have tumor progression on imaging. The median survival for the whole population was 13.7 months with a median PFS of 8.8 months.

The similarities in patient characteristics between the ConvRT and HF60 groups are in contrast to that of patients in the HF40 group. Patients in the ConvRT and HF60 groups were more likely to have gross tumor resection (GTR), to have had repeat surgery at the time of recurrence, and to have received chemotherapy at some point during their treatment. Patients in the HF40 group were older in age, with a median age of 72, and had a more limited performance status, with close to half of these patients having a KPS of less than 70.

Median survival was 16 months in the ConvRT group and 15 months in the HF60 group (P = .3487, Figure 1a). Survival in the HF40 arm was significantly lower than ConvRT with a median survival of 8 months (P < .00001). This difference in survival was also sustained at 1 and 2 years after diagnosis, with a 2-year OS of 23.1% and 19.7% (P = .347) in the ConvRT and HF60 groups, respectively. Compared to these results, the OS was extremely poor in the HF40 group, with only 2.32% of patients still alive at 2 years. Our data also showed that for patients 65 year and older, the median survival was 10.0 months in the ConvRT group, compared to 9.13 (P = .357) for the HF60 patients, and 7.6 months for the HF40 patients (P = .0049). We did not find any indication in our cohort of patients of an interaction between age and treatment status (Wald Chi-squared test = 0.0261; p = 0.8717). We detected a borderline significant interaction between KPS score and treatment regimen group (Wald Chi-squared test = 5.8949; p = 0.0525).

These trends were also observed in terms of PFS (Figure 1b). Median PFS was close to 9 months in both the ConvRT and the HF60 groups, compared to 5.4 months in the HF40 groups (P = .0002). The PFS at 1 year was 37.4% in the ConvRT group, 31.4% in the HF60 group (P = .6894) and 7.0% in the HF40 group (P = .0007).

Multivariate analysis showed that treatment regimen (ConvRT and HF60), methylation status, use of chemotherapy, extent of resection and repeat surgery at the time of recurrence were the most significant independent prognostic factors for survival (Table 2). There was no significant difference in treatment outcome when HF60 was compared to ConvRT (HR: 1.27, 95% CI, 0.93-1.74). This finding is in contrast to the HF40 treatment group, which showed significantly worse outcome (HR 2.22; 95% CI, 1.30 to 3.80).

Extent of resection was also found to significantly affect outcome. At the time of initial surgery, both STR (HR 0.62; CI, 0.39-0.99; P = .048) and GTR (HR 0.39; CI, 0.22 to 0.67; P = .001) were found to be superior to biopsy alone. The effect of repeat debulking surgery at the time of recurrence was also found to correlate with survival (HR 0.62; CI, 0.44 to 0.86; P = .005). Upon looking at factors affecting individual treatment regimens, we found that in both the ConvRT and HF60 groups, methylation status, and extent of resection were the most significant prognostic factors affecting OS (Table 3). In addition, in the HF60 group, age, use of chemotherapy and repeat surgery also correlated with better outcome. In comparison to ConvRT, only performance status significantly affected survival in the HF40 group.

Pseudoprogression was found to develop at a median time of 3.8 months in 10.8% of patients. Patients who developed pseudoprogression had a median survival of 25.16 months with a 2- year OS of 46.6%, vs. a median survival of 13.4 months and a 2-year OS of 16.6% for those who did not (P = .002). Pseudoprogression did not show an association with either the MGMT methylation status of the tumor (P = .4506) or the RT regimen received (P = .70).

Methylation of MGMT promoter was found as an independent prognostic factor for OS. On Kaplan-Meier analysis, median survival in the ConvRT group of patients with MGMT promoter methylation was 21 months, vs. 13.4 months for unmethylated tumors (P = .001; Figure 2). Similar results were seen in the HF60 group, with MS of 20.6 months for methylated tumors, vs. 13.6 months for unmethylated tumors (P = .0325). In the HF40 group, patients with methylated tumour had a median survival of 10.2 months, compared to 7.9 months for unmethylated tumours. The outcome of patients with methylated (P = .674) or unmethylated MGMT (P = .891) was not affected by the RT regimen being used.

The pattern of recurrence was evaluated and correlated with the MGMT methylation status. We found that distal recurrences were more frequently associated with MGMT methylated tumors (P = .020). We have also identified a trend for central recurrence to be more often linked to unmethylated tumors (P = .070). However, both methylated and unmethylated tumors were more likely to recur centrally than to recur distally (P < .0001).