We prospectively collected and followed all cases of recurrent or secondary GBM treated with HFReRT after salvage surgery at our institutions between November 2011 and December 2014. All patients were presented to our institute's multidisciplinary neuro-oncology tumor conference, which included neurosurgeon, radiation oncologist, neuro-oncologist, and neuro-radiologist. Tumor recurrences were identified by the appearance of new contrast-enhanced lesions on T1-weighted magnetic resonance imaging (MRI) scans or increase in the volumes of enhanced lesions that had been recorded as treatment-related changes after initial treatment. Eventually, a diagnosis of recurrent GBM was established via histologic analysis after salvage surgery, and HFReRT was offered to selected patients who met the following inclusion criteria, based on our previous work: 1) pathologically confirmed recurrent GBM treated with salvage surgery; 2) age of 18–70 years; 3) KPS ≥60 at the time of referral for HFReRT; and 4) an interval of ≥12 months between initial RT and HFReRT.16 This study was approved by our Institutional Review Board (4-2017-0878).

Patients with recurrent or secondary GBM underwent maximal surgical resection of the enhancing lesions identified on MRI scans. HFReRT was offered to patients within 4 weeks after salvage surgery. Each patient underwent a computed tomography simulation in the supine position while immobilized with a customized thermoplastic mask device. MRI scans at recurrence and after salvage surgery for the recurrent tumor were transferred to the planning system; after image fusion, the gross tumor volume (GTV) was delineated on T1-weighted MRI scans, based on the operative cavity and residual contrast (gadolinium)-enhanced lesion plus a 0.5-cm margin. The clinical target volume (CTV) was defined as the GTV with a 1–2-cm expansion to account for peritumoral edema on T2 fluid-attenuated inversion recovery MRI sequences. HFReRT was delivered via helical tomotherapy with 6-MV photons (Tomo-Therapy, Madison, WI, USA) using a simultaneous integrated boost technique. Doses of 45 Gy in 15 fractions and 37.5 Gy in 15 fractions were delivered to the GTV and CTV, respectively.

All patients underwent MRI evaluations at 1 month after salvage treatment for recurrent GBM, and every 3 months thereafter. During follow-up, the treatment outcome and toxicity assessments included a thorough history taking, physical examination (particularly the KPS and neurological status), and analysis of all available imaging data, including contrast-enhanced brain MRI scans. Toxicity after salvage treatment was assessed according to the criteria of the Radiation Therapy Oncology Group. However, no specific tool was used to assess the patients' neurocognitive status or quality of life. The primary endpoints for analysis were overall survival (OS) and progression-free survival (PFS). OS was calculated from the date of surgery for recurrence to the date of death from any cause, and PFS was measured from the date of recurrence until the date of tumor progression or death. The following prognostic factors associated with OS were included: age, gender, KPS, initial histology, eloquent area involvement at recurrence, extent of surgery at recurrence, O6-methylguanine-DNA-methyltransferase status at recurrence, isocitrate dehydrogenase (IDH) mutation at recurrence, tumor volume at recurrence, GTV for HFReRT, and interval between RT courses. In addition, we divided patients into prognostic groups using the NIH recurrent GBM scale, which is a practical scoring system devised by the National Institutes of Health (NIH),9 and assessed OS by group. The secondary endpoints included any documented treatment-related toxicity observed during follow-up examinations and on MRI scans.

Survival rates were estimated using the Kaplan-Meier method. A univariate analysis of risk factors was performed by comparing survival rates using the log-rank test. A multivariate analysis was performed using a Cox regression model with a stepwise method to identify prognostic factors. A p value <0.05 was considered statistically significant. All analyses were performed using IBM SPSS, version 20.0 (IBM Corp., Armonk, NY, USA).

Our study population included 25 patients (11 men, 14 women) with a median age at recurrence of 45 years (range, 23–61 years) and median and minimum KPSs of 90 and 60, respectively. Among all patients, 8 patients had secondary GBM, who were initially diagnosed with WHO grade II/III glioma. The median recurrent tumor volume was 32 mL (range, 4–90 mL). All patients underwent surgery for pathologic diagnosis and maximal tumor debulking; 15 underwent gross total resection (GTR) and 10 underwent subtotal resection, defined as the removal of >90% of the contrast-enhanced tumor. All patients had received a median postoperative RT dose of 60 Gy (range, 54–70 Gy) at the time of the initial diagnosis, and the median time interval between the initial RT and re-irradiation courses was 30 months (range, 12–72 months; 29 months for primary recurrent GBM, 30 months for secondary GBM). At the time of HFReRT, the median GTV and CTV were 97 mL (range, 21–261 mL) and 270 mL (range, 101–1353 mL), respectively. Thirteen patients received sequential systemic therapy with temozolomide after HFReRT. Details of the baseline characteristics are described in Table 1.

Patients were followed up for a median of 13 months (range, 4–41 months) after a recurrence. At the time of the last observation, 5 patients (20%) remained alive at 15, 16, 22, 28, and 29 months after recurrence; the other 20 (80%) had died. The median PFS duration and 1-year PFS rate were 13 months (range, 1–41 months) and 51.4%, respectively, and the median OS duration and 1-year OS rate were 16 months (range, 7–41 months) and 60%, respectively (Fig. 1). A univariate analysis of factors related to OS after a recurrence identified the KPS (p=0.026) (Fig. 2A), eloquent area involvement at recurrence (p=0.030) (Fig. 2B), GTV at HFReRT (p=0.005) (Fig. 2C), and administration of sequential CTx after HFReRT (p=0.033) (Fig. 2D) as significant factors. In contrast, age (p=0.385), initial histology (p=0.232), the extent of surgery at recurrence (p=0.391), IDH mutation status at recurrence (p=0.626), tumor volume at recurrence (p=0.075), and interval between RT (p=0.489) did not affect OS. All results from the univariate analysis are shown in Table 2. In a multivariate stepwise Cox regression model analysis, the GTV at HFReRT (≥100 mL vs. <100 mL: hazard ratio=3.717, 95% confidence interval: 1.356–10.187, p=0.011) was confirmed as an independent prognostic factor affecting OS (Table 2).

We classified patients into prognostic groups based on the NIH recurrent GBM scale, which included the involvement of eloquent regions, KPS ≤80, and recurrent tumor volume ≥50 mL.9 Because our cohort was small, we defined the following groups: NIH recurrent GBM scale of 0 or 1 (low-risk group, n=14) and NIH recurrent GBM scale of 2 or 3 (high-risk group, n=11). The NIH recurrent GBM scale was found to be a significant prognosticator of OS (low-risk group vs. high-risk group: 15 months vs. 10 months, p=0.025) (Fig. 3).

During treatment, no severe acute toxicity events requiring the interruption of HFReRT were recorded. During the follow-up period, corticosteroids, antiepileptic drugs, and mannitol were administered in 22 (88%), 22 (88%), and 7 (28%) cases, respectively. At a median of 9 months after HFReRT, 16 patients (64%) harbored suspected areas of RN, based on gadolinium-enhanced T1-weighted MRI scans. However, these lesions could not be completely differentiated from tumor progression. Of these 16 patients, 8 were hospitalized for neurological deterioration with suspected grade 3 RN. Grade 3 RN developed in 18.8% (3/16) of cases involving an interval of >24 months between RT courses and in 55.6% (5/9) of cases with an interval of <24 months (p=0.062). No differences in the incidence of grade 3 RN according to GTV and CTV were observed.