Introduction

While the primary multidisciplinary treatment of glioblastoma multiforme (GBM) is well established through randomized trials [1, 2], the optimal therapeutic approach for recurrent disease remains a controversial issue in neuro-oncology with no standard treatment recommended at present. Owing to the infiltrative nature of GBM [3], almost all patients experience local failure after surgery and postoperative chemoradiotherapy with macroscopic relapses manifested mainly adjacent to or within the surgical tumor bed [4]. Despite improvements in neurosurgical techniques, advances in radiation oncology and the introduction of new systemic agents, the prognosis of GBM remains dismal [5].

In the clinical setting of recurrent GBM, re-irradiation has been shown to be a valuable treatment option with the potential for palliation and survival prolongation in a subgroup of patients [6, 7]. However, repeat radiotherapy (RT) is hampered by the intrinsic radiosensitivity of pretreated normal brain tissue and the risk of increased toxicity [8]. Several RT techniques have been implemented in clinical practice including stereotactic radiosurgery (SRS; [9]), fractionated stereotactic radiotherapy (FSRT; [6]) as well as interstitial brachytherapy (BRT; [10]). Interstitial high-dose-rate (HDR) BRT represents a high-precision RT technique [11] which enables the safe treatment of larger gliomas [12, 13] compared to SRS and FSRT [5, 13, 14].

In the present single-institute analysis we report our experience with computed tomography (CT)-guided iridium-192 interstitial HDR BRT for the re-irradiation of inoperable recurrent GBM. To the best of our knowledge, this represents the largest study on HDR BRT for recurrent GBM.

Patients and methods

From 1995–2014, a total of 135 patients underwent HDR BRT for recurrent primary GBM. All patients had shown disease progression after initial treatment consisting of surgery (gross total or subtotal resection) and external beam radiotherapy (EBRT) with chemotherapy. Since April 2005 all patients had received postoperative chemoradiotherapy and adjuvant chemotherapy with temozolomide. Patients with radiologically confirmed disease progression were offered treatment with interstitial HDR BRT, if they were medically or technically inoperable or refused repeat surgery. Clinical judgment was used to define eligibility for BRT with patients fulfilling the following criteria: Karnofsky performance score (KPS) ≥50, unilateral tumor growth not involving the ventricles or corpus callosum as well as maximal tumor diameter ≤10 cm. Each patient was required to provide written consent. The study was approved by the local research ethics board. Epidemiological and clinical characteristics are summarized in Table 1.

Table 1 Patient, tumor, and treatment characteristics
Full size table

Treatment technique

Our BRT technique has been described in detail elsewhere [10]. In short, catheter implantation was performed with neurosurgical assistance using CT-guidance under local anesthesia and sedoanalgesia along with intravenous peri-interventional anticonvulsive/anti-edematous prophylaxis. Implantation technique was transcranial insertion utilizing a acrylic template sutured to the scalp without requirement of a stereotactic frame (Fig. 1a). Positional control of the catheters was obtained by generating contrast-enhanced CT images with the catheters in situ under registration with a preinterventional contrast-enhanced magnetic resonance imaging (MRI). Thus, maximum insertion depth, direction and position of the catheters were estimated by interactive CT scanning (Fig. 1b). Treatment planning with anatomy-based three-dimensional (3D) dose optimization was conducted initially by Plato BPS (Nucletron, Veenendaal, The Netherlands) followed by Oncentra Brachy (ELEKTA, Veenendaal, The Netherlands). Gross tumor volume (GTV) was defined as the gadolinium-enhanced lesion in T1-weighted MRI images without the addition of further margins, thus, planning target volume (PTV) equaled GTV. PTV coverage was defined as the proportion of the PTV receiving at least the prescription dose defined as the average dose value on the PTV surface, representing the 100% isodose (Fig. 1c). Our HDR protocol evolved over time under consideration of available radio-oncological knowledge. Initially, based on the experiences by Prados et al. [15] and Gutin et al. [16], the total physical HDR dose was escalated in 10 Gy increments from 30 to 40 Gy and finally 50 Gy by fractional doses of 5.0 Gy twice a day. However, an interim analysis showing no additional survival benefit with increased incidence of radionecrosis led to dose de-escalation and implementation of the current protocol consisting of 40 Gy delivered over four consecutive days in twice-daily fractions of 5.0 Gy with an interfractional interval of at least 6 h which equals a biological effective dose of 65 Gy (a/β = 8 Gy; [17]). Catheters were removed after the last treatment fraction. All irradiations were performed using an iridium-192 HDR-afterloading system (micro Selectron-HDR, Nucletron). Written informed consent was obtained from all patients.

Fig. 1
figure 1

Interstitial high-dose-rate brachytherapy implant for a recurrent glioblastoma multiforme of the right temporal lobe. a Macroscopic template view with 13 implanted brachytherapy catheters. The template is sutured at the capillitium of the right temporal region. b Multiplanar three-dimensional view of the planning computed tomography data set of the same implant coregistered with a pretreatment magnetic resonance imaging data set. The brachytherapy catheters are identifiable in situ. c Sagittal and axial view of computed tomography/magnetic resonance imaging coregistration of the same implant with overlaid isodose distribution. The volumetrically calculated lesion size amounted 109 cm3. The isodose lines color code convention is: rose = 300% {isodose = 15.0 Gy}; yellow = 200% {isodose = 10.0 Gy}; green = 100% {isodose = 5.0 Gy}; light blue = 50% {isodose = 2.5 Gy}

Full size image

Follow-up

Follow-up consisted of clinical and radiological evaluation (contrast-enhanced CT until 2001 and contrast-enhanced MRI ever since) performed initially at 6 weeks after BRT and every 3 months thereafter. Patient data were collected from a prospectively maintained database and by retrospective clinical chart review with data collection allowing also information acquisition from referring neurosurgeons and radiation oncologists. Tumor response was initially assessed according to the Macdonald criteria [18] and after 2010 according to the “Response Assessment in Neuro-oncology” (RANO) criteria [19]. Overall survival (OS) was calculated from primary diagnosis, and survival after BRT was calculated from initiation of BRT. Progression-free survival after BRT (PFS) was defined as the time from initiation of BRT until tumor progression or death (by any cause), whichever occurred first.

Statistics

Primary outcome measures considered for analysis were OS, survival after BRT and treatment-related toxicity. Univariate log-rank test and Cox regression models were used to identify a possible correlation with following factors: extent of resection, BRT dose, median time from EBRT to BRT, median tumor volume, median age, KPS at BRT and chemotherapy after BRT. Furthermore, we investigated how survival after BRT correlated with the revised recursive partitioning analysis (RPA) classification for GBM [20]. The estimated likelihood of events was calculated using the Kaplan–Meier method. A two-sided P value ≤0.05 was considered statistically significant. Statistical analysis was performed using the WinStat® software (R. Fitch Software, Bad Krozingen, Germany).

Results

Oncological outcomes

After a median follow-up of 9.2 months, median OS after BRT and after primary diagnosis was 9.2 and 20.5 months, respectively (Fig. 2), whereas median PFS after BRT was 4.6 months with a PFS rate of 47% at 6 months and 13.5% at 12 months. Of the prognostic factors evaluated in the univariate analysis extent of surgery at diagnosis (gross total resection vs. subtotal resection), median tumor volume at recurrence and median time from EBRT to BRT reached statistical significance (Table 2), retained also in multivariate analysis (Table 3). According to the revised RPA classification for GBM the corresponding median OS after BRT for the classes I–III, IV and V–VI was 9.6, 9.4 and 8.9 months, respectively (p = 0.07; Fig. 3).

Table 2 Univariate results for survival after BRT
Full size table
Table 3 Multivariate results for survival after BRT
Full size table
Fig. 2
figure 2

Kaplan–Meier survival curves: a survival after brachytherapy (BRT), b survival after primary diagnosis

Full size image
Fig. 3
figure 3

Survival after brachytherapy (BRT) according to recursive partitioning analysis (RPA) classification (I–III, IV, V–VI)

Full size image

Toxicity

Treatment was generally well tolerated. Eight out of 135 patients (5.9%) developed moderate to severe complications. Among those, 4 (2.9%) developed intracerebral bleeding after catheter implantation or explanation and one (0.7%) of them died due to massive hemorrhage. Another 3 (2.2%) suffered the consequences of symptomatic focal radionecrosis which was diagnosed with a mean latent interval of 3.5 months (2–5 months) and were treated surgically. One (0.7%) patient experienced an epileptic grand mal episode which responded adequately to anticonvulsive treatment. The median KPS of the entire patient population was 80 (50–100) at the time of BRT. The corresponding median KPS 6 weeks, 3 and 6 months after BRT was 80 (50–100), 80 (50–100), and 70 (50–100), respectively, suggesting no severe deterioration in the first 6 months following BRT. All patients received during hospitalization anticonvulsive and corticosteroid based antiedematous treatment. With the exception of the 3 patients experiencing symptomatic radionecrosis, antiedematous treatment could be tapered off within 3 weeks after treatment.

Discussion

Despite maximal surgical excision and subsequent chemoradiotherapy, GBM recurs locally nearly without exception with high rate lethality [21]. Nowadays, several therapeutic approaches are available for the treatment of recurrent GBM, including re-operation, re-irradiation and systemic agents.

Contrary to first-line treatment where a significant contribution of radical resection on OS is proven [22], the role of salvage surgery and its indications for recurrent GBM [23] still remain unclear. In the absence of randomized trials, however, several studies report a positive impact on OS for repeated resection of recurrent disease [24, 25]. Montemurro et al. [26] could show in a systematic review comprising 2279 patients from 28 studies an improved OS after repeat surgery of median 9.7 months. However, there are several series not corroborating these data [27, 28], thus, indicating no survival benefit. Considering that mortality and morbidity rates range between 0–5% and 2.1–33% [29], respectively, any decision for repeat neurosurgical treatment is being made individually based on a number of prognostic factors among which performance status and tumor size seem to be the most crucial [30].

During the natural course of disease, most patients with recurrent GBM will be also exposed to some kind of systemic therapy. Until now, various antineoplastic regimens, either as monotherapy or in combination, have been tested in several phase I–III trials generating a median survival after systemic therapy in the order of 3.5–11.0 months [31]. After its introduction, temozolomide became the most widely used chemotherapy [32] with nitrosoureas remaining a good alternative [33] and bevacizumab gaining growing clinical acceptance in recent years [34].

Historically, re-irradiation for recurrent malignant gliomas was prescribed reluctantly due to the increased risk of excessive morbidity. With the implementation of modern treatment techniques, however, repeat irradiation has become well-tolerated with improved therapeutic ratio. Stereotactic radiosurgery, FSRT as well as BRT, even though sharing different physical characteristics, have been all used to treat patients with recurrent gliomas.

Selected studies on SRS by Redmond et al. [14], Amelio et al. [13] and Niyazi et al. [5] could show an inferred median survival ranging from 5.3 to 13.0 months for recurrent lesion volumes of median 6.2 to 28.0 cm3 and associated radionecrosis as well as re-operation rates of 0–31.3% and 2.0–22.0%, respectively. Similarly, several authors have published their experience on FSRT for the treatment of recurrent GBM [6, 35, 36] reporting a median survival in the range of 6.5–11 months for tumor volumes of median 7–50 cm3. Overall, FSRT is reported to be equally well tolerated as SRS with minor moderate toxicity and only one case of symptomatic radionecrosis in the largest published series [6, 35, 36].

Data regarding the value of BRT in the treatment of recurrent GBM are to some extent inconclusive [7, 10, 37, 38]. Experience on implantation of iodine-125 comprise the largest body of the literature, either as low-dose-rate or HDR techniques, reporting a median survival after BRT ranging from 7.6 to 17.25 months for median lesion volumes from 12.5 to 51 cm3 [12, 13]. In most of those studies patients underwent BRT after maximal recurrent tumor debulking with radionecrosis observed in 0–23% of cases. Further consideration of the dose–rate application shows that HDR BRT yields the best local tumor control, however with increased risk for symptomatic radionecrosis. Archavlis et al. [7] compared the results of 111 patients with recurrent GBM who were treated either by repeated surgery versus (vs.) sole interstitial HDR BRT vs. palliative chemotherapy with temozolomide. High-dose-rate BRT was delivered in eight fractions of 5.0 Gy twice-daily, whereas temozolomide was administered as a “one week on, one week of” scheme with a daily dose of 150 mg/m2. Patients after re-operation or BRT received temozolomide as adjunctive treatment. With a mean tumor volume of 51 cm3 for BRT vs. 43 cm3 for re-operation vs. 45 cm3 in the temozolomide group, survival was 37, 30, and 24 weeks, respectively (p < 0.05).

The present study expands our experience [10] with interstitial HDR irradiation for unresectable recurrent GBM reporting oncological outcomes comparable to the existing literature on hypofractionated, high-precision re-irradiation. Fogh et al. [6] reported on 147 patients with high-grade glioma recurrences, among which 105 with GBM, treated with FSRT. Median tumor volume was 22 cm3 with FSRT delivering daily fractions of 3.5 Gy up to a total median dose of 35 Gy. Median survival after treatment was 11 months for patients with primary GBM. Of note, 84 (57%) patients underwent repeated surgery prior to re-irradiation, whereas 48 (32.6%) received concurrent chemotherapy using various agents. Combs et al. [35] treated a total of 172 patients with recurrent low- and high-grade gliomas. Of those, 59 were diagnosed with primary GBM. Fractionated stereotactic RT was performed in 2 Gy daily fractions up to a total dose of median 36 Gy. Median survival for GBM after re-irradiation was 8 months. In our current series, median OS after BRT was 9.2 months for median recurrent lesion volumes of 42 cm3. In this context, stereotactic RT is a non-invasive treatment modality with proven efficacy in the treatment of small- to medium-sized tumors with image-based interstitial HDR BRT being a meaningful additional modality that can be implemented for the treatment of larger lesions or when stereotactic radiation delivery systems are not available. Here, the inherently nonhomogeneous dose distribution in HDR parallels the intrinsic capability of SRS/FSRT to perform simultaneous intratumoral dose-boosting. Compared to SRS/FSRT, however, HDR provides a higher degree of intratumoral dose heterogeneity with no upper dose limits. A recently published dosimetric study by Milickovic et al. comparing SRS with interstitial HDR BRT for recurrent GBM could show HDR achieving at least equal conformity and a steeper dose gradient at the target volume margin [11].

This study represents, to our knowledge, the largest series to examine the efficacy and tolerability of HDR BRT in the re-irradiation of unresectable recurrent GBM. To the study’s strengths belong the relatively large patient number and the uniformity with respect to technique and dose fractionation. However, the present results are not obtained from a prospective randomized study. In addition, it must be pointed out that in recent years an increasing number of publications have been communicating the impact of new systemic agents on the treatment of progressive high-grade gliomas [39, 40]. However, none of our patients was treated with such substances before or simultaneously with BRT. From this perspective, the impact of highly conformal interstitial re-irradiation by means of HDR BRT on patient survival may not be adequately evaluated. The possible synergistic effect of hypofractionated RT in combination with newer systemic treatments represents a topic of interest and is therefore subject of ongoing research.

Conclusions

We have demonstrated that interstitial HDR BRT is a safe and effective radiotherapy method for the treatment of inoperable recurrent glioblastoma in selected patients. Our results warrant a prospective evaluation and investigation in future studies in order to conclusively elucidate the role of HDR BRT in the treatment of recurrent GBM.