Up until now, whole-brain radiation therapy (WBRT) has been the standard adjuvant therapy following resection of brain metastases, since it decreases the rate of local and distant recurrence in the brain [16]. WBRT with the addition of a boost to the resection cavity has been shown to increase local control (LC) rates. Since no survival benefit has been demonstrated for WBRT in addition to surgery or SRS, there is interest avoiding the potential neurocognitive sequelae associated with this treatment [12]. Recently, several retrospective reports have demonstrated that stereotactic radiosurgery (SRS) or stereotactic hypofractionated radiotherapy (SHRT) directed at the resection cavity can reduce local failure rates [19, 25]. Most patients were treated with frame-based stereotactic systems such as the Gamma Knife (Elekta, Stockholm, Sweden) or dedicated stereotactic LINAC systems [7, 10, 15]. Frameless image-guided intracranial stereotactic LINAC radiosurgery for brain metastases has recently been introduced and clinical outcomes are comparable to those after frame-based radiosurgery techniques [4, 11]. Here we report on clinical outcome and LC in patients who underwent adjuvant frameless image-guided LINAC-based SRS and SHRT after resection of brain metastases.

Materials and methods

Patients and study design

This retrospective study was approved by the local ethics committee. Between March 2009 and February 2012, 44 surgical cavities in 42 patients were treated with frameless image-guided LINAC-based SRS or SHRT. Of these patients, 35 had total gross resection of the metastasis, which was confirmed by MRI within 24 h after surgery. Patients with one or two further brain metastases were included and treated by SRS. Patients with prior WBRT were excluded. For each patient, the Radiation Therapy Oncology Group recursive partitioning analysis (RTOG RPA) classification and Graded Prognostic Assessment (GPA) scores were calculated [24]. Acute and late toxicities were evaluated using the Common Toxicity Criteria for Adverse Events (CTCAE) version 4.0 grading system. Clinical status evaluation and imaging examinations were performed at 3–6-month intervals. To assess local recurrence, new distant brain metastases and radionecrosis, all posttreatment MRIs were reviewed by a radiation oncologist, a neuroradiologist and a neurosurgeon. Radionecrosis was scored according to Late Effects in Normal Tissue—Subjective, Objective, Management and Analytic (LENT-SOMA) criteria [20].

Frameless SRS and SHRT procedures

All patients were immobilized using a thermoplastic stereotactic frameless head mask (BrainLAB, Feldkirchen, Germany). Postoperative helical CT images of 1.5-mm slice thickness were obtained and fused with postoperative T1 contrast-enhanced MPRAGE and T2-3D sequences that were not older than 2 weeks prior to irradiation. The clinical target volume (CTV) was defined as the resection cavity including the surgical defect and any contrast enhancement, plus a 2-mm margin in all directions. For definition of planning target volume (PTV), a 1-mm margin was added to the CTV in all directions. Planning was carried out using the BrainLAB® iPlan planning system versions  4.1 and 4.5 (BrainLAB). All patients were irradiated with a single isocenter. Dose–volume histograms (DVH) were calculated for the target volumes and organs at risk. We used a conformity index (CI) defined according the following formula: (1 + volume of tissue outside PTV receiving at least the prescribed dose/volume of PTV receiving at least the prescribed dose). For SRS planning we evaluated the normal brain volume irradiated with 12 Gy (V12 Gy) or 10 Gy (V10 Gy), according to previously published reports [3, 13]. For SHRT planning we evaluated the volume of normal brain irradiated with more than 4 Gy (V4 Gy) per fraction [8]. The prescribed SRS dose range was 17–18 Gy for PTVs with a volume ≤ 10 cm3. Dose was prescribed in order to cover at least 95 % of the PTV. For larger cavities, the dose concepts of SHRT were 4 × 6 Gy, 6 × 4 Gy and 10 × 4 Gy. The fractionation scheme of 4 × 6 Gy was used for smaller PTVs with a volume ≤ 20 cm3. Initially, the dose concept of 6 × 4 Gy was used for PTVs > 20 cm3, but in order to increase the equivalent dose in 2-Gy fractions (EQD2), 40 Gy was applied in 10 fractions. All treatments were delivered using the Novalis TX® LINAC (Varian Medical Systems Inc., Palo Alto, CA, USA and BrainLAB) in the 6-MV stereotactic mode. Patient setup was performed using the ExacTrac system (BrainLAB). ExacTrac delivers the patient setup error in six dimensions. In this study, the tolerances for the patient setup were 0.8 mm for the three translational axes (longitudinal, lateral and vertical) and 1.0° for the three rotational axes (pitch, roll and couch rotation).

Treatment was delivered using conformal dynamic arcs, intensity-modulated radiation therapy (IMRT) field techniques and hybrid arcs (dynamic arcs and IMRT fields).

Statistics

The analyzed endpoints were LC, distant brain control (DC) and overall survival (OS). LC was defined as the absence of new nodular contrast enhancement adjacent to the resection cavity on MRI. Local recurrence (LR) was defined as new contrast-enhancing lesions within 3 mm of the resection cavity, i.e. within the PTV. New distant brain metastases were defined as new contrast-enhancing lesions outside the PTV. All time-to-event endpoints were measured from the beginning of radiotherapy to either the last follow-up MRI (for recurrence rates), the beginning of salvage radiotherapy (for salvage therapy) or the date of death for OS. Survival rates were calculated using the Kaplan–Meier product limit methodology. Comparison of survival rates according to treatment (SRS vs. SHRT) was performed using a two-sided log-rank test. All analyses were conducted using SPSS version 17.0 (SPSS Inc., Chicago, IL, USA).

Results

Patient characteristics

Patient characteristics are shown in Tab. 1. In 7 patients (16 %) the postoperative MRI showed residual tumor after resection. At the time of irradiation, extracranial disease was present in 60 % of patients. The median time from surgery to irradiation was 40 days (range 15–73 days).

Tab. 1 Patient characteristics
Full size table

Treatment parameters

Of the 44 surgical cavities in the 42 patients in the current study, 23 lesions (52 %) were treated with SRS. The median dose prescription to the PTV margin was 17 Gy (range 16–18 Gy) with maximum and minimum median PTV doses of 17.7 Gy (range 17–20.4 Gy) and 16.8 Gy (range 14.6–18 Gy), respectively. One patient received 16 Gy to the PTV margin because the lesion was near critical structures. The median V10 Gy and V12 Gy values were 24.3 cm3 (range 0.6–45.0 cm3) and 18.0 cm3 (range 0.4–28.9 cm3), respectively. The median PTV for SHRT was 22.3 cm3; the V4 Gy/fraction was 5.9 cm3. Mean patient setup errors for SRS and SHRT are summarized in Tab. 4.

Local and distant brain control

The median follow-up was 9.6 months (range 0.9–27.4 months). The median follow-up of living patients (23) was 10.8 months (range 2.4–27.4 months). Radiological follow-up data were not available for 4 patients. Median time to local brain recurrence was 7.3 months (range 3.5–9 months). Median time to any intracranial failure (local or distant) was 5.9 months. LC rates after 6 and 12 months were 91 and 77 %, respectively. No statistically significant difference in LC rates between the SRS and SHRT treatments was observed (Fig. 1). A total of 4 patients presented local recurrence: 1 patient after SRS treatment with 17 Gy and 3 patients after SHRT treatment with 4 × 6 Gy, 6 × 4 Gy and 10 × 4 Gy. Details of patients with local recurrence are summarized in Tab. 2. Of the 7 patients with residual tumor after surgery, 6 presented no local recurrence; radiological follow-up data were not available for 1 patient. DBC rates at 6 and 12 months were 61 and 33 %, respectively. A total of 23 patients (61 %) developed brain metastases at new sites during the follow-up period. Tumor growth along the surgical access route was observed in 2 patients, suggestive of leptomeningeal seeding. Median survival after regional recurrence was 6.4 months. OS at 6 and 12 months was 87 and 63.5 %, respectively, with a median OS of 15.9 months (Fig. 2). At the last follow-up, 19 of the 42 patients had passed away. Salvage radiotherapy was applied in 16 patients (38 %), 15 patients received WBRT and 1 patient was treated using radiosurgery. The median estimated time to salvage irradiation was 13.4 months.

Fig. 1
figure 1

Local control. a All patients, b patients stratified by treatment. SRS stereotactic radiosurgery (solid line in b), SHRT stereotactic hypofractionated radiotherapy (dotted line in b)

Full size image
Tab. 2 Characteristics of patients with a local recurrence
Full size table
Fig. 2
figure 2

Overall survival of all patients

Full size image

Toxicity

Symptomatic and pathologically proven radionecrosis occurred in 1 patient treated by SRS. This patient received a single dose of 17 Gy. The PTV, V12 Gy and V10 Gy values were 13.3 cm3, 21.5 cm3 and 29.6 cm3, respectively. The most frequent acute toxicities were mild headaches and nausea. No acute grade 2 or higher toxicity was observed.

Discussion

In patients with limited brain metastases, the positive effects of WBRT in decreasing the rate of intracranial progression do not translate into survival or quality of life benefits [22]. Up until now, no prospective study including a quality of life assessment has investigated stereotactic irradiation of the resection cavity as an alternative to upfront WBRT after surgery. In light of these findings, it is our practice to omit or defer WBRT in favor of SRS or SHRT in postoperative patients with a limited number of brain metastases. In our study, the LC rates after 6 and 12 months were 91 and 77 %, respectively, with a median follow-up of 9.6 months. These results are comparable to the LC rates of 75–90 % achieved previously with adjuvant WBRT [17].

Frameless image-guided SRS and SHRT

Noninvasive patient immobilization and frameless image guidance as applied in SRS of brain metastases are techniques that have been recently introduced. A number of reports regarding the accuracy of image-guided methods have demonstrated that submillimeter accuracies can be achieved and that accuracy is comparable to the traditional frame-based approach [5, 9]. In our experience, mean patient setup errors for the SRS and SHRT treatments were comparable to published reports. The number of reports exploring the efficacy and morbidity associated with frameless image-guided SRS and SHRT of the resection cavity is limited ([6, 11, 18, 21, 23, 25, 26], Tab. 3). The SRS studies suggest that LC rates of 74–89 % can be obtained using a radiosurgical dose of 18 Gy [12, 18, 21, 23]. However, the SRS dose to the resection cavity in the absence of WBRT remains a topic of investigation. In multivariate analysis, smaller PTV volumes and marginal doses < 18 Gy were predictive for reduced LC [18]. In our analysis, we observed only a single recurrence in the SRS group, with a dose of 17 Gy and a PTV volume of 11.1 cm3. The SHRT studies suggest that LC rates of 76–89 % can be obtained using regimens of 3–10 fractions with total doses ranging from 20 to 40 Gy [6, 25, 26]. However, it is difficult to compare the results of these studies due to the large hetereogeneity of the fractionation regimens. A recent report comparing different dose concepts in SHRT showed that EQD2s of ≥ 35 Gy seem to be the most effective concept in patients with primary or recurrent limited primary brain metastases [14]. We compared SRS to different hypofractionated regimens and failed to find any fractionation-associated differences in LC due to the high diversity of dose concepts. However, in our study the LC rate in the SHRT group was 60 % and in 2 out 3 patients with recurrence, the EQD2 was < 35 Gy.

Tab. 3 Summary of published frameless image-guided series
Full size table
Tab. 4 Patient setup accuracy
Full size table

The importance of other key issues, such as target volume definition and the use of margins, also has to be established.

Target volume

At present, there are no well established guidelines concerning the definition of the target after surgical resection of brain metastases. There are no prospective data showing that the inclusion of the surgical track has an impact on LC. However, surgical resection seems to be crucial in terms of local recurrence incidence rates. Patel and colleagues reported that resection of the tumor in a piecemeal fashion significantly increased the incidence of local recurrence in comparison with en bloc resection [17]. The resection of metastatic lesions in contact with or involved with the cerebrospinal fluid (CSF) pathway is associated with a significantly higher incidence of leptomeningeal seeding than resection of tumors separated from the CSF pathway by brain parenchyma [1]. The addition of margins around the surgical cavity remains controversial. Neuropathological studies have shown that infiltration may be responsible for the presence of clinically undetectable cancer islands showing a maximum infiltration depth of 1–3 mm [2].

Toxicity

Choi and colleagues reported the first prospective data showing that the addition of a 2-mm margin to the resection cavity resulted in a decreased local failure rate at 12 months from 16 to 3 %, without increasing toxicity [6]. In our series, we observed symptomatic and pathologically proven radionecrosis in 1 patient treated by a single fraction of 17 Gy. Clinical data on the toxicity profile of postoperative hypofractionated SRS and SHRT remain limited. Wang and colleagues reported a combined rate of all toxicities (radionecrosis, prolonged steroid use and new-onset seizures) of 9 % using Cyberknife® (Accuray, Sunnyvale, CA, USA) hypofractionated SRS with 3 fractions of 8 Gy daily [26]. No toxicity grade 2 or higher was reported by Steinmann and colleagues using three different fractionation concepts with SHRT to the resection cavity [25]. For most lesions, 40 Gy in 10 fractions was applied according to the guidelines reported in the previous phase II trial of SHRT, which recommended that the V4 Gy per fraction for normal brain should not exceed 20 cm3 [8]. In our SHRT treated patient group, the median normal brain V4 Gy/fraction was 5.9 cm3 and we did not observe acute grade 2 or higher toxicity.

Conclusion

Frameless image-guided LINAC-based adjuvant SRS and SHRT is a safe and effective treatment after resection of brain metastases in patients with oligometastatic disease. The system’s accuracy is comparable to that of frame-based systems. In the current study, we found SHRT to be comparable to single-fraction SRS in terms of local tumor control and toxicity. This treatment strategy and its correlation with quality of life should be explored by additional studies.