Background. 

The standard treatment for primary intracranial ependymomas is maximal safe resection followed by fractionated radiation therapy. The extent of the resection has been shown to influence survival and tumour control. Stereotactic Radiosurgery by definition is a highly conformal, high dose radiation to the target with a rapid decline of the dose peripherally and is used to treat recurrent, residual or initially unresectable ependymomas.

Methods.

A retrospective review of 94 tumours in 51 consecutive patients with ependymoma and treated with Gamma Knife Stereotactic Radiosurgery (GKSRS) at the National Centre for Stereotactic Radiosurgery in Sheffield from 1993 to 2021, was undertaken. GKSRS planning and delivery was performed using Leksell Gamma Knife Models S, C, 4C and Perfexion. The diagnosis and grading of ependymomas were carried out according to the WHO criteria and a multidisciplinary team validated the indication for GKSRS treatment. The number of surgeries, location, time between surgery and GKSRS, prescribed dose, target volume, tumour control rate, overall (OS) and progression free survival were recorded.

Results.

Of the 51 patients treated, 38 patients underwent GKSRS for a single lesion, 7 patients had single session GKSRS for multiple lesions and 6 patients had multiple sessions for recurrent lesions. All patients received 12 - 25Gy (mean 18.23Gy) for tumour volumes between 0.053 - 48.30cm³ (mean 1.872cm³). In addition to the 88 cases of WHO grade II and III ependymomas, 6 subependymomas (WHO I) were also included in the analysis. OS in WHO grade III tumours (14 patients) was 71.42%, at 1 year (range 181–5750, mean 988.79 days). WHO grade II tumours (25 patients) fared better with verified survivorship and tumour control of 96% at 1 year (range 85-10408, mean 3953.38 days). WHO grade I tumours OS at 1 year was 100% (range 1969–6918, mean 3774 days). Post treatment adverse events were reported in 5 patients (9.8%). 80% of these were post radiation effects which occurred at 6 months post GKSRS and resolved with a short course of steroids.

 Conclusions.

 GKSRS is an excellent primary and adjuvant treatment for ependymomas including recurrent tumours with a good rate of local control. Patients with WHO grade I and II ependymomas have better local and regional control of the disease compared to WHO grade III tumours. GKSRS is currently underutilised as an adjuvant treatment in the management of residual or recurrent ependymoma in the U.K., despite being clinically commissioned by the National Health Service since 2016.

Purpose:

Pilocytic astrocytomas (PA) are the most common gliomas (WHO I) in children. According to lots of trials in different countries Stereotactic Radiation Therapy (SRT) and  Radiosurgery (SRS) provide tumor growth control if surgical removal is not possible or in case of recurrence. In this paper, we have summarized our own experience with radiation treatment of PA, which composes the largest series available.

Materials and  Methods:

The study included 410 consecutive patients who received radiation treatment at the N. N. Burdenko National Medical Research Center of Neurosurgery between April 2005 and January 2018. There were 225 female and 185 male patients. The study group consisted of 110 adults and 300 children (including those under the age of 18). The median age was 10.8 years (Q1-Q3: 5.4-19.3 years). Histopathological verification was performed in 307 cases (74.8%). In 103 patients (25.2%) the diagnosis was established according to clinical and neuroimaging data (MRI, CT-perfusion and PET). The indication for the radiation treatment was a residual tumor after surgery (204 pts, 49.7%) or recurrence after previous removal or chemotherapy (206 pts, 50.3%). Standard fractionation was used most frequently (292 cases, 71.2%). Hypofractionation and radiosurgery were used in 61 (14.9%) and 57 (13.9%) cases, respectively.

Results:

The median follow-up period was 68 months (range - 3-300 months) from the diagnosis establishment. 391 patients (95.4%) were available for the follow-up. The median follow-up period after irradiation was 45 months (range - 3-162 months). The 5-year PFS was 97,5% and 5-year OS was 99%. Undesirable events occurred in 77 (19.7%) patients: pseudoprogression – in 67 patients (89.5%), tumor progression – in 8 patient (4 local and 4 distant repalses), brain stem necrosis and intratumoral hemorrhage – in 1 case each.  In cases where radiotherapy was delayed until disease progression, the tumor volume was larger and the patients' state was worse than when irradiation was performed immediately if there was a residual after surgery.

Conclusion:

Stereotactic irradiation is the effective method of treatment for PA in patients with residual tumors, patients with PA relapse and patients with progressive disease. The treatment should be indicated as early as possible after partial resection. The most common adverse event after irradiation is pseudoprogression.

Purpose: Using SBRT for reirradiation of skull base malignancies is one of the most challenging tasks due to the strict requirements on patient immobilization and the high complexity of treatment planning. We present our experience and outcomes of the SBRT treatment for skull base recurrences.

Methods: Forty-three recurrent skull base patients were treated with SBRT between December 2014 and April 2020. All patients were immobilized using a three-point customized cushion/mask/bite-block system. High quality volumetric modulated arc therapy conformal plans were generated in Pinnacle or RayStation treatment planning system. Clinical goals for critical structures were customized for each patient based on prescription dose, distances from nearby organ-at-risk (OAR) structures to target, OAR tolerance, as well as the patient specific prior radiation treatment history. Patients were treated on Varian Truebeam STx with Exactrac and CBCT imaging guidance. Dosimetry was summarized and the outcome included overall survival, local control and toxicities.

Results: Forty-two percent of cases (n=18) were of squamous cell histology. Median follow up was 18.2 months (range 7.8-60.6 months). Twenty-seven percent (n=12) of patients received induction systemic therapy, 58% (n=25) received concurrent and 23% (n=10) received adjuvant therapy. The median SBRT dose was 45Gy (40-45Gy in 5 fractions (n=40) or 36Gy in 4 fractions (n=3)). Median primary target volume was 17.7 cm³ (range 1.5-58.5 cm³). The median target coverage was 95% (range 79%-100%) and target mean dose was 105% (range 101.9%-114.1%) of prescription dose. Brainstem achieved maximum dose of 11Gy or ALARA for 60% (n=26) of the patients, while 4 patients had brainstem abutting target and received brainstem dose V21Gy < 0.5 cm³. Other OARs (optic pathway, carotid, cochlea, temporal lobe) also achieved clinical goals. The use of 2mm PTV margin seemed adequate for skull base SBRT based on our previous report of intra-fractional positional accuracy using Exactrac/CBCT imaging system. The two-year survival and local control were 71.8% and 75.4%, respectively. There were no Grade 3+ acute toxicities. Two patients presented Grade 3+ late toxicities, including one Grade 3 toxicity associated with temporal lobe necrosis, and one Grade 4 osteoradionecrosis.

Conclusion: with the combination of improved immobilization, high quality conformal treatment planning and advanced image guidance, our optimized SBRT workflow for skull base reirradiation showed excellent outcomes with acceptable toxicities.

Background

Sulfasalazine (SAS), a well-known anti-inflammatory drug has shown tumor selective radiosensitizing properties in preclinical studies. The antioxidant glutathione (GSH) produced at high levels in glioma cells normally protects against radiation injury by scavenging the reactive oxygen species produced during radiation therapy (RT). SAS blocks the uptake of cysteine through the _xCT-channel, a rate-limiting step for glioma cell GSH production. We have previously shown that tumor growth slowed down and overall survival (OS) was prolonged when SAS was combined with gamma knife surgery (GKS) in a glioma animal model compared to either treatment alone. Our hypothesis is that SAS potentiates the efficacy of SRS in glioblastoma (GBM) patients with a low risk of side effects. This open-label, single-arm trial aims to evaluate the maximum tolerated dose (MTD) and the preliminary efficacy of SAS combined with GKS in first or second relapse of GBM following standard of care.

Methods

The trial is designed as a standard 3 + 3 phase 1-dose escalation study with 3-6 patients in each dose-cohort. The trial will enroll 12-24 patients for treatment with oral SAS (1.5, 3.0, 4.5 or 6.0 g/day) 3 days prior to single session GKS with 12 Gy prescription dose to the tumor margin. The dose will be escalated depending on the absence/presence of severe toxicity in the previous cohort of treated patients. If more than 1 patient in a cohort of 3 or 6 patients (≥33%) experiencing grade 3 or higher toxicity levels, the study will be terminated and the dose will be defined as MDT. Toxicity is graded using the Common Toxicity Criteria for Adverse Events (CTCAE) v5.0 recorded the first 30 days. Primary end-point is to establish the recommended dose for efficacy testing in phase II/III trials. Secondary end-points are assessments of 1) intratumoral GSH production evaluated with GSH-spectroscopy before and after SAS treatment, 2) late adverse radiation effects utilizing 11C-MET-MRI-PET 3) changes in KPS/quality of life (FACT-Br), 4) need for steroidal treatment, 5) progression free survival (PFS) utilizing Response Assessment in Neuro-Oncology criteria and 6) OS. PFS and OS will be compared with historical controls treated between 2018-2022.

Discussion: Due to the dismal prognosis for GBM patients, novel treatment modalities are urgently needed. The trial will establish the recommended dose for SAS repurposed as a radiosensitizer for a future phase 2/3 trial. Our proposed treatment may ultimately lead to increased effect of radiation therapy for these highly radioresistant tumors.

Background and Objective: Possible management strategies for recurrent glioblastomas of the brain include surgery, chemotherapy, radiotherapy, and combined treatments.  Over several treatment approaches, reoperation and reirradiation have shown promising results.  Indications for surgical resection rather than stereotactic radiosurgery are still debated.  We hereby conduct an expert consensus to provide guidelines for the treatment of recurrent glioblastomas.

Methods: To aggregate expert opinions in an anonymous fashion we adopt the conventional e(lectronic)-Delphi method.  This takes place over a series of rounds where communication occurs through electronic exchange.  First, a questionnaire is formulated upon a set of assumptions and solutions or options.  We search PubMed, MEDLINE, and EMBASE from 2000 to 2022 for studies reporting on surgical resection and/or stereotactic radiosurgery for recurrent glioblastomas of the brain.  Two independent reviewers select studies and abstract data.  We list the variables considered to define the treatment modality and design the questionnaire.  Second, an expert panel is identified and invited to provide opinions.  Members are recruited based on their grounded expertise in both neuroradiosurgical and neurosurgical fields.  Representative members of intercontinental and international stereotactic radiosurgical and neurosurgical societies join efforts to elaborate a pragmatic worldwide set of guidelines.  Based on predetermined criteria, responses are analyzed and ranked.  A second questionnaire is developed employing the results and feedback from the first round.  Responses are again collected and assessed for consensus.  The process terminates when an acceptable degree of consensus is reached. 

Results: Three rounds are usually sufficient to achieve consensus with the largest adjustments usually occurring between the first and second rounds.  Consensus is generally reached at the end of the third round for the majority of the items originally posed.  Results bring out discrepancies among intercontinental healthcare backgrounds, featuring thorough reasons for treatment choices.  Guidelines are advanced in the shape of the items gaining consensus.  

Conclusions: The proposed expert consensus document highlights adherence and divergencies among experts on the treatment of recurrent glioblastoma.  Taking into account clear items of agreement between members, we postulate a pragmatic worldwide set of guidelines.   Across social and cultural heterogeneities of healthcare environments, we aim to provide universal determinants to best select patients for surgery or stereotactic radiosurgery, through a decision-making process embodying a multidisciplinary neuro-oncology team.  Larger prospective studies are still warranted to provide a higher level of evidence.

Glioblastoma (GBM) is the most common and devastating primary brain tumor in adults. The current standard of care includes maximal safe resection followed by concurrent radiation and temozolomide-based chemotherapy.

Near uniform recurrence forces us to pursue alternative approaches to improve patient outcomes. Neoadjuvant irradiation is used in other cancer types to reduce the tumor burden before resection, however, its role in the preoperative setting has not been evaluated in the treatment of primary brain tumors.

In this preclinical study, we evaluated if preoperative irradiation before surgical resection improves survival.  39 immunocompromised adult athymic nude female rats were orthotopically engrafted with a patient-derived glioma cell line and randomized to receive different treatments: not treated (NT); tumor resection only (RE),  tumor resection followed by radiotherapy (RERT), and radiotherapy prior to tumor resection (RTRE). The rats received 30 Gy hyperfractionated stereotactic irradiation in 5 fractions of 6 Gy. RNAseq was performed from the samples of at least 2 rats per group. Kaplan-Meier analysis showed a significant survival benefit of the  RTRE group of 3.43 weeks over the RERT group (P-value <0.0001). RNA expression analysis showed 2,451 differentially expressed genes. Among the genes of interest, we found upregulation of miR186, which is a well-known antioncogenic micro-RNA, and downregulation of metallothionein 1E (MT1E), a gene that has been related to a decrease in the activity of the tumor suppressor gene p53. Moreover, the small nucleolar RNA, SNORD60, appears downregulated. The reduction on SNORD60 drives a reduction in cholesterol trafficking which could be associated with less angiogenesis.

This study demonstrates in principle the benefits of neoadjuvant therapy in GBM, and highlights the importance of epigenetic regulation in the treatment outcomes, suggesting these non-coding RNAs as potential new druggable targets. Further clinical studies are required and underway to determine the impact of these novel radiation strategies in patients.

Introduction

Distinguish between radiation necrosis (RN) and tumor progression, in patients with irradiated primary or metastatic brain tumors, is a diagnostic challenge. Also the use of new MRI sequences, like diffusion, perfusion-weighted and spectroscopy, or PET with new amino acid tracers, is not always able to differentiate these two entities.

To overcome this crucial problem, encouraging results have been obtained using the analysis of delayed contrast extravasation MRI to calculate high resolution maps, called “treatment response assessment maps” (TRAMs).

Aim of this exploratory analysis is to assess TRAM ability in differentiate between radiation effect and tumor progression in a small cohort of brain tumor patients treated with radiation therapy (RT).

Materials and methods

Thirty-four patients irradiated for primary and metastatic brain tumors were evaluated.

12patients have primary brain tumors, 22patients have brain metastases from different solid tumors.

Distinguish by histological subtypes and type of treatment, the 12patients with primary brain tumors were: 8glioblastoma, 2anaplastic astrocitoma, 1pleomorphic xanthoastrocytoma WHO grade II, and 1anaplastic xanthoastrocytoma WHO grade III, treated with surgery followed by RT and concomitant and\or adjuvant chemotherapy with temozolomide. Among brain metastatic patients, primary tumor was: 18non-small cell lung cancer, 2malignant melanoma, 1breast cancer and 1renal cell carcinoma.All of them were treated with stereotactic radiosurgery at the dose of 20-24Gy in 1fr, or with hypofractionated stereotactic radiotherapy at the dose of 27-30Gy in 3fr.

All images were uploaded and elaborate into the image workstation ([Brainlab AG, Olof-Palme-Straße 9,  81829 Munich]).

TRAMs were calculated by subtracting T1 MRI images acquired 5 minutes after contrast injection from the T1 MRI images acquired 60-105 minutes later. On TRAMs, radiation effects appeared as red areas whereas persistent tumoral lesion appeared as blue areas.

Results

From February 2021, 34 patients have been evaluated, in a prospective study, with this novel MRI modality. During their follow-up, 13patients(38%) showed a clinicoradiologic suspicion of a persistent tumoral lesion or progressive disease, and 21(62%) a suspicion of RN. For 14patients a brain MET-PET has been performed. TRAMs analysis have shown a fair agreement with clinicoradiologic diagnosis, perfusion-weighted MRI, and PET imaging. Moreover, 7patients underwent surgical resection, with histopathological confirm of persistent disease in 4 and radionecrosis in 3.

Conclusions

These preliminary results show the ability of TRAMs evaluation in distinguish between RN and progressive disease. The recruitment of new patients continues, and further evaluations are ongoing to evaluate sensitivity and positive predictive value of TRAMs analysis.

Introduction: The optimal treatment paradigm for brain metastasis that recurs locally after initial radiosurgery (BMRS) remains an area of active investigation.  Here, we report outcomes for patients with BMRS treated with stereotactic laser ablation (SLA, also known as laser interstitial thermal therapy, LITT) followed by consolidation radiosurgery (cSRS).

Methods: Clinical outcomes of 20 patients with 21 histologically confirmed BMRS treated with SLA followed by consolidation SRS and > 6 months follow-up were collected retrospectively across three participating institutions.

Results: Consolidation SRS (5 Gy x 5 or 6 Gy x 5) was carried out 16-73 days (median of 26 days) post-SLA in patients with BMRS. There were no new neurological deficits after SLA/cSRS. While 3/21 (14.3%) patients suffered temporary Karnofsky Performance Score (KPS) decline after SLA, no KPS decline was observed after cSRS. There were no 30-day mortalities or wound complications. Two patients required re-admission within 30 days of cSRS (severe headache that resolved with steroid therapy (n=1) and new onset seizure (n=1)).  With a median follow-up of 228 days (range: 178-1367 days), the local control rate at 6 and 12 months (LC₆, LC₁₂) was 100%.  All showed diminished FLAIR volume surrounding the SLA/cSRS treated BMRS at the six-month follow-up; none of the patients required steroid for symptoms attributable to these BMRS.  These results compare favorably to the available literature for repeat SRS or SLA-only treatment of BMRS.

Conclusions: This multi-institutional experience supports further investigations of SLA/cSRS as a treatment strategy for BMRS.

Objective: We evaluate the clinical outcome in patients with large, high-risk brain metastases (BM) treated with different dose-strategies of two-fraction dose-staged Gamma Knife radiosurgery (GKRS).

Methods: 142 patients from two centers, who had been treated with two-fraction dose-staged GKRS between June 2015 and January 2020, were analyzed. Depending on the changes in marginal dose between the first and second GKRS treatments, the study population was divided into three treatment groups: dose-escalation, dose-maintenance and dose-deescalation.

Results: In our study population, 166 large, high-risk BM were treated with the two-fraction dose-staged GKRS treatments. The median volume decreased significantly between the first and second GKRS and to last FU.  We could show that the outcome was influenced by different dose strategies and primary tumor origin. Of note, the vast majority of dose-staged BM did not show any significant post-radiosurgical complications. 

Conclusions: In patients with large or high-risk BM, all three dose-staged GKRS treatment strategies represent effective local treatment methods with excellent local tumor control rates. Complication rates are acceptable and in line with previously reported single-fraction, but also with dose-staged radiosurgical complication rates. Depending on the primary tumor origin and initial treatment volume, different dose-strategy options are available.

Introduction

Brain metastases (BM) can be treated with stereotactic radiosurgery (SRS) to achieve long term control. In the event of local progression beyond 6 months from first SRS, BM can be re-treated with salvage SRS. Factors such as patient fitness and local control duration (LCD) should be considered when deciding on salvage SRS. Short LCD after first SRS might predict poor response to salvage SRS. We report the outcomes of salvage SRS at our centre, and evaluate if LCD after first SRS predicts outcomes after salvage SRS.

Method

Patients who had BM re-treated with salvage SRS were included in this study.  Statistical analysis was performed with SPSS Version 26. Kaplan-Meier analysis was performed to compare LCD after first SRS and salvage SRS.

Results

20 patients had 27 BM re-treated with SRS. Median age was 62 years at first SRS and 62.5 years at salvage SRS. All patients were ECOG PS 0 or 1. 15 patients had 1 BM retreated, 3 patients had 2 BM retreated, and 2 patients had 3 BM retreated. Median BM volume was 1.8ml at first SRS and 1.1ml at salvage SRS.  At first SRS, 14 BM (51.9%) were treated with 22Gy but at salvage SRS only 5 BM (18.5%) received 22Gy. Other doses at salvage SRS were 15Gy (7.4%), 18Gy (29.6%) and 27Gy in 3 fractions (40.7%)

Median LCD after first SRS was 336 days. In comparison median LCD after salvage SRS was 732 days which was significantly longer (log rank p = 0.034). At the time of analysis, of the 27 re-treated BM, 10 experienced further local progression while the remaining 17 were censored; 9 BM due to 8 patient deaths and 8 BM were stable at the time. As continuous variables, first LCD and salvage LCD were moderately correlated (Pearson’s correlation = 0.444, p = 0.020) but on Cox regression first LCD was not significantly associated with salvage LCD (HR = 0.997, 95% CI 0.994 – 1.001, p = 0.107)

Conclusion

Salvage LCD is not inferior but in fact significantly longer than first LCD despite lower salvage SRS dose. First LCD is not a predictor for salvage LCD therefore salvage SRS should be considered even if first LCD was relatively short provided the patient is otherwise suitable for re-treatment. Our sample is relatively small resulting therefore we aim to repeat analysis with a larger cohort.  

Background:  Radiation therapy (RT) and immune checkpoint inhibitors (ICIs) can be synergistic, although abscopal responses are uncommon.  We hypothesized that stereotactic body radiation therapy (SBRT) delivered to two lesions sequentially – effectively delivering a radiation “booster shot” after initial in situ RT tumor "vaccination" – with concurrent ICI may potentiate a clinically meaningful immune response.

Methods:  Patients with metastatic carcinoma who experienced progression in 1-5 lesions while previously receiving ICI monotherapy were enrolled in a single arm phase 2 trial (NCT04376502). Eligible patients had 3+ measurable lesions, 2 of which sequentially received SBRT separated by 1 week (#1: 40 Gy/5 fractions; #2: 30 Gy/5 fractions) on a 0.35T MR Linac using soft tissue tracking and automatic beam gating.  The same ICI was continued until disease progression or unacceptable toxicity.  The primary objective was to evaluate best overall response rate (ORR) in non-irradiated lesion(s) and was evaluated using Simon’s two-stage design.  

Results:  In this interim analysis, 9 patients with non-small cell lung cancer (NSCLC) (78%) or melanoma (22%) completed SBRT typically with concurrent pembrolizumab (44%) or nivolumab (33%).  SBRT was most commonly delivered to lung (22%), adrenal gland (22%), and liver (17%) metastases.  Median follow up was 10.7 months (range, 4.6-22.2) from enrollment.  8 patients (89%) were alive at the time of analysis.  Best ORR in non-irradiated lesions was complete response (CR), partial response (PR), stable disease (SD), and progressive disease (PD) in 0, 1 (11%), 4 (44%), and 4 (44%) patients, respectively; 5 (55%) achieved clinical benefit with PR/SD.  Best ORR in irradiated lesions was CR, PR, SD, or PD in 2 (22%), 4 (44%), 3 (33%), and 0 patients, respectively. One NSCLC patient who initially progressed on pembrolizumab achieved a PR at a non-irradiated site (left adrenal) and CR at both irradiated sites (lung, right adrenal).  He experienced distant progression (solitary brain metastasis) 6.5 months after SBRT, was salvaged with stereotactic radiosurgery, and remains on pembrolizumab without further progression now 21.7 months after SBRT.  Seven patients (77.8%) experienced distant progression after a median 3.7 months (range, 1.8-6.5) from SBRT.  No grade 2+ toxicity occurred.  

Conclusions:  Despite experiencing disease progression on immunotherapy, SBRT “revaccination” while continuing the same immunotherapy appears to achieve a favorable response at both irradiated and non-irradiated sites without causing significant toxicity.  These early results suggest that this novel treatment strategy provides clinically meaningful benefit and could delay the need to change systemic therapy.  

The efficacy of combining radiation therapy with immune checkpoint inhibitor blockade to treat brain tumors is currently the subject of multiple investigations and holds significant therapeutic promise.  However, the long-term effects of this combination therapy on the normal brain tissue are unknown. Here, we examined mice that were intracranially implanted with murine glioma cell line, and treated with a combination of 10 Gy cranial irradiation (RT) and anti-PD-1 checkpoint blockade (aPD-1). This combination treatment cured 75% of the mice over 90 days of study duration. We examined the contralateral normal brain in the long-term survivors after the combined treatment. Post-mortem analysis of the cerebral hemisphere contralateral to tumor implantation showed that neural stem cells were well preserved in subventricular zone.  In addition, we observed a drastic reduction in the number of mature oligodendrocytes in the subcortical white matter. Importantly, this observation was evident specifically in the combined (RT + aPD-1) treatment group but not in the single treatment arm of either RT alone or aPD-1 alone. Elimination of microglia with a small molecule inhibitor of colony stimulated factor-1 receptor (PLX5622) prevented the loss of mature oligodendrocytes. These results identify for the first time a unique pattern of normal tissue changes in the brain secondary to combination treatment with radiotherapy and immunotherapy. The results also suggest a role for microglia as key mediators of the adverse treatment effect.

Purpose: We evaluated the efficacy and safety of simultaneous stereotactic radiosurgery for multiple lesions using a single isocenter and a volume-and critical structure-adapted dosing strategy in patients with 4 to 10 brain metastases.

Study Design: Adult patients with 4 to 10 brain metastases were enrolled on this prospective single-institution trial. The primary endpoint was overall survival. Secondary endpoints were local recurrence, distant brain failure, neurologic death, and rate of adverse events. Exploratory objectives were neurocognition, quality of life, dosimetric data, salvage rate, and radionecrosis. Dose was prescribed in a single fraction per RTOG 90-05 or as 5 Gy × 5 fractions for lesions ≥3 cm diameter, lesions involving critical structures, or single-fraction brain V_12Gy >20 mL.

Results: Forty patients were treated on the study. The median age was 61 years, median Karnofsky performance status 90, and median number of brain lesions was 6. We used the graded prognostic assessment to estimate expected survival, and twenty-two patients lived longer than expected after SRS, with 1 living patient who had not reached that milestone at the time of publication. Median overall survival was 8.1 months with a 1-year overall survival of 35.7%. The 1-year local recurrence rate was 5% (10 of 204 of evaluable lesions) in 12.5% (4 of 32) of the patients. Distant brain failure was observed in 19 of 32 patients with a 1-year rate of 35.8%. There were no grade 3-5 treatment-related adverse events. Radionecrosis was observed in only 5 lesions, with a 1-year rate of 1.5%. Rate of neurologic death was 20%. Neurocognition and quality of life did not significantly change 3 months after SRS compared with pretreatment.

Conclusions: These results suggest that single-isocenter multitarget stereotactic radiosurgery using adapted dose for target volume and critical structures is an effective and safe treatment for patients with 4 to 10 brain metastases.

Currently, the biological effectiveness of photon beam radiosurgery applications is assessed using the concept of Biologically Equivalent Dose (BED) and assuming instant dose delivery. In practice however, radiosurgery sessions are finite and fairly diverse in time, spanning from several min to more than 1h for the same tumor location and dose prescription. This reflects majorly the corresponding diversity in technical configuration among the clinically available radiation systems. In a previous work, the treatment data of 30 patients who underwent CyberKnife radiosurgery for the treatment of vestibular schwannomas (VS) were retrospectively studied to acquire corresponding time-resolved dose and dose-rate distributions for a marginal dose prescription of 13Gy in a single fraction. It was demonstrated that CyberKnife radiosurgery deviates from the “acute exposure” approach, comprising highly inhomogeneous dose and dose-rate distributions in both the spatial and the temporal domain. This induces local variations in the sublethal DNA damage repair (SDR), which, in turn, will affect the biological response of a treatment provided its delivery time is comparable to the repair kinetics rate. Using a BED calculation framework accounting for SDR effects, via a bi-phasic repair model with a slow (t_1/2=129.6min) and a fast (t_1/2=11.4min) component, along with the best available radiobiological parameters for the human central nervous system tissue, it was shown that each physical dose iso-surface is associated with a range of BED_SDR levels, typically spanning 2% (1σ) around the mean value, while the marginal BED_SDR delivered to the tumor by the prescription dose iso-surface deteriorates with total treatment time. To enable marginal BED_SDR estimations prior to- and during treatment planning, the available BED_SDR data were re-evaluated against the ratio of the weighted mean collimator size, _wCS, used to irradiate the planned dose distributions and the tumor physical volume (TV). For each case, _wCS was calculated weighting the contribution of each collimator by its relative number of radiation beams. As shown in Figure 1, there is a statistically significant monotonic relation between marginal BED_SDR and the ratio _wCS/TV. For the planning strategies applied to the studied patient cohort, it can be expressed using a two-parameter power fitting model (see Figure 1). Overall, results suggest that the selection of collimator sizes and their relative contribution in dose delivery can be optimized during CK treatment planning, to aid BED delivery standardization in parallel to other physical dose goals, such as conformality and dose gradient indices, where collimator size is also of crucial importance.

Introduction:

Vestibular schwannomas (VSs) are benign, slow-growing tumors. Management options include observation, surgery, and radiation. In this retrospective trial, we aimed at evaluating whether biologically effective dose (BED) plays a role in tumor volume changes after single-fraction first intention stereotactic radiosurgery (SRS) for VS.

Method:

We compiled a single-institution experience (n = 159, Lausanne University Hospital, Switzerland). The indication for SRS was decided after multidisciplinary discussion. Only cases with minimum 3 years follow-up were included. The Koos grading, a reliable method for tumor classification was used. Radiosurgery was performed using Gamma Knife (GK) and a uniform marginal prescription dose of 12 Gy. Mean BED was 66.3 Gy (standard deviation 3.8, range 54.1-73.9). The mean follow-up period was 5.1 years (standard deviation 1.7, range 3-9.2). The primary outcome was changes in 3D volumes after SRS as function of BED and of integral dose received by the VS.

Results:

Random-effect linear regression model showed that tumor volume significantly and linearly decreased over time with higher BED (p < 0.0001). Changes in tumor volume were also significantly associated with age, sex, number of isocenters, gradient index, and Koos grade. However, the effect of BED on tumor volume change was moderated by time after SRS and Koos grade. Lower integral doses received by the VSs were inversely correlated with BED in relationship with tumor volume changes (p < 0.0001). Six (3.4%) patients needed further intervention.

Conclusion:

For patients having uniformly received the same marginal dose prescription, higher BED linearly and significantly correlated with tumor volume changes after SRS for VSs. BED could represent a potential new treatment paradigm for patients with benign tumors, such as VSs, for attaining a desired radiobiological effect. This could further increase the efficacy and decrease the toxicity of SRS not only in benign tumors but also in other SRS indications.

Introduction
In an earlier study by King’s College London (KCL), a framework for the automatic segmentation of vestibular schwannomas (VS) from MRI scans was proposed. They demonstrated that artificial intelligence is capable of  annotating and calculating VS tumor volumes, which can benefit tumor progression monitoring. In this study, their algorithm is validated using data from another institution. The obtained results are subsequently compared to a multi-institutional model, in which the data from both centers is combined.

Methods
In addition to the original KCL dataset of 375 contrast-enhanced T1-weighted (ceT1) MRI scans, a total of 1,115 ceT1 scans of individual VS patients from the ETZ hospital in Tilburg, the Netherlands, were used. All tumors were manually annotated by a neurosurgeon at time of treatment. Employing the original 2.5D convolutional neural network, two different models were subsequently trained. First, one model is trained solely on 176 scans from the KCL dataset, using an additional 20 and 46 scans for hyper-parameter tuning and internal validation, respectively. This single-institution model is externally validated on the entire ETZ dataset. The second model employed 242 scans from the KCL dataset and all of the ETZ data from before 2015 (733 scans), with the remainder of the ETZ data equally distributed between a tuning and validation set. The remaining 133 scans from the KCL dataset were also used for validation.

Results
The single-institutional model achieved internal and external validation mean Dice scores of 92.0±5.1% and 64.5±32.4%, respectively. The external validation set included 175 scans where the model failed to recognize any VS, resulting in a Dice score of zero for these scans, which skews the results. The second model yielded validation mean dice scores of 92.5±5.1% for the ETZ scans and 89.1±9.6% for the KCL scans. During validation, most of the low-scoring segmentations are of tumors that are either very small or cystic.

Conclusions
The significant difference in performance between internal and external validation of the first model, thereby surmising a poor generalization, can be explained by variations in spatial resolution and image acquisition parameters between the two datasets. The second model shows performances approaching inter-observer variability of human annotators (cf. 93.8±3.1%). The results show that creating a robust and well-generalizing model from single-institution data is challenging. However, the multi-institutional results empower the earlier demonstrated capabilities of the framework for the automatic segmentation of VS.

Introduction

Approximately one third of all vestibular schwannomas (VS) show a volumetric increase one year after Gamma Knife radiosurgery (GKRS) treatment, which can indicate transient tumor enlargement (TTE) and may cause symptoms due to increased mass effect. It is therefore clinically relevant to be able to predict tumor response relatively early after GKRS. Computer-aided prediction models using classical machine learning have already shown promising results in predicting GKRS tumor response in VS patients. The objective of this study is to investigate the feasibility of the modern machine learning method of deep learning to predict early tumor response after GKRS.

Methods

A total of 1118 contrast-enhanced T1 MRI scans, obtained during GKRS treatment of VS patients, were used. These images were cropped to include only the tumor and its direct surroundings using a bounding box. This data was subsequently used to train a three-dimensional autoencoder model to extract 256 tumor image features automatically. The learned features were subsequently employed in a neural network classifier to predict tumor response based on the treatment scan. For this classification task, tumor enlargement and shrinkage were defined as at least a 15% volumetric increase and decrease with respect to the tumor volume at the time of GKRS. Furthermore, patients without a one-year follow-up scan were excluded, as well as patients with tumors smaller than 0.25cc due to the lack of sufficient image features.

Results

The exclusion criteria resulted in the inclusion of 395 patients with a mean one-year follow-up time of 12.14±0.70 months. Using our volumetric criteria, enlargement was observed in 122 (30.9%) tumors and shrinkage in 273 (69.1%). The prediction model achieved a sensitivity and specificity of 72.3% and 82.5%, respectively.

Conclusion

The specificity shows that the model is more tuned to predicting tumor shrinkage. Although tumor enlargement prediction performance (i.e., sensitivity) is lower, the overall performance of the model indicates that deep learning can be of predictive value for tumor response in VS patients. These methods should therefore be further investigated to improve upon the presented results.

Introduction: Fully automated artificial intelligence (AI) frameworks have recently reached outstanding performance for automatic segmentation of vestibular schwannomas (VS) from MRI. However, they often generalize poorly when the training data is acquired with scanners or imaging sequences that are different from those of the testing data. This problem strongly reduces the applicability of AI frameworks in real world radiosurgery settings.

To increase the robustness of AI tools, various technical approaches have been proposed. However, these techniques have been validated either on private datasets or on small publicly available datasets. To tackle these limitations, the authors organized the Cross-Modality Domain Adaptation (crossMoDA) challenge in conjunction with the 24th International Conference on Medical Image Computing and Computer Assisted Intervention (MICCAI 2021). 

Methods: The challenge’s goal was to segment two key brain structures involved in the follow-up and treatment planning of VS: the tumor and the cochlea. While contrast-enhanced T1 (ceT1) scans are commonly used for VS segmentation, recent work has demonstrated that high-resolution T2 (hrT2) imaging could be a reliable, safer, and lower-cost alternative to ceT1. For these reasons, the authors proposed an unsupervised cross-modality challenge to benchmark the generalization capability of techniques developed on images acquired with one sequence (ceT1) and tested on images acquired with another one (hrT2). Specifically, participants had access to a training set of unpaired annotated ceT1 (N=105) and non-annotated hrT2 (N=105). The automated segmentation tools developed by the challenge participants were then tested on a private hrT2 evaluation set (N=137). Images were collected on consecutive patients with a single sporadic VS treated with Gamma Knife stereotactic radiosurgery.

Results: A total of 341 teams registered for the challenge, allowing them to download the data. 55 teams from 16 countries submitted predictions to the validation leaderboard. Among them, 16 teams from 9 countries submitted their algorithm for the evaluation phase. The level of performance reached by the top-performing teams is strikingly high (best median Dice score - VS: 88.4%; Cochleas: 85.7%) and close to a state-of-the-art model developed using hrT2 scans and their corresponding annotations (median Dice score - VS: 92.5%; Cochleas: 87.7%).

Conclusions: The authors organized the first international benchmark assessing the robustness of AI frameworks for stereotactic surgery planning of VS. The excellent results obtained by the top-performing teams suggest that AI tools can be robust to different imaging sequences. The next challenge edition will assess the robustness of AI tools for Koos grade classification.

Our aim is to develop an automatic method for the accurate and reliable longitudinal volumetric radiological evaluation of brain metastases following SRS on high-resolution MRI using the available baseline metastases delineation and a deep learning network trained on various annotated datasets. 

Our method consists of five steps: 1) resolution matching between the baseline and the follow-up scans; 2) registration of the baseline and follow-up scans by Normalized Mutual Information; 3) Region of Interest (ROI) computation for each brain metastasis in the follow-up scan based on their baseline segmentation; 4) automatic segmentation of the brain metastases in the follow-up scan; 5) simultaneous analysis and visualization of the brain metastases segmentations in the baseline and follow-up scans. 

The brain metastases segmentation is performed with two 3D U-Net classifiers: one for small metastases (diameter <= 10mm) and one for large metastases (diameter > 10mm).

Clinical datasets consist of 267 patients with 374 image data sets in which there were 1157 brain metastases (average 4.3 metastases per patient). We used 329 image data sets with 1023 brain metastases to train our algorithm and 45 image data sets with 134 metastases to test the algorithm performances.

We developed a practical viewer (Fig. 1): the main viewer window where one can view the scans, a summary window detailing a number of important statistics such as change in tumor burden and a timeline window detailing the change over time between multiple scans. The window also allows the selection of a specific brain metastases in order to view information about it such as volume and volume difference between baseline and follow up scans. The summary window displays global information such as total volume change between scans, number of disappeared brain metastases. The timeline window displays the total tumor burden across multiple scans giving the physician an overview of total disease progression over time.

Our method achieves an average Dice score of 0.81 on a 5-fold cross validation study, surpassing the observer variability. It also includes a scan viewer that allows simultaneous viewing of pairs of pre- and post-SRS MRI scans.

The novelties of our method are that incorporates expert prior knowledge from the pre-treatment brain metastases delineation, that it uses relatively few brain metastases manual segmentations top train the classification deep network, and that it provides a scans viewer that enables the simultaneous visualization of the metastases segmentations and their analysis for longitudinal volumetric scan studies.