Faculty Bibliography

BACKGROUND:Differentiating brain metastasis progression from radiation effects or radiation necrosis (RN) remains challenging. Golden-angle radial sparse parallel (GRASP) dynamic contrast-enhanced MRI provides high spatial and temporal resolution to analyze tissue enhancement, which may differ between tumor progression (TP) and RN. OBJECTIVE:To investigate the utility of longitudinal GRASP MRI in distinguishing TP from RN after gamma knife stereotactic radiosurgery (SRS). METHODS:We retrospectively evaluated 48 patients with brain metastasis managed with SRS at our institution from 2013 to 2020 who had GRASP MRI before and at least once after SRS. TP (n = 16) was pathologically confirmed. RN (n = 16) was diagnosed on either resected tissue without evidence of tumor or on lesion resolution on follow-up. As a reference, we included a separate group of patients with non-small-cell lung cancer that showed favorable response with tumor control and without RN on subsequent imaging (n = 16). Mean contrast washin and washout slopes normalized to the superior sagittal sinus were compared between groups. Receiver operating characteristic analysis was performed to determine diagnostic performance. RESULTS:After SRS, progression showed a significantly steeper washin slope than RN on all 3 follow-up scans (scan 1: 0.29 ± 0.16 vs 0.18 ± 0.08, P = .021; scan 2: 0.35 ± 0.19 vs 0.18 ± 0.09, P = .004; scan 3: 0.32 ± 0.12 vs 0.17 ± 0.07, P = .002). No significant differences were found in the post-SRS washout slope. Post-SRS washin slope differentiated progression and RN with an area under the curve (AUC) of 0.74, a sensitivity of 75%, and a specificity of 69% on scan 1; an AUC of 0.85, a sensitivity of 92%, and a specificity of 69% on scan 2; and an AUC of 0.87, a sensitivity of 63%, and a specificity of 100% on scan 3. CONCLUSION:Longitudinal GRASP MRI may help to differentiate metastasis progression from RN.

BACKGROUND:Intraventricular metastases (IVMs) are uncommon, and their optimal management remains debatable. OBJECTIVE:The aim is to define the safety and efficacy of stereotactic radiosurgery (SRS) in the treatment of IVMs. METHODS:This retrospective, multicenter study included patients managed with SRS for IVMs. SRS-induced adverse events, local tumor or intracranial progression, and the frequency of new-onset hydrocephalus or leptomeningeal spread were documented. Analyses of variables related to patient neuroimaging or clinical outcomes were also performed. RESULTS:The cohort included 160 patients from 11 centers who underwent SRS for treatment of 1045 intracranial metastases, of which 196 were IVMs. The median survival from SRS was 10 months. Of the 154 patients and 190 IVMs with imaging follow-up, 94 patients (61%) experienced distant intracranial disease progression and 16 IVMs (8.4%) progressed locally. The 12- and 24-month local IVM control rates were 91.4% and 86.1%, respectively. Sixteen (10%) and 27 (17.5%) patients developed hydrocephalus and leptomeningeal dissemination post-SRS, respectively. Adverse radiation effects were documented in 24 patients (15%). Eleven patients (6.9%) died because of intracranial disease progression. CONCLUSION/CONCLUSIONS:SRS is an effective treatment option for IVMs, with a local IVM control rate comparable with SRS for parenchymal brain metastases. Leptomeningeal spread and hydrocephalus in patients with IVM occur in a minority of patients, but these patients warrant careful follow-up to detect these changes.

BACKGROUND:The development of accurate machine learning algorithms requires sufficient quantities of diverse data. This poses a challenge in health care because of the sensitive and siloed nature of biomedical information. Decentralized algorithms through federated learning (FL) avoid data aggregation by instead distributing algorithms to the data before centrally updating one global model. OBJECTIVE:To establish a multicenter collaboration and assess the feasibility of using FL to train machine learning models for intracranial hemorrhage (ICH) detection without sharing data between sites. METHODS:Five neurosurgery departments across the United States collaborated to establish a federated network and train a convolutional neural network to detect ICH on computed tomography scans. The global FL model was benchmarked against a standard, centrally trained model using a held-out data set and was compared against locally trained models using site data. RESULTS:A federated network of practicing neurosurgeon scientists was successfully initiated to train a model for predicting ICH. The FL model achieved an area under the ROC curve of 0.9487 (95% CI 0.9471-0.9503) when predicting all subtypes of ICH compared with a benchmark (non-FL) area under the ROC curve of 0.9753 (95% CI 0.9742-0.9764), although performance varied by subtype. The FL model consistently achieved top three performance when validated on any site's data, suggesting improved generalizability. A qualitative survey described the experience of participants in the federated network. CONCLUSION/CONCLUSIONS:This study demonstrates the feasibility of implementing a federated network for multi-institutional collaboration among clinicians and using FL to conduct machine learning research, thereby opening a new paradigm for neurosurgical collaboration.

BACKGROUND: While most meningiomas are considered benign tumors, a subset of these tumors are characterized by a more aggressive clinical course and require multimodal treatment. Beyond surgical and radiotherapeutic options, there are no effective medical treatments available. Somatostatin receptor 2 (SSTR2) is expressed by the majority of meningiomas. 177Lu-DOTATATE is a SSTR2-targeting radionuclide that has been successful in neuroendocrine tumors. Here we report the results of the interim analysis of an ongoing clinical trial (NCT03971461) that is evaluating the effect of 177Lu-DOTATATE in treating progressive intracranial meningiomas.
METHOD(S): In this Simon two-stage design phase II study, adults with advanced intracranial meningiomas received 177Lu-DOTATATE 7.4 GBq (200 mCi) every eight weeks for four doses. 68Ga-DOTATATE PET-MRI was performed before and at the end of treatment. The primary endpoint was progression-free survival at 6 months (PFS-6). Correlative studies evaluated the association of PFS-6, objective response rate, progression-free survival, overall survival with radiographic tumor measurements, 68Ga-DOTATATE uptake on PET-MRI, SSTR2 expression in tumor, and meningioma methylation subclass.
RESULT(S): Fourteen patients (F = 11, M = 3) with progressive meningiomas (WHO I = 3, II = 10, III = 1) have been enrolled. Median age was 63.1 (range 49-78) years. All patients previously underwent tumor resection and at least one course of radiation. Treatment with 177Lu-DOTATATE was well tolerated, no treatment-limiting toxicities were observed. Six of 14 patients (42%) achieved PFS-6. Radiographically, all six patients had achieved Stable Disease. A functional alteration of tumoral SSTR2 expression by 68Ga-DOTATATE PET-MR imaging was observed in three patients.
CONCLUSION(S): Treatment with SSTR2- targeting 177Lu-DOTATATE is well tolerated. In this interim analysis, six of 14 patients achieved PFS-6. This exceeds the predefined threshold to continue to stage two of this study. This clinical trial is now open to patient enrollment at two study sites in the US

OBJECTIVE: Immune checkpoint inhibitors (ICI) and stereotactic radiosurgery (SRS) are commonly utilized in the management of brain metastases. Treatment-related imaging changes (TRIC) are a frequently observed clinical manifestation and are commonly classified as radiographic radiation necrosis. However, these findings are not well characterized and may predict for response to SRS and ICI.
METHOD(S): The diagnosis-specific graded prognostic assessment was used to guide variable selection. TRIC were determined based upon MRI, PET/CT, or MR spectroscopy and a consensus by local clinical providers was required.
RESULT(S): The analysis included 697 patients with 4,536 brain metastases across 11 institutions in 4 countries. The median follow-up after SRS was 13.6 months. The median age was 66 years, 54.1% of patients were male, and 57.3%, 36.3%, and 6.4% were non-small cell lung cancer, melanoma, and renal cell carcinoma (RCC) histology, respectively. TRIC were observed in 9.8%. On univariable analysis, Karnofsky Performance Status (KPS) (hazard ratio [HR]: 0.98; p < 0.001), presence of TRIC (HR: 0.67; p = 0.03), female sex (HR: 0.67; p < 0.001), and prior resection (HR: 0.60; p = 0.03) were associated with improved OS. On multivariable analysis, KPS (HR: 0.98; p < 0.001) and the presence of TRIC (HR: 0.66; p = 0.03) were associated with improved OS. A V12 Gy >= 10 cm3 (Odds Ratio [OR]: 2.78; p < 0.001), prior whole brain radiation therapy (OR: 3.46; p = 0.006), and RCC histology (OR: 3.10; p = 0.01) were associated with an increased probability of developing TRIC. The median OS in patients with and without TRIC was 29.0 and 23.1 months, respectively (log-rank p = 0.03).
CONCLUSION(S): TRIC following ICI and SRS are associated with a median OS benefit of approximately 6 months. Further prospective study is warranted to further elucidate the role and etiology of this common clinical scenario

Purpose/Objective(s): Stereotactic radiosurgery (SRS) and immune checkpoint inhibitors (ICI) are highly effective treatments for brain metastases (BrM), particularly when these therapies are administered concurrently. However, there are limited data reporting the risk of radiation necrosis (RN) in this setting. Materials/Methods: Patients with BrM from primary non-small cell lung cancer, renal cell carcinoma, or melanoma treated with SRS and ICI were considered. Recursive partitioning analysis (RPA) was utilized for model development, and a loop of potential models was analyzed, with the highest-fidelity model selected.
Result(s): Six hundred fifty-seven patients with 4,182 BrM across 11 international institutions were analyzed. Rates of RN and symptomatic RN (SRN) for all patients were 10% and 6.8%, respectively. The highest-fidelity models consistently identified V12 Gy as the dominant variable predictive of RN. Three risk groups were identified using V12 Gy: (1) < 12 cm³; (2) 20 cm³ <= V12 Gy >= 12 cm³; (3) > 20 cm³. Odds ratios for RN and SRN with cases of V12 Gy >= 12 cm³ compared with < 12 cm³ were 3.05 (p < 0.001) and 3.72 (p < 0.001), respectively. Rates of RN and SRN are presented in the table below. Concurrent ICI use rates were equivalent among these resulting groups, and the addition of concurrent ICI use did not improve the model's fidelity. Using RPA, 80% of the highest-fidelity models failed to incorporate concurrent ICI as a predictive variable. Even after exclusion of V12 Gy as a candidate variable, concurrent ICI remained unused in 85% of the highest-fidelity models. These models yielded 94% accuracy for the validation set and 92% accuracy for the test set.
Conclusion(s): Utilization of SRS and ICI results in a low risk of RN and SRN. This risk is not increased when ICI and SRS are administered concurrently. Therefore, ICI can safely be administered within 4-weeks of SRS. In patients receiving SRS and ICI, three risk groups based on V12 Gy were identified, which clinicians may consider to further reduce rates of RN.
Copyright

BACKGROUND:Melanoma brain metastases are commonly treated with stereotactic radiosurgery (SRS) and immune checkpoint inhibitors (ICIs). However, the toxicity of these 2 treatments is largely unknown when administered concurrently. OBJECTIVE:To evaluate the risk of radiation necrosis (RN) with concurrent and nonconcurrent SRS and ICIs. METHODS:The guidelines from the Strengthening the Reporting of Observational Studies in Epidemiology checklist were used. Inverse probability of treatment weighting, univariable and multivariable logistic regression, and the Kaplan-Meier method was utilized. RESULTS:There were 203 patients with 1388 brain metastases across 11 international institutions in 4 countries with a median follow-up of 15.6 months. The rates of symptomatic RN were 9.4% and 8.2% in the concurrent and nonconcurrent groups, respectively ( P =.766). On multivariable logistic regression, V12 ≥ 10 cm 3 (odds ratio [OR]: 2.76; P =.006) and presence of BRAF mutation (OR: 2.20; P =.040) were associated with an increased risk of developing symptomatic RN; the use of concurrent over nonconcurrent therapy was not associated with an increased risk (OR: 1.06; P =.877). There were 20 grade 3 toxic events reported, and no grade 4 events reported. One patient experienced a grade 5 intracranial hemorrhage. The median overall survival was 36.1 and 19.8 months for the concurrent and nonconcurrent groups (log-rank P =.051), respectively. CONCLUSION/CONCLUSIONS:Concurrent administration of ICIs and SRS are not associated with an increased risk of RN. Tumors harboring BRAF mutation, or perhaps prior exposure to targeted agents, may increase this risk. Radiosurgical optimization to maintain V12 < 10 cm 3 is a potential strategy to reduce the risk of RN.

The fundamental challenge in machine learning is ensuring that trained models generalize well to unseen data. We developed a general technique for ameliorating the effect of dataset shift using generative adversarial networks (GANs) on a dataset of 149,298 handwritten digits and dataset of 868,549 chest radiographs obtained from four academic medical centers. Efficacy was assessed by comparing area under the curve (AUC) pre- and post-adaptation. On the digit recognition task, the baseline CNN achieved an average internal test AUC of 99.87% (95% CI, 99.87-99.87%), which decreased to an average external test AUC of 91.85% (95% CI, 91.82-91.88%), with an average salvage of 35% from baseline upon adaptation. On the lung pathology classification task, the baseline CNN achieved an average internal test AUC of 78.07% (95% CI, 77.97-78.17%) and an average external test AUC of 71.43% (95% CI, 71.32-71.60%), with a salvage of 25% from baseline upon adaptation. Adversarial domain adaptation leads to improved model performance on radiographic data derived from multiple out-of-sample healthcare populations. This work can be applied to other medical imaging domains to help shape the deployment toolkit of machine learning in medicine.

BACKGROUND:Previous studies comparing hearing outcomes in patients managed with stereotactic radiosurgery (SRS) and a watch-and-wait strategy were limited by small sample sizes that prevented controlling for potential confounders, including initial hearing status, tumor size, and age. OBJECTIVE:To compare hearing outcomes for patients with vestibular schwannomas (VS) managed with observation and SRS while controlling for confounders with propensity score matching. METHODS:Propensity score matching was used to compare 198 patients with unilateral VS with initial serviceable hearing (99 treated with SRS and 99 managed with observation alone) and 116 with initial class A hearing (58 managed with SRS and 58 with observation), matched by initial hearing status, tumor volume, age, and sex. Kaplan-Meier survival methods were used to compare risk of losing class A and serviceable hearing. RESULTS:Between patients with VS managed with SRS or observation alone, there was no significant difference in loss of class A hearing (median time 27.2 months, 95% CI 16.8-43.4, and 29.2 months, 95% CI 20.4-62.5, P = .88) or serviceable hearing (median time 37.7 months, 95% CI 25.7-58.4, and 48.8 months, 95% CI 38.4-86.3, P = .18). For SRS patients, increasing mean cochlear dose was not related to loss of class A hearing (hazard ratio 1.3, P = .17) but was associated with increasing risk of serviceable hearing loss (hazard ratio of 1.5 per increase in Gy, P = .017). CONCLUSION/CONCLUSIONS:When controlling for potential confounders, there was no significant difference in loss of class A or serviceable hearing between patients managed with SRS or with observation alone.

BACKGROUND:Late pathology after vestibular schwannoma radiosurgery is uncommon. The authors presented a case of a resected hemorrhagic mass 13 years after radiosurgery, when no residual tumor was found. OBSERVATIONS/METHODS:A 56-year-old man with multiple comorbidities, including myelodysplastic syndrome cirrhosis, received Gamma Knife surgery for a left vestibular schwannoma. After 11 years of stable imaging assessments, the lesion showed gradual growth until a syncopal event occurred 2 years later, accompanied by progressive facial weakness and evidence of intralesional hemorrhage, which led to resection. However, histopathological analysis of the resected specimen showed hemorrhage and reactive tissue but no definitive residual tumor. LESSONS/CONCLUSIONS:This case demonstrated histopathological evidence for the role of radiosurgery in complete elimination of tumor tissue. Radiosurgery for vestibular schwannoma carries a rare risk for intralesional hemorrhage in select patients.