Search

Abstract Purpose Artificial intelligence-based tools can be leveraged to improve detection and segmentation of brain metastases for stereotactic radiosurgery (SRS). VBrain by Vysioneer Inc. is a deep learning algorithm with recent FDA clearance to assist in brain tumor contouring. We aimed to assess the performance of this tool by various demographic and clinical characteristics among patients with brain metastases treated with SRS. Materials and methods We randomly selected 100 patients with brain metastases who underwent initial SRS on the CyberKnife from 2017 to 2020 at a single institution. Cases with resection cavities were excluded from the analysis. Computed tomography (CT) and axial T1-weighted post-contrast magnetic resonance (MR) image data were extracted for each patient and uploaded to VBrain. A brain metastasis was considered “detected” when the VBrain- “predicted” contours overlapped with the corresponding physician contours (“ground-truth” contours). We evaluated performance of VBrain against ground-truth contours using the following metrics: lesion-wise Dice similarity coefficient (DSC), lesion-wise average Hausdorff distance (AVD), false positive count (FP), and lesion-wise sensitivity (%). Kruskal–Wallis tests were performed to assess the relationships between patient characteristics including sex, race, primary histology, age, and size and number of brain metastases, and performance metrics such as DSC, AVD, FP, and sensitivity. Results We analyzed 100 patients with 435 intact brain metastases treated with SRS. Our cohort consisted of patients with a median number of 2 brain metastases (range: 1 to 52), median age of 69 (range: 19 to 91), and 50% male and 50% female patients. The primary site breakdown was 56% lung, 10% melanoma, 9% breast, 8% gynecological, 5% renal, 4% gastrointestinal, 2% sarcoma, and 6% other, while the race breakdown was 60% White, 18% Asian, 3% Black/African American, 2% Native Hawaiian or other Pacific Islander, and 17% other/unknown/not reported. The median tumor size was 0.112 c.c. (range: 0.010–26.475 c.c.). We found mean lesion-wise DSC to be 0.723, mean lesion-wise AVD to be 7.34% of lesion size (0.704 mm), mean FP count to be 0.72 tumors per case, and lesion-wise sensitivity to be 89.30% for all lesions. Moreover, mean sensitivity was found to be 99.07%, 97.59%, and 96.23% for lesions with diameter equal to and greater than 10 mm, 7.5 mm, and 5 mm, respectively. No other significant differences in performance metrics were observed across demographic or clinical characteristic groups. Conclusion In this study, a commercial deep learning algorithm showed promising results in segmenting brain metastases, with 96.23% sensitivity for metastases with diameters of 5 mm or higher. As the software is an assistive AI, future work of VBrain integration into the clinical workflow can provide further clinical and research insights.

ObjectiveThe 2-year incidence of brain metastases (BrMs) in stage III non-small lung cell cancer (NSCLC) has been estimated to be around 30%. However, recent clinical trials have demonstrated considerably lower BrMs rates in this patient population. In this study, we aimed to review the real-world incidence, surveillance, and treatment patterns of BrMs in stage III NSCLC.Materials and methodsUsing a retrospective single-center study design, we identified patients with stage III NSCLC who received radiation with curative intent over a 10-year period. Outcome variables included BrMs incidence, overall survival (OS), and survival from date of BrMs. Additionally, we assessed patterns of BrMs surveillance in stage III NSCLC and treatment.ResultsWe identified a total of 279 stage III NSCLC patients, of which 160 with adequate records were included in the final analyses [adenocarcinoma (n = 96), squamous cell carcinoma (n = 53), other histology subtype (n = 11)]. The median OS for the entire cohort was 41 months (95% CI, 28-53), while the median time from BrMs to death was 19 months (95% CI, 9-21). Twenty-three patients (14.4%) received planned surveillance brain MRIs at 6, 12, and 24 months after completion of treatment. The remaining 137 patients (85.6%) received brain MRIs at systemic recurrence (restaging) or when neurologically symptomatic. A total of 37 patients (23%) developed BrMs, with a 2-year cumulative BrMs incidence of 17% (95% CI, 11-23). A higher incidence of BrMs was identified in patients with adenocarcinoma relative to those with squamous cell carcinoma (p < 0.01). Similarly, a higher 2-year BrMs incidence was observed in patients who received planned surveillance brain MRI relative to those who did not, although statistical significance was not reached. Stereotactic radiosurgery (SRS) treated 29 of BrMs patients (78.4%) and was preferred over WBRT, which treated only 3 patients (8.1%).ConclusionsAt our center, BrMs incidence in stage III NSCLC patients was lower than historically reported but notably higher than the incidence described in recent clinical trials. Routine BrMs surveillance potentially allows earlier detection of asymptomatic BrMs. However, asymptomatic BrMs were mostly detected on restaging MRI at the time of recurrence.

Purpose: Postoperative stereotactic radiosurgery (SRS) is associated with up to 30% risk of subsequent leptomeningeal disease (LMD). Radiographic patterns of LMD (classical sugarcoating [cLMD] vs. nodular [nLMD]) in this setting has been shown to be prognostic. However, the association of these findings with neurologic death (ND) is not well described. Methods and Materials: The records for patients with brain metastases who underwent surgical resection and adjunctive SRS to 1 lesion (SRS to other intact lesions was allowed) and subsequently developed LMD were combined from 7 tertiary care centers. Salvage radiation therapy (RT) for LMD was categorized according to use of whole-brain versus focal cranial RT. Results: The study cohort included 125 patients with known cause of death. The ND rate in these patients was 79%, and the rate in patients who underwent LMD salvage treatment (n = 107) was 76%. Univariate logistic regression demonstrated radiographic pattern of LMD (cLMD vs. nLMD, odds ratio: 2.9; P = .04) and second LMD failure after salvage treatment (odds ratio: 3.9; P = .02) as significantly associated with ND. The ND rate was 86% for cLMD versus 68% for nLMD. Whole-brain RT was used in 95% of patients with cLMD and 52% with nLMD. In the nLMD cohort (n = 58), there was no difference in ND rate based on type of salvage RT (whole-brain RT: 67% vs. focal cranial RT: 68%, P = .92). Conclusions: LMD after surgery and SRS for brain metastases is a clinically significant event with high rates of ND. Classical LMD pattern (vs. nodular) and second LMD failure after salvage treatment were significantly associated with a higher risk of ND. Patients with nLMD treated with salvage focal cranial RT did not have higher ND rates compared with WBRT. Methods to decrease LMD and the subsequent high risk of ND in this setting warrant further investigation.

Introduction: Gliosarcomas are clinically aggressive tumors, histologically distinct from glioblastoma. Data regarding the impact of extent of resection and post-operative adjuvant therapy on gliosarcoma outcomes are limited.Methods: Patients with histologically confirmed gliosarcoma diagnosed between 1999 and 2019 were identified. Clinical, molecular, and radiographic data were assembled based on historical records. Comparisons of categorical variables used Pearson's Chi-square and Fisher's exact test while continuous values were compared using the Wilcoxon signed-rank test. Survival comparisons were assessed using Kaplan-Meier statistics and Cox regressions.Results: Seventy-one gliosarcoma patients were identified. Secondary gliosarcoma was not associated with worse survival when compared to recurrent primary gliosarcoma (median survival 9.8 [3.8 to 21.0] months vs. 7.6 [1.0 to 35.7], p = 0.7493). On multivariable analysis, receipt of temozolomide (HR = 0.02, 95% CI 0.001–0.21) and achievement of gross total resection (GTR; HR = 0.13, 95% CI 0.02–0.77) were independently prognostic for improved progression-free survival (PFS) while only receipt of temozolomide was independently associated with extended overall survival (OS) (HR = 0.03, 95% CI 0.001–0.89). In patients receiving surgical resection followed by radiotherapy and concomitant temozolomide, achievement of GTR was significantly associated with improved PFS (median 32.97 [7.1–79.6] months vs. 5.45 [1.8–26.3], p = 0.0092) and OS (median 56.73 months [7.8–104.5] vs. 14.83 [3.8 to 29.1], p = 0.0252).Conclusion: Multimodal therapy is associated with improved survival in gliosarcoma. Even in patients receiving aggressive post-operative multimodal management, total surgical removal of macroscopic disease remains important for optimal outcomes.