Faculty Bibliography

BACKGROUND AND OBJECTIVES/OBJECTIVE:Advances in targeted therapies and wider application of stereotactic radiosurgery (SRS) have redefined outcomes of patients with brain metastases. Under modern treatment paradigms, there remains limited characterization of which aspects of disease drive demise and in what frequencies. This study aims to characterize the primary causes of terminal decline and evaluate differences in underlying intracranial tumor dynamics in patients with metastatic brain cancer. These fundamental details may help guide management, patient counseling, and research priorities. METHODS:Using NYUMets-Brain-the largest, longitudinal, real-world, open data set of patients with brain metastases-patients treated at New York University Langone Health between 2012 and 2021 with SRS were evaluated. A review of electronic health records allowed for the determination of a primary cause of death in patients who died during the study period. Causes were classified in mutually exclusive, but collectively exhaustive, categories. Multilevel models evaluated for differences in dynamics of intracranial tumors, including changes in volume and number. RESULTS:Of 439 patients with end-of-life data, 73.1% died secondary to systemic disease, 10.3% died secondary to central nervous system (CNS) disease, and 16.6% died because of other causes. CNS deaths were driven by acute increases in intracranial pressure (11%), development of focal neurological deficits (18%), treatment-resistant seizures (11%), and global decline driven by increased intracranial tumor burden (60%). Rate of influx of new intracranial tumors was almost twice as high in patients who died compared with those who survived (P < .001), but there was no difference in rates of volume change per intracranial tumor (P = .95). CONCLUSION/CONCLUSIONS:Most patients with brain metastases die secondary to systemic disease progression. For patients who die because of neurological disease, tumor dynamics and cause of death mechanisms indicate that the primary driver of decline for many may be unchecked systemic disease with unrelenting spread of new tumors to the CNS rather than failure of local growth control.

Physicians make critical time-constrained decisions every day. Clinical predictive models can help physicians and administrators make decisions by forecasting clinical and operational events. Existing structured data-based clinical predictive models have limited use in everyday practice owing to complexity in data processing, as well as model development and deployment^1-3. Here we show that unstructured clinical notes from the electronic health record can enable the training of clinical language models, which can be used as all-purpose clinical predictive engines with low-resistance development and deployment. Our approach leverages recent advances in natural language processing^4,5 to train a large language model for medical language (NYUTron) and subsequently fine-tune it across a wide range of clinical and operational predictive tasks. We evaluated our approach within our health system for five such tasks: 30-day all-cause readmission prediction, in-hospital mortality prediction, comorbidity index prediction, length of stay prediction, and insurance denial prediction. We show that NYUTron has an area under the curve (AUC) of 78.7-94.9%, with an improvement of 5.36-14.7% in the AUC compared with traditional models. We additionally demonstrate the benefits of pretraining with clinical text, the potential for increasing generalizability to different sites through fine-tuning and the full deployment of our system in a prospective, single-arm trial. These results show the potential for using clinical language models in medicine to read alongside physicians and provide guidance at the point of care.

INTRODUCTION/BACKGROUND:Historical reservations regarding radiosurgery (SRS) for small-cell-lung-cancer (SCLC) brain metastases (BrM) include concerns for short-interval/diffuse CNS-progression, poor prognoses, and increased neurological mortality specific to SCLC histology. We compared SRS outcomes for SCLC and non-small-cell-lung-cancer (NSCLC) where SRS is well established. METHODS:Multicenter first-line SRS outcomes for SCLC and NSCLC from 2000-2022 were retrospectively collected (N=892-SCLC/N=4,785-NSCLC). Data from the prospective JLGK0901 SRS trial were analyzed as a comparison cohort (N=98-SCLC/N=794-NSCLC). OS and CNS-progression were analyzed using Cox-Proportional-Hazard and Fine-Gray models, respectively, with multivariable (MV) adjustment (including age/sex/performance-status/year/extracranial disease/BrM-number/BrM-volume). Mutation-stratified analyses were performed in propensity score-matched (PSM) retrospective cohorts of EGFR/ALK-positive-NSCLC, mutation-negative-NSCLC, and SCLC. RESULTS:OS was superior with NSCLC over SCLC in the retrospective dataset (median-OS, 10.5 vs 8.6 months, MV-p<0.001) and JLGK0901. Hazard estimates for first CNS-progression favoring NSCLC were similar in both datasets but reached significance in the retrospective dataset only (MV-HR:0.82 [95%-CI:0.73-0.92], p=0.001). In the PSM cohorts, there were continued OS advantages for NSCLC (median-OS, 23.7 [EGFR/ALK-positive-NSCLC] vs 13.6 [mutation-negative-NSCLC] vs 10.4 months [SCLC], pairwise-p-values<0.001), but no significant differences in CNS-progression. Neurological mortality and number of lesions at CNS-progression were similar for NSCLC and SCLC patients. Leptomeningeal-progression was increased in NSCLC patients in the retrospective dataset only (MV-HR:1.61 [95%-CI:1.14-2.26], p=0.007). CONCLUSION/CONCLUSIONS:After SRS, SCLC was associated with shorter OS compared to NSCLC. CNS progression occurred earlier in SCLC overall but was similar in patients matched on baseline characteristics. Neurological mortality, lesions at CNS-progression, and leptomeningeal-progression were comparable. These findings may better inform clinical decision-making for SCLC patients.

BACKGROUND AND OBJECTIVES/OBJECTIVE:Meningiomas in children are uncommon, with distinct characteristics that set them apart from their adult counterparts. The existing evidence for stereotactic radiosurgery (SRS) in this patient population is limited to only case series. The objective of this study was to evaluate the safety and efficacy of SRS in managing pediatric meningiomas. METHODS:Children and adolescents who had been treated for meningioma with single-fraction SRS were included in this retrospective, multicenter study. The assessment included local tumor control, any complications related to the tumor or SRS, and the emergence of new neurological deficits after SRS. RESULTS:The cohort included 57 patients (male-to-female ratio 1.6:1) with a mean age of 14.4 years who were managed with single-fraction SRS for 78 meningiomas. The median radiological and clinical follow-up periods were 69 months (range, 6-268) and 71 months (range, 6-268), respectively. At the last follow-up, tumor control (tumor stability and regression) was achieved in 69 (85.9%) tumors. Post-SRS, new neurological deficits occurred in 2 (3.5%) patients. Adverse radiation effects occurred in 5 (8.8%) patients. A de novo aneurysm was observed in a patient 69 months after SRS. CONCLUSION/CONCLUSIONS:SRS seems to be a safe and effective up-front or adjuvant treatment option for surgically inaccessible, recurrent, or residual pediatric meningiomas.

BACKGROUND AND OBJECTIVES/OBJECTIVE:Stereotactic radiosurgery (SRS) of larger arteriovenous malformations (AVM) is associated with an elevated incidence of adverse radiation effects (ARE). To date, volume-response and dose-response models have been used to predict such effects. To understand radiological outcomes and their hemodynamic effects on the regional brain. METHODS:A retrospective analysis was conducted at our institution using a prospective registry of patients managed between 2014 and 2020. We included patients with AVM with a nidus larger than 5 cc who received either single-session or volume-staged Gamma Knife radiosurgery. AVM volume changes, volumes of parenchymal response, and obliteration were analyzed and correlated with transit times and diameters of feeding arteries and draining veins. RESULTS:Sixteen patients underwent single-session SRS, and 9 patients underwent volume-staged SRS. The average AVM volume was 12.6 cc (5.5-23). The AVM locations were predominantly lobar (80%) and 17 (68%) were in critical locations. The mean margin dose was 17.2 Gy (15-21), and the median V12Gy was 25.5 cc. Fourteen (56%) AVMs had a transit time shorter than 1 second. The median vein-artery ratio (sum diameter of the veins/sum diameter of feeding arteries) was 1.63 (range, 0.60-4.19). Asymptomatic parenchymal effects were detected in 13 (52%) patients and were symptomatic in 4 (16%) patients. The median time to ARE was 12 months (95% CI 7.6-16.4). On univariate analysis, significant predictors of ARE were lower vein-artery ratio (P = .024), longer transit time (P = .05), higher mean dose (P = .028), and higher D95 (P = .036). CONCLUSION/CONCLUSIONS:Transit times and vessel diameters are valuable predictors of the subsequent parenchymal response after SRS. A more quantitative understanding of blood flow is critical for predicting the effects on the regional brain after AVM radiosurgery.

BACKGROUND:Long-term follow-up using volumetric measurement could significantly assist in the management of vestibular schwannomas (VS). Manual segmentation of VS from MRI for treatment planning and follow-up assessment is labor-intensive and time-consuming. This study aims to develop a deep learning technique to fully automatically segment VS from MRI. METHODS:This study retrospectively analyzed MRI data of 737 patients who received gamma knife radiosurgery for VS. Treatment planning T1-weighted isotropic MR and manually contoured gross tumor volumes (GTV) were used for model development. A 3D convolutional neural network (CNN) was built on ResNet blocks. Spatial attenuation and deep supervision modules were integrated in each decoder level to enhance the training for the small tumor volume on brain MRI. The model was trained and tested on 587 and 150 patient data, respectively, from this institution (n = 495) and a publicly available dataset (n = 242). The model performance were assessed by the Dice similarity coefficient (DSC), 95% Hausdorff distance (HD95), average symmetric surface (ASSD) and relative absolute volume difference (RAVD) of the model segmentation results against the GTVs. RESULTS:Measured on combined testing data from two institutions, the proposed method achieved mean DSC of 0.91 ± 0.08, ASSD of 0.3 ± 0.4 mm, HD95 of 1.3 ± 1.6 mm, and RAVD of 0.09 ± 0.15. The DSCs were 0.91 ± 0.09 and 0.92 ± 0.06 on 100 testing patients of this institution and 50 of the public data, respectively. CONCLUSIONS:A CNN model was developed for fully automated segmentation of VS on T1-Weighted isotropic MRI. The model achieved good performance compared with physician clinical delineations on a sizeable dataset from two institutions. The proposed method potentially facilitates clinical workflow of radiosurgery for VS patient management.

BACKGROUND:Percutaneous rhizotomy of the trigeminal nerve is a common surgery to manage medically refractory trigeminal neuralgia. Traditionally, these procedures have been performed based on anatomic landmarks with fluoroscopic guidance. Augmented reality (AR) relays virtual content on the real world and has the potential to improve localization of surgical targets based on preoperative imaging. OBJECTIVE:To study the potential application and benefits of AR as an adjunct to traditional fluoroscopy-guided glycerol rhizotomy (GR). METHODS:We used traditional fluoroscopy-guided percutaneous GR technique as previously described, performed under general anesthesia. Anatomic registration to the Medivis SurgicalAR system was performed based on the patient's preoperative computerized tomography, and the surgeon was equipped with the system's AR goggles. AR was used as an adjunct to fluoroscopy for trajectory planning to place a spinal needle into the medial aspect of the foramen ovale. RESULTS:A 50-year-old woman with multiple sclerosis-related right-sided classical trigeminal neuralgia had persistent pain, refractory to medications, previous gamma knife stereotactic radiosurgery, and percutaneous radiofrequency rhizotomy performed elsewhere. The patient underwent AR-assisted fluoroscopy-guided percutaneous GR. The needle was placed into the right trigeminal cistern within seconds. She was discharged home after a few hours of observation with no complications and reported pain relief. CONCLUSION/CONCLUSIONS:AR-assisted percutaneous rhizotomy may enhance the learning curve of these types of procedures and decrease surgery duration and radiation exposure. This allowed rapid and correct placement of a spinal needle through the foramen ovale.