Faculty Bibliography

BACKGROUND:Pituitary neuroendocrine tumors (PitNETs) are the most common intracranial neuroendocrine tumors. PitNETs can be challenging to classify, and current recommendations include a large immunohistochemical panel to differentiate among 14 WHO-recognized categories. METHODS:In this study, we analyzed clinical, immunohistochemical and DNA methylation data of 118 PitNETs to develop a clinico-molecular approach to classifying PitNETs and identify epigenetic classes. RESULTS:CNS DNA methylation classifier has an excellent performance in recognizing PitNETs and distinguishing the three lineages when the calibrated score is ≥0.3. Unsupervised DNA methylation analysis separated PitNETs into two major clusters. The first was composed of silent gonadotrophs, which form a biologically distinct group of PitNETs characterized by clinical silencing, weak hormonal expression on immunohistochemistry, and simple copy number profile. The second major cluster was composed of corticotrophs and Pit1 lineage PitNETs, which could be further classified using DNA methylation into distinct subclusters that corresponded to clinically functioning and silent tumors and are consistent with transcription factor expression. Analysis of promoter methylation patterns correlated with lineage for corticotrophs and Pit1 lineage subtypes. However, the gonadotrophic genes did not show a distinct promoter methylation pattern in gonadotroph tumors compared to other lineages. Promoter of the NR5A1 gene, which encodes SF1, was hypermethylated across all PitNETs clinical and molecular subtypes including gonadotrophs with strong SF1 protein expression indicating alternative epigenetic regulation. CONCLUSION/CONCLUSIONS:Our findings suggest that classification of PitNETs may benefit from DNA methylation for clinicopathological stratification.

OBJECTIVES/OBJECTIVE:To assess a decade of experience of treating patients with high hearing risk cerebellopontine angle (CPA) epidermoid lesions and examine factors influencing postoperative outcomes, particularly hearing preservation. STUDY DESIGN/METHODS:Retrospective chart review. SETTING/METHODS:Single tertiary-referral center. PATIENTS/METHODS:Adults with CPA epidermoid lesions who presented with hearing loss or evidence of lesion involving vestibulocochlear nerve. INTERVENTIONS/METHODS:The studied intervention was microsurgical resection. MAIN OUTCOME MEASURES/METHODS:Main outcome measures included extent of resection, hearing preservation rate for patients with postoperative audiograms, and disease progression. RESULTS:Twenty-three adults with an average tumor volume of 15.63 ± 16.2 cm3 were included. Five lesions (22%) involved the full internal auditory canal (IAC), 11 (48%) had partial involvement, and 5 (22%) were IAC sparing. Most patients with IAC involvement (88%) had circumferential invasion of the canal. Patients underwent either a retrosigmoid (18, 79%) or combined retrolabyrinthine transpetrosal approach (5, 22%), and gross total resection was achieved in most cases (13, 57%). Of 12 patients with postoperative audiograms, 10 (83%) had preoperative hearing preserved. There was no statistically significant change in hearing scores with treatment based on preoperative extent of IAC involvement. Ten patients (43%) had residual lesions postoperatively, and 6 exhibited progression. One patient ultimately required reoperation 6 years after initial surgery. CONCLUSIONS:Preoperative hearing was preserved in the majority of the patients who underwent resection of CPA epidermoids via a retrosigmoid or transpetrosal approach. CPA epidermoids often invaded the IAC; however, degree of invasion was not associated with hearing outcomes.

BACKGROUND AND OBJECTIVES/OBJECTIVE:Hydrocephalus after Gamma Knife® stereotactic radiosurgery (SRS) for vestibular schwannomas is a rare but manageable occurrence. Most series report post-SRS communicating hydrocephalus in about 1% of patients, thought to be related to a release of proteinaceous substances into the cerebrospinal fluid. While larger tumor size and older patient age have been associated with post-SRS hydrocephalus, the influence of baseline ventricular anatomy on hydrocephalus risk remains poorly defined. METHODS:A single-institution retrospective cohort study examining patients who developed symptomatic communicating hydrocephalus after undergoing Gamma Knife® SRS for unilateral vestibular schwannomas from 2011 to 2021 was performed. Patients with prior hydrocephalus and cerebrospinal fluid diversion or prior surgical resection were excluded. Baseline tumor volume, third ventricle width, and Evans Index (EI)-maximum width of the frontal horns of the lateral ventricles/maximum internal diameter of the skull-were measured on axial postcontrast T1-weighted magnetic resonance imaging. RESULTS:A total of 378 patients met the inclusion criteria; 14 patients (3.7%) developed symptomatic communicating hydrocephalus and 10 patients (2.6%) underwent shunt placement and 4 patients (1.1%) were observed with milder symptoms. The median age of patients who developed hydrocephalus was 69 years (IQR, 67-72) and for patients younger than age 65 years, the risk was 1%. For tumor volumes <1 cm3, the risk of requiring shunting was 1.2%. The odds of developing symptomatic hydrocephalus were 5.0 and 7.7 times higher in association with a baseline EI > 0.28 (P = .024) and tumor volume >3 cm3 (P = .007), respectively, in multivariate analysis. Fourth ventricle distortion on pre-SRS imaging was significantly associated with hydrocephalus incidence (P < .001). CONCLUSION/CONCLUSIONS:Patients with vestibular schwannoma with higher baseline EI, larger tumor volumes, and fourth ventricle deformation are at increased odds of developing post-SRS hydrocephalus. These patients should be counseled regarding risk of hydrocephalus and carefully monitored after SRS.

BACKGROUND:The diagnosis and treatment of tumors often depend on molecular-genetic data. However, rapid and iterative access to molecular data is not currently feasible during surgery, complicating intraoperative diagnosis and precluding measurement of tumor cell burdens at surgical margins to guide resections. METHODS:Here, we introduce Ultra-Rapid droplet digital PCR (UR-ddPCR), a technology that achieves the fastest measurement, to date, of mutation burdens in tissue samples, from tissue to result in 15 min. Our workflow substantially reduces the time from tissue biopsy to molecular diagnosis and provides a highly accurate means of quantifying residual tumor infiltration at surgical margins. FINDINGS/RESULTS: = 0.995). CONCLUSIONS:The technology and workflow developed here enable intraoperative molecular-genetic assays with unprecedented speed and sensitivity. We anticipate that our method will facilitate novel point-of-care diagnostics and molecularly guided surgeries that improve clinical outcomes. FUNDING/BACKGROUND:This study was funded by the National Institutes of Health and NYU Grossman School of Medicine institutional funds. Reagents and instruments were provided in kind by Bio-Rad.

OBJECTIVE:Limited evidence guides the optimal timing of treatment after the detection of tumor growth during the observation of sporadic vestibular schwannoma (VS). The current work aimed to inform the timing of radiosurgical intervention based on an analysis of patient outcomes among those who ultimately failed stereotactic radiosurgery (SRS) and underwent salvage microsurgery. STUDY DESIGN/METHODS:A historical cohort study. SETTING/METHODS:Seven centers across the United States and Norway. METHODS:Adults with sporadic VS who underwent salvage microsurgery following failed primary SRS were included. The primary outcome of interest was the association between tumor size at the time of primary SRS and the ability to achieve gross total resection (GTR) and maintain postoperative House-Brackmann (HB) facial nerve grade I at the last follow-up after salvage microsurgery. RESULTS:Among 96 patients, the median (interquartile range [IQR]) cerebellopontine angle (CPA) tumor size at primary SRS was 14.5 mm (10.0-19.0). Each 1-mm increase in CPA tumor size at the time of primary SRS was associated with a 13% increased likelihood of near-total/subtotal resection or most recent postoperative HB grade >I (odds ratio [OR] 1.13, 95% confidence interval [CI] 1.05-1.21, P = .001), with an optimal tumor size threshold to distinguish this outcome of 12 mm of CPA extension (c-index 0.73). Similarly, for each 1-mm increase in CPA tumor size at the time of primary SRS, a 9% increase in any postoperative complication with salvage microsurgery was observed (OR 1.09, 95% CI 1.02-1.15, P = .009). CONCLUSION/CONCLUSIONS:Corroborated by size threshold surveillance data informing the timing of primary microsurgical resection, the current study suggests that VS outcomes are optimized when primary radiosurgical intervention is undertaken on growing tumors when they harbor 10-15 mm of cerebellopontine angle extension or less.

The adoption of large language models (LLMs) in healthcare demands a careful analysis of their potential to spread false medical knowledge. Because LLMs ingest massive volumes of data from the open Internet during training, they are potentially exposed to unverified medical knowledge that may include deliberately planted misinformation. Here, we perform a threat assessment that simulates a data-poisoning attack against The Pile, a popular dataset used for LLM development. We find that replacement of just 0.001% of training tokens with medical misinformation results in harmful models more likely to propagate medical errors. Furthermore, we discover that corrupted models match the performance of their corruption-free counterparts on open-source benchmarks routinely used to evaluate medical LLMs. Using biomedical knowledge graphs to screen medical LLM outputs, we propose a harm mitigation strategy that captures 91.9% of harmful content (F1 = 85.7%). Our algorithm provides a unique method to validate stochastically generated LLM outputs against hard-coded relationships in knowledge graphs. In view of current calls for improved data provenance and transparent LLM development, we hope to raise awareness of emergent risks from LLMs trained indiscriminately on web-scraped data, particularly in healthcare where misinformation can potentially compromise patient safety.

BACKGROUND:Accurate intraoperative diagnosis is crucial for differentiating between primary CNS lymphoma (PCNSL) and other CNS entities, guiding surgical decision-making, but represents significant challenges due to overlapping histomorphological features, time constraints, and differing treatment strategies. We combined stimulated Raman histology (SRH) with deep learning to address this challenge. METHODS:We imaged unprocessed, label-free tissue samples intraoperatively using a portable Raman scattering microscope, generating virtual H&E-like images within less than three minutes. We developed a deep learning pipeline called RapidLymphoma based on a self-supervised learning strategy to (1) detect PCNSL, (2) differentiate from other CNS entities, and (3) test the diagnostic performance in a prospective international multicenter cohort and two additional independent test cohorts. We trained on 54,000 SRH patch images sourced from surgical resections and stereotactic-guided biopsies, including various CNS neoplastic/non-neoplastic lesions. Training and test data were collected from four tertiary international medical centers. The final histopathological diagnosis served as ground-truth. RESULTS:In the prospective test cohort of PCNSL and non-PCNSL entities (n=160), RapidLymphoma achieved an overall balanced accuracy of 97.81% ±0.91, non-inferior to frozen section analysis in detecting PCNSL (100% vs. 77.77%). The additional test cohorts (n=420, n=59) reached balanced accuracy rates of 95.44% ±0.74 and 95.57% ±2.47 in differentiating IDH-wildtype diffuse gliomas and various brain metastasis from PCNSL. Visual heatmaps revealed RapidLymphoma's capabilities to detect class-specific histomorphological key features. CONCLUSIONS:RapidLymphoma proves reliable and valid for intraoperative PCNSL detection and differentiation from other CNS entities. It provides visual feedback within three minutes, enabling fast clinical decision-making and subsequent treatment strategy planning.

BACKGROUND AND OBJECTIVES/OBJECTIVE:The management of World Health Organization (WHO) grade 2 meningiomas is complicated by their diverse clinical behaviors. Stereotactic radiosurgery (SRS) can be an effective management option. Literature on SRS dose selection is limited but suggests that a higher dose is better for tumor control. We characterize the predictors of post-SRS outcomes that can help guide planning and management. METHODS:We reviewed a cohort of consecutive patients with pathologically-proven WHO grade 2 meningiomas who underwent SRS at a single institution between 2011 and 2023. RESULTS:Ninety-nine patients (median age 62 years) underwent SRS, 11 of whom received hypofractionated SRS in 5 fractions. Twenty-two patients had received previous irradiation. The median follow-up was 49 months. The median overall survival was 119 months (95% CI 92-NA) with estimated 5- and 10-year survival of 83% and 27%, respectively. The median progression-free survival (PFS) was 40 months (95% CI 32-62), with 3- and 5-year rates at 54% and 35%, respectively. The median locomarginal PFS was 63 months (95% CI 51.8-NA) with 3- and 5-year rates at 65% and 52%. Nine (9%) patients experienced adverse events, 2 Common Terminology Criteria for Adverse Events grade 3 and 7 grade 2, consisting of worsening neurologic deficit from edema. In the single-session cohort, Ki-67 significantly predicted both overall survival and intracranial PFS. Tumors with Ki-67 >10% had 2.17 times the risk of locomarginal progression compared with Ki-67 ≤10% (P = .018) adjusting for covariates. Sex, prescription dose, tumor volume, and location also predicted tumor control. In tumors with Ki-67 >10%, margin dose ≥14 Gy was associated with significantly better tumor control but not for tumors with Ki-67 ≤10%. CONCLUSION/CONCLUSIONS:The management of WHO grade 2 meningiomas requires a multimodality approach. This study demonstrates the value of a targeted SRS approach in patients with limited disease and further establishes predictive biomarkers that can guide planning through a personalized approach.

The detection and tracking of metastatic cancer over the lifetime of a patient remains a major challenge in clinical trials and real-world care. Advances in deep learning combined with massive datasets may enable the development of tools that can address this challenge. We present NYUMets-Brain, the world's largest, longitudinal, real-world dataset of cancer consisting of the imaging, clinical follow-up, and medical management of 1,429 patients. Using this dataset we developed Segmentation-Through-Time, a deep neural network which explicitly utilizes the longitudinal structure of the data and obtained state-of-the-art results at small (<10 mm³) metastases detection and segmentation. We also demonstrate that the monthly rate of change of brain metastases over time are strongly predictive of overall survival (HR 1.27, 95%CI 1.18-1.38). We are releasing the dataset, codebase, and model weights for other cancer researchers to build upon these results and to serve as a public benchmark.

PURPOSE/OBJECTIVE:DNA methylation profiling stratifies isocitrate dehydrogenase (IDH)-mutant astrocytomas into methylation low-grade and high-grade groups. We investigated the utility of the T2-FLAIR mismatch sign for predicting DNA methylation grade and cyclin-dependent kinase inhibitor 2A/B (CDKN2A/B) homozygous deletion, a molecular biomarker for grade 4 IDH-mutant astrocytomas, according to the 2021 World Health Organization (WHO) classification. EXPERIMENTAL DESIGN/METHODS:Preoperative MRI scans of IDH-mutant astrocytomas subclassified by DNA methylation profiling (n=71) were independently evaluated by two radiologists for the T2-FLAIR mismatch sign. The diagnostic utility of T2-FLAIR mismatch in predicting methylation grade, CDKN2A/B status, copy number variation, and survival was analyzed. RESULTS:The T2-FLAIR mismatch sign was present in 21 of 45 (46.7%) methylation low-grade and 1 of 26 (3.9%) methylation high-grade cases (p<0.001), resulting in 96.2% specificity, 95.5% positive predictive value, and 51.0% negative predictive value for predicting low methylation grade. The T2-FLAIR mismatch sign was also significantly associated with intact CDKN2A/B status (p=0.028) with 87.5% specificity, 86.4% positive predictive value, and 42.9% negative predictive value. Overall multivariable Cox analysis showed that retained CDKN2A/B status remained significant for PFS (p=0.01). Multivariable Cox analysis of the histologic grade 3 subset, which was nearly evenly divided by CDKN2A/B status, CNV, and methylation grade, showed trends toward significance for DNA methylation grade with OS (p=0.045) and CDKN2A/B status with PFS (p=0.052). CONCLUSIONS:The T2-FLAIR mismatch sign is highly specific for low methylation grade and intact CDKN2A/B in IDH-mutant astrocytomas.