Faculty Bibliography

INTRODUCTION/UNASSIGNED:Outcomes for NSCLC remain suboptimal. Recent data suggest that cryoablation can generate antitumor immune effects. In this first-in-human phase I clinical trial, we investigated the safety and feasibility of bronchoscopic cryoimmunotherapy (BCI) delivered during standard-of-care bronchoscopy and explored associated systemic immune responses. METHODS/UNASSIGNED:Subjects with known or suspected advanced-stage NSCLC were recruited. BCI was delivered in dose-escalated freeze-thaw cycles to determine maximum dose tolerance. Feasibility assessment was determined with a pre-set goal of achieving successful BCI in more than or equal to 80% of subjects. Safety was assessed by review of BCI-related complications, including grades 2 to 3 bleeding, pneumothorax requiring intervention, and National Cancer Institute Common Terminology Criteria for Adverse Events grade 3 to 5 adverse events. Pre- and post-BCI blood samples were collected to explore changes in the systemic immune profile. RESULTS/UNASSIGNED:Subjects with predominantly clinical TNM stage 3 or 4 adenocarcinoma or squamous cell carcinoma were enrolled. We reached the maximum dose of 30 seconds with 100% feasibility and no BCI-related adverse events. In peripheral blood analysis, we observed a significant decrease in derived neutrophil-to-lymphocyte ratio in the high-dose BCI group in comparison to the low-dose BCI cohort. We also observed increases in inflammatory cytokines-GM-CSF, IFN-γ, IL-1β, IL-17A, and IL-2-and effector memory T cells post-BCI. CONCLUSION/UNASSIGNED:BCI is safe and feasible. In addition, we provide preliminary evidence that at higher dose levels there is a systemic immune response consistent with a cytotoxic profile. Further immune analyses will determine the potential of BCI as an adjunctive therapy in combination with immune checkpoint inhibition in NSCLC treatment.

UNLABELLED:Melanoma being one of the most common and deadliest skin cancers has been increasing since the past decade. Patients at advanced stages of the disease have very poor prognoses, as opposed to at the earlier stages. Nowadays, the standard of care of advanced melanoma is resection, followed by immune checkpoint inhibition-based immunotherapy. However, a substantial proportion of patients either do not respond or develop resistance. This underscores a need for novel approaches and therapeutic targets as well as a better understanding of the mechanisms of melanoma pathogenesis. Long noncoding RNAs (lncRNA) comprise a poorly characterized class of functional players and promising targets in promoting malignancy. Certain lncRNAs have been identified to play integral roles in melanoma progression and drug resistance; however, systematic screens to uncover novel functional lncRNAs are scarce. In this study, we profile differentially expressed lncRNAs in patient-derived short-term metastatic cultures and BRAF-MEK inhibition-resistant cells. We conduct a focused growth-related CRISPR inhibition screen of overexpressed lncRNAs, validate, and functionally characterize lncRNA hits with respect to cellular growth, invasive capacities, and apoptosis in vitro as well as the transcriptomic impact of our lead candidate the novel lncRNA XLOC_030781. In sum, we extend the current knowledge of ncRNAs and their potential relevance in melanoma. SIGNIFICANCE/UNASSIGNED:LncRNAs have emerged as novel players in regulating many cellular aspects also in melanoma. The number of functional significances of most lncRNAs remains elusive. We provide a comprehensive strategy to identify functionally relevant lncRNAs in melanoma by combining expression profiling with CRISPR inhibition growths screens. Our results broaden the characterized lncRNAs as potential targets for future therapeutic applications.

Despite recent advances, the regulation of anticancer and antimicrobial bioactive compound (AABC) production by leukocytes remains poorly understood. Here, we demonstrate that inactivation of the DNA- and RNA-based Teazeled receptors of the Universal Receptive System in human leukocytes generated so called "Leukocyte-Tells," which showed enhanced AABC production. Comprehensive analysis of the AABCs produced by Leukocyte-Tells based on LC/MS identified 707 unique or differentially produced peptide or nonpeptide metabolites. Functional testing demonstrated that many of these metabolites exhibited increased antibacterial, antifungal, and anticancer activities. The AABCs produced by the Leukocyte-Tells were effective against different multidrug-resistant clinical isolates of fungi and gram-positive and gram-negative bacteria (including their biofilms), as well as various cancer cell lines, with >100,000-fold activity than AABCs derived from control leukocytes. Notably, the AABCs produced by the Leukocyte-Tells exhibited greater stability under adverse environmental conditions. Our findings highlight the important role of the Universal Receptive System in regulating AABC production through a process named here as cell genome-memory management, offering new insights into immune functions and suggesting potential therapeutic applications.

BACKGROUND AND OBJECTIVES/OBJECTIVE:Classical biomedical data science models are trained on a single modality and aimed at one specific task. However, the exponential increase in the size and capabilities of the foundation models inside and outside medicine shows a shift toward task-agnostic models using large-scale, often internet-based, data. Recent research into smaller foundation models trained on specific literature, such as programming textbooks, demonstrated that they can display capabilities similar to or superior to large generalist models, suggesting a potential middle ground between small task-specific and large foundation models. This study attempts to introduce a domain-specific multimodal model, Congress of Neurological Surgeons (CNS)-Contrastive Language-Image Pretraining (CLIP), developed for neurosurgical applications, leveraging data exclusively from Neurosurgery Publications. METHODS:We constructed a multimodal data set of articles from Neurosurgery Publications through PDF data collection and figure-caption extraction using an artificial intelligence pipeline for quality control. Our final data set included 24 021 figure-caption pairs. We then developed a fine-tuning protocol for the OpenAI CLIP model. The model was evaluated on tasks including neurosurgical information retrieval, computed tomography imaging classification, and zero-shot ImageNet classification. RESULTS:CNS-CLIP demonstrated superior performance in neurosurgical information retrieval with a Top-1 accuracy of 24.56%, compared with 8.61% for the baseline. The average area under receiver operating characteristic across 6 neuroradiology tasks achieved by CNS-CLIP was 0.95, slightly superior to OpenAI's Contrastive Language-Image Pretraining at 0.94 and significantly outperforming a vanilla vision transformer at 0.62. In generalist classification, CNS-CLIP reached a Top-1 accuracy of 47.55%, a decrease from the baseline of 52.37%, demonstrating a catastrophic forgetting phenomenon. CONCLUSION/CONCLUSIONS:This study presents a pioneering effort in building a domain-specific multimodal model using data from a medical society publication. The results indicate that domain-specific models, while less globally versatile, can offer advantages in specialized contexts. This emphasizes the importance of using tailored data and domain-focused development in training foundation models in neurosurgery and general medicine.

Chromosomal structural variants (SVs) are major contributors to cancer development. Although multiple methods exist for detecting SVs, they are limited in throughput, such as fluorescent in situ hybridization and targeted panels, and use RNA, which degrades in formalin-fixed, paraffin-embedded (FFPE) blocks and is unable to detect SVs that do not produce a fusion transcript. High-throughput chromosome conformation capture (Hi-C) is a DNA-based next-generation sequencing (NGS) method that preserves the spatial conformation of the genome, capturing long-range genetic interactions and SVs. Herein, a retrospective study analyzing 71 FFPE specimens from 10 different solid tumors was performed. Results showed high concordance (98%) with clinical fluorescent in situ hybridization and RNA NGS in detecting known SVs. Furthermore, Hi-C provided insight into the mechanism of SV formation, including chromothripsis and extrachromosomal DNA, and detected rearrangements between genes and regulatory regions, all of which are undetectable by RNA NGS. Lastly, SVs were detected in 71% of cases in which previous clinical methods failed to identify a driver. Of these, 14% were clinically actionable based on current medical guidelines, and an additional 14% were not in medical guidelines but involve targetable biomarkers. Current data suggest that Hi-C is a robust and accurate method for genome-wide SV analyses from FFPE tissue and can be incorporated into current clinical NGS workflows.

While the mutational landscape across early T-cell precursor acute lymphoblastic leukemia (ETP-ALL) and ETP-like leukemia is known, establishing a unified framework that activates stem cell genes characteristic of these tumors remains elusive. Using complementary mouse and human models, chromatin mapping, and enhancer profiling, we show that the coactivator ZMIZ1 promotes normal and malignant ETP population growth by inducing the transcription factor MYB in feedforward circuits to convergently activate oncogenes (MEF2C, MYCN, and BCL2) through essential enhancers. A key superenhancer, the N-Myc regulating enhancer (NMRE), drives malignant ETP population growth but is dispensable for normal lymphopoiesis. This network of stem cell superenhancers identifies treatment-resistant tumors and poor survival outcomes; unifies diverse ETP-ALLs; and contributes to cardinal features of the recently genomically identified high-risk bone marrow progenitor-like (BMP-like) ETP-ALL tumor-stem cell/myeloid gene expression, inhibited NOTCH1-induced T-cell development, aggressive clinical behavior, and venetoclax sensitivity. Since ZMIZ1 is dispensable for essential homeostasis, it might be possible to safely target this network to treat high-risk diseases.

Here we used a series of CTCF mutations to explore CTCF's relationship with chromatin and its contribution to gene regulation. CTCF's impact depends on the genomic context of bound sites and the unique binding properties of WT and mutant CTCF proteins. Specifically, CTCF's signal strength is linked to changes in accessibility, and the ability to block cohesin is linked to its binding stability. Multivariate modeling reveals that both CTCF and accessibility contribute independently to cohesin binding and insulation, but CTCF signal strength has a stronger effect. CTCF and chromatin have a bidirectional relationship such that at CTCF sites, accessibility is reduced in a cohesin-dependent, mutant-specific fashion. In addition, each mutant alters TF binding and accessibility in an indirect manner, changes which impart the most influence on rewiring transcriptional networks and the cell's ability to differentiate. Collectively, the mutant perturbations provide a rich resource for determining CTCF's site-specific effects.

Self-supervised learning (SSL) automates the extraction and interpretation of histopathology features on unannotated hematoxylin-eosin-stained whole slide images (WSIs). We train an SSL Barlow Twins encoder on 435 colon adenocarcinoma WSIs from The Cancer Genome Atlas to extract features from small image patches (tiles). Leiden community detection groups tiles into histomorphological phenotype clusters (HPCs). HPC reproducibility and predictive ability for overall survival are confirmed in an independent clinical trial (N = 1213 WSIs). This unbiased atlas results in 47 HPCs displaying unique and shared clinically significant histomorphological traits, highlighting tissue type, quantity, and architecture, especially in the context of tumor stroma. Through in-depth analyses of these HPCs, including immune landscape and gene set enrichment analyses, and associations to clinical outcomes, we shine light on the factors influencing survival and responses to treatments of standard adjuvant chemotherapy and experimental therapies. Further exploration of HPCs may unveil additional insights and aid decision-making and personalized treatments for colon cancer patients.

High-dimensional multiplexed imaging can reveal the spatial organization of tumour tissues at the molecular level. However, owing to the scale and information complexity of the imaging data, it is challenging to discover and thoroughly characterize the heterogeneity of tumour microenvironments. Here we show that self-supervised representation learning on data from imaging mass cytometry can be leveraged to distinguish morphological differences in tumour microenvironments and to precisely characterize distinct microenvironment signatures. We used self-supervised masked image modelling to train a vision transformer that directly takes high-dimensional multiplexed mass-cytometry images. In contrast with traditional spatial analyses relying on cellular segmentation, the vision transformer is segmentation-free, uses pixel-level information, and retains information on the local morphology and biomarker distribution. By applying the vision transformer to a lung-tumour dataset, we identified and validated a monocytic signature that is associated with poor prognosis.