Faculty Bibliography

PURPOSE/OBJECTIVE:Necrosis quantification in the neoadjuvant setting using pathology slide review is the most important validated prognostic marker in conventional osteosarcoma. Herein, we explored three deep learning strategies on histology samples to predict outcome for OSA in the neoadjuvant setting. EXPERIMENTAL DESIGN/METHODS:Our study relies on a training cohort from New York University (New York, NY) and an external cohort from Charles university (Prague, Czechia). We trained and validated the performance of a supervised approach that integrates neural network predictions of necrosis/tumor content, and compared predicted overall survival (OS) using Kaplan-Meier curves. Furthermore, we explored morphology-based supervised and self-supervised approaches to determine whether intrinsic histomorphological features could serve as a potential marker for OS in the setting of neoadjuvant. RESULTS:Excellent correlation between the trained network and the pathologists was obtained for the quantification of necrosis content (R2=0.899, r=0.949, p < 0.0001). OS prediction cutoffs were consistent between pathologists and the neural network (22% and 30% of necrosis, respectively). Morphology-based supervised approach predicted OS with p-value=0.0028, HR=2.43 [1.10-5.38]. The self-supervised approach corroborated the findings with clusters enriched in necrosis, fibroblastic stroma, and osteoblastic morphology associating with better OS (lg2HR; -2.366; -1.164; -1.175; 95% CI=[-2.996; -0.514]). Viable/partially viable tumor and fat necrosis were associated with worse OS (lg2HR;1.287;0.822;0.828; 95% CI=[0.38-1.974]). CONCLUSIONS:Neural networks can be used to automatically estimate the necrosis to tumor ratio, a quantitative metric predictive of survival. Furthermore, we identified alternate histomorphological biomarkers specific to the necrotic and tumor regions themselves which can be used as predictors.

BACKGROUND:Risk of early-stage lung adenocarcinoma (LUAD) recurrence after surgical resection is significant, and post-recurrence median survival is approximately two years. Currently there are no commercially available biomarkers that predict recurrence. Here, we investigated whether microbial and host genomic signatures in the lung can predict recurrence. METHODS:In 91 early-stage (Stage IA/IB) LUAD-patients with extensive follow-up, we used 16s rRNA gene sequencing and host RNA-sequencing to map the microbial and host transcriptomic landscape in tumor and adjacent unaffected lung samples. RESULTS:23 out of 91 subjects had tumor recurrence over 5-year period. In tumor samples, LUAD recurrence was associated with enrichment with Dialister, Prevotella, while in unaffected lung, recurrence was associated with enrichment with Sphyngomonas and Alloiococcus. The strengths of the associations between microbial and host genomic signatures with LUAD recurrence were greater in adjacent unaffected lung samples than in the primary tumor. Among microbial-host features in the unaffected lung samples associated with recurrence, enrichment with Stenotrophomonas geniculata and Chryseobacterium were positively correlated with upregulation of IL-2, IL-3, IL-17, EGFR, HIF-1 signaling pathways among the host transcriptome. In tumor samples, enrichment with Veillonellaceae Dialister, Ruminococcacea, Haemophilus Influenza, and Neisseria were positively correlated with upregulation of IL-1, IL-6, IL17, IFN, and Tryptophan metabolism pathways. CONCLUSIONS:Overall, modeling suggested that a combined microbial/transcriptome approach using unaffected lung samples had the best biomarker performance (AUC=0.83). IMPACT/CONCLUSIONS:This study suggests that LUAD recurrence is associated with distinct pathophysiological mechanisms of microbial-host interactions in the unaffected lung rather than those present in the resected tumor.

Diverse cellular insults converge on activation of the heat shock factor 1 (HSF1), which regulates the proteotoxic stress response to maintain protein homoeostasis. HSF1 regulates numerous gene programmes beyond the proteotoxic stress response in a cell-type- and context-specific manner to promote malignancy. However, the role(s) of HSF1 in immune populations of the tumour microenvironment remain elusive. Here, we leverage an in vivo model of HSF1 activation and single-cell transcriptomic tumour profiling to show that augmented HSF1 activity in natural killer (NK) cells impairs cytotoxicity, cytokine production and subsequent anti-tumour immunity. Mechanistically, HSF1 directly binds and regulates the expression of key mediators of NK cell effector function. This work demonstrates that HSF1 regulates the immune response under the stress conditions of the tumour microenvironment. These findings have important implications for enhancing the efficacy of adoptive NK cell therapies and for designing combinatorial strategies including modulators of NK cell-mediated tumour killing.

Partitioning of repressive from actively transcribed chromatin in mammalian cells fosters cell-type-specific gene expression patterns. While this partitioning is reconstructed during differentiation, the chromatin occupancy of the key insulator, CCCTC-binding factor (CTCF), is unchanged at the developmentally important Hox clusters. Thus, dynamic changes in chromatin boundaries must entail other activities. Given its requirement for chromatin loop formation, we examined cohesin-based chromatin occupancy without known insulators, CTCF and Myc-associated zinc-finger protein (MAZ), and identified a family of zinc-finger proteins (ZNFs), some of which exhibit tissue-specific expression. Two such ZNFs foster chromatin boundaries at the Hox clusters that are distinct from each other and from MAZ. PATZ1 was critical to the thoracolumbar boundary in differentiating motor neurons and mouse skeleton, while ZNF263 contributed to cervicothoracic boundaries. We propose that these insulating activities act with cohesin, alone or combinatorially, with or without CTCF, to implement precise positional identity and cell fate during development.

As efforts to study the mechanisms of melanoma metastasis and novel therapeutic approaches multiply, researchers need accurate, high-throughput methods to evaluate the effects on tumor burden resulting from specific interventions. We show that automated quantification of tumor content from whole slide images is a compelling solution to assess in vivo experiments. In order to increase the outflow of data collection from preclinical studies, we assembled a large dataset with annotations and trained a deep neural network for the quantitative analysis of melanoma tumor content on histopathological sections of murine models. After assessing its performance in segmenting these images, the tool obtained consistent results with an orthogonal method (bioluminescence) of measuring metastasis in an experimental setting. This AI-based algorithm, made freely available to academic laboratories through a web-interface called MetFinder, promises to become an asset for melanoma researchers and pathologists interested in accurate, quantitative assessment of metastasis burden.

OBJECTIVE:Early-stage lung adenocarcinoma is treated with local therapy alone, although patients with grade 3 stage I lung adenocarcinoma have a 50% 5-year recurrence rate. Our objective is to determine if analysis of the tumor microenvironment can create a predictive model for recurrence. METHODS:Thirty-four patients with grade 3 stage I lung adenocarcinoma underwent surgical resection. Digital spatial profiling was used to perform genomic (n = 31) and proteomic (n = 34) analyses of pancytokeratin positive and negative tumor cells. K-means clustering was performed on the top 50 differential genes and top 20 differential proteins, with Kaplan-Meier recurrence curves based on patient clustering. External validation of high-expression genes was performed with Kaplan-Meier plotter. RESULTS:There were no significant clinicopathologic differences between patients who did (n = 14) and did not (n = 20) have recurrence. Median time to recurrence was 806 days; median follow-up with no recurrence was 2897 days. K-means clustering of pancytokeratin positive genes resulted in a model with a Kaplan-Meier curve with concordance index of 0.75. K-means clustering for pancytokeratin negative genes was less successful at differentiating recurrence (concordance index 0.6). Genes upregulated or downregulated for recurrence were externally validated using available public databases. Proteomic data did not reach statistical significance but did internally validate the genomic data described. CONCLUSIONS:Genomic difference in lung adenocarcinoma may be able to predict risk of recurrence. After further validation, stratifying patients by this risk may help guide who will benefit from adjuvant therapy.

Relapsed pediatric B-cell acute lymphoblastic leukemia (B-ALL) remains one of the leading causes of cancer mortality in children. We performed Hi-C, ATAC-seq, and RNA-seq on 12 matched diagnosis/relapse pediatric leukemia specimens to uncover dynamic structural variants (SVs) and 3D chromatin rewiring that may contribute to relapse. While translocations are assumed to occur early in leukemogenesis and be maintained throughout progression, we discovered novel, dynamic translocations and confirmed several fusion transcripts, suggesting functional and therapeutic relevance. Genome-wide chromatin remodeling was observed at all organizational levels: A/B compartments, TAD interactivity, and chromatin loops, including some loci shared by 25% of patients. Shared changes were found to drive the expression of genes/pathways previously implicated in resistance as well as novel therapeutic candidates, two of which (ATXN1 and MN1) we functionally validated. Overall, these results demonstrate chromatin reorganization under the selective pressure of therapy and offer the potential for discovery of novel therapeutic interventions.

UNLABELLED:Melanoma being one of the most common and deadliest skin cancers, has been rising 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 resistances. This underscores a need for novel approaches and therapeutic targets as well as a better understanding of the mechanisms of melanoma pathogenesis. Long non-coding RNAs (lncRNAs) 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 resistances, however systematic screens to uncover novel functional lncRNAs are scarce. Here, 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 on melanoma. SIGNIFICANCE/UNASSIGNED:Previously considered as transcriptional noise, lncRNAs have emerged as novel players in regulating many cellular aspects in health and disease including melanoma. However, the number and as well as the extent of functional significance 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 lowering the experimental effort. We also provide a larger resource of differentially expressed lncRNAs with potential implications in melanoma growth and invasion. Our results broaden the characterized of lncRNAs as potential targets for future therapeutic applications.

Aberrations in the capacity of DNA/chromatin modifiers and transcription factors to bind non-coding regions can lead to changes in gene regulation and impact disease phenotypes. However, identifying distal regulatory elements and connecting them with their target genes remains challenging. Here, we present MethNet, a pipeline that integrates large-scale DNA methylation and gene expression data across multiple cancers, to uncover cis regulatory elements (CREs) in a 1 Mb region around every promoter in the genome. MethNet identifies clusters of highly ranked CREs, referred to as 'hubs', which contribute to the regulation of multiple genes and significantly affect patient survival. Promoter-capture Hi-C confirmed that highly ranked associations involve physical interactions between CREs and their gene targets, and CRISPR interference based single-cell RNA Perturb-seq validated the functional impact of CREs. Thus, MethNet-identified CREs represent a valuable resource for unraveling complex mechanisms underlying gene expression, and for prioritizing the verification of predicted non-coding disease hotspots.

The transfer of regulatory information between distal loci on chromatin is thought to involve physical proximity, but key biophysical features of these contacts remain unclear. For instance, it is unknown how close and for how long two loci need to be in order to productively interact. The main challenge is that it is currently impossible to measure chromatin dynamics with high spatiotemporal resolution at scale. Polymer simulations provide an accessible and rigorous way to test biophysical models of chromatin regulation, yet there is a lack of simple and general methods for extracting the values of model parameters. Here we adapt the Nelder-Mead simplex optimization algorithm to select the best polymer model matching a given Hi-C dataset, using the MYC locus as an example. The model's biophysical parameters predict a compartmental rearrangement of the MYC locus in leukemia, which we validate with single-cell measurements. Leveraging trajectories predicted by the model, we find that loci with similar Hi-C contact frequencies can exhibit widely different contact dynamics. Interestingly, the frequency of productive interactions between loci exhibits a non-linear relationship with their Hi-C contact frequency when we enforce a specific capture radius and contact duration. These observations are consistent with recent experimental observations and suggest that the dynamic ensemble of chromatin configurations, rather than average contact matrices, is required to fully predict productive long-range chromatin interactions.