Search

Immune checkpoint blockade (ICB) has shown immense promise for treating patients with various cancer types, but its effectiveness relies on our ability to identify likely responders. Here, we examined the association between mutations in 25 core DNA repair genes and ICB outcomes in 6619 patients across 9 cancer types with advanced disease and MSK-IMPACT tumor sequencing. Notably, we observed that mutations in 7 of the DNA repair genes (ATM, ATR, POLE, ERCC4, NBN, RAD50, PARP1) were significantly associated with improved overall survival in ICB-treated patients (p0.05 for all) and had significant interaction with treatment (p

Purpose:Perineural invasion (PNI) is associated with aggressive tumor behavior, recurrence, and metastasis, and can influence the administration of adjuvant treatment. However, standard histopathologic examination has limited sensitivity in detecting PNI and does not provide insights into its mechanistic underpinnings.Experimental Design:A multivariate Cox regression was performed to validate associations between PNI and survival in 2,029 patients across 12 cancer types. Differential expression and gene set enrichment analysis were used to learn PNI-associated programs. Machine learning models were applied to build a PNI gene expression classifier. A blinded re-review of hematoxylin and eosin (H&E) slides by a board-certified pathologist helped determine whether the classifier could improve occult histopathologic detection of PNI.Results:PNI associated with both poor overall survival [HR, 1.73; 95% confidence interval (CI), 1.27–2.36; P < 0.001] and disease-free survival (HR, 1.79; 95% CI, 1.38–2.32; P < 0.001). Neural-like, prosurvival, and invasive programs were enriched in PNI-positive tumors (Padj < 0.001). Although PNI-associated features likely reflect in part the increased presence of nerves, many differentially expressed genes mapped specifically to malignant cells from single-cell atlases. A PNI gene expression classifier was derived using random forest and evaluated as a tool for occult histopathologic detection. On a blinded H&E re-review of sections initially described as PNI negative, more specimens were reannotated as PNI positive in the high classifier score cohort compared with the low-scoring cohort (P = 0.03, Fisher exact test).Conclusions:This study provides salient biological insights regarding PNI and demonstrates a role for gene expression classifiers to augment detection of histopathologic features.

Supplementary Data Legends

Supplementary Figure S1

Supplementary Figure S2

Supplementary Table 1

Pancreatic ductal adenocarcinoma (PDAC) is projected to be the second leading cause of cancer mortality in the United States by 2030. Given that resistance to cytotoxic therapy is pervasive, there is a critical need to elucidate salient gene expression programs and spatial relationships among malignant and stromal cells in the tumor microenvironment (TME), particularly in residual disease. We developed and applied a single-nucleus RNA-seq (snRNA-seq) technique to 43 banked frozen primary PDAC specimens that either received neoadjuvant therapy (n=25) or were treatment-naïve (n=18). We discovered expression programs across malignant cell and fibroblast profiles that formed the basis for a refined molecular taxonomy, including a novel neural-like progenitor (NRP) malignant program enriched with neoadjuvant treatment in tumors and organoids, and associated with the worst prognosis in bulk profiles from independent cohorts. To elucidate how neoadjuvant treatment and cancer cell- and fibroblast-intrinsic programs modulate the composition of multicellular neighborhoods, we performed spatial profiling with the GeoMx[1] platform (NanoString) on 21 formalin-fixed paraffin-embedded sections using the human whole transcriptome atlas (WTA). Each tumor showed intra-tumoral heterogeneity in tissue architecture and regions of interest (ROIs) with diverse patterns of neoplastic cells, cancer-associated fibroblasts (CAFs), and immune cells were selected for profiling. We deconvolved the WTA data with our snRNA-seq cell type signatures and mapped expression programs onto the tumor architecture to reveal three distinct multicellular neighborhoods, which we annotated as classical, squamoid-basaloid, and treatment-enriched. The observed enrichment in post-treatment residual disease of multiple spatially-defined receptor-ligand interactions and a neighborhood featuring the NRP program, neurotropic CAF program, and CD8+ T cells may open new therapeutic opportunities. Next, we mapped malignant/CAF programs and immune cell subsets at single-cell spatial resolution by performing spatial molecular imaging (SMI[2]; NanoString CosMx) using a panel of 960 RNA targets on a subset of seven tumors (2 untreated, 5 treated) and captured over 200,000 cells with an average of more than 450 transcripts detected per cell. Correlating ROIs from whole-transcriptome DSP to matched fields of view in kiloplex SMI enabled further dissection of PDAC architecture and treatment-associated remodeling of cell type distributions and receptor-ligand interactions. Ongoing functional studies have begun to elucidate the key regulatory elements underlying the distinct treatment-associated NRP malignant program and its interactions with the TME. Overall, the complementary combination of snRNA-seq, whole-transcriptome DSP, and kiloplex SMI provides a high-resolution molecular framework that can be harnessed to augment precision oncology efforts in pancreatic cancer. [1] GeoMx DSP is for Research Use Only and not for use in diagnostic procedures. [2] CosMx SMI is for Research Use Only and not for use in diagnostic procedures. Citation Format: William L. Hwang, Karthik A. Jagadeesh, Jimmy A. Guo, Hannah I. Hoffman, Carina Shiau, Jennifer Su, Payman Yadollahpour, Jason W. Reeves, Youngmi Kim, Sean Kim, Mark Gregory, Prajan Divakar, Eric Miller, Michael Rhodes, Sarah Warren, Erroll Rueckert, Kit Fuhrman, Daniel R. Zollinger, Robin Fropf, Joseph M. Beechem, Arnav Mehta, Toni Delorey, Cristin McCabe, Jaimie L. Barth, Piotr Zelga, Cristina R. Ferrone, Motaz Qadan, Keith D. Lillemoe, Rakesh K. Jain, Jennifer Y. Wo, Theodore S. Hong, Ramnik Xavier, Orit Rozenblatt-Rosen, Andrew J. Aguirre, Carlos Fernandez-Del Castillo, Andrew S. Liss, Mari Mino-Kenudson, David T. Ting, Tyler Jacks, Aviv Regev. Multicellular spatial community featuring a novel neuronal-like malignant phenotype is enriched in pancreatic cancer after neoadjuvant chemotherapy and radiotherapy [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2022; 2022 Apr 8-13. Philadelphia (PA): AACR; Cancer Res 2022;82(12_Suppl):Abstract nr SY12-04.

Pancreatic ductal adenocarcinoma (PDAC) is a highly lethal and treatment-refractory cancer. Molecular stratification in pancreatic cancer remains rudimentary and does not yet inform clinical management or therapeutic development. Here, we construct a high-resolution molecular landscape of the cellular subtypes and spatial communities that compose PDAC using single-nucleus RNA sequencing and whole-transcriptome digital spatial profiling (DSP) of 43 primary PDAC tumor specimens that either received neoadjuvant therapy or were treatment naive. We uncovered recurrent expression programs across malignant cells and fibroblasts, including a newly identified neural-like progenitor malignant cell program that was enriched after chemotherapy and radiotherapy and associated with poor prognosis in independent cohorts. Integrating spatial and cellular profiles revealed three multicellular communities with distinct contributions from malignant, fibroblast and immune subtypes: classical, squamoid-basaloid and treatment enriched. Our refined molecular and cellular taxonomy can provide a framework for stratification in clinical trials and serve as a roadmap for therapeutic targeting of specific cellular phenotypes and multicellular interactions.

With increasing attention on the essential roles of the tumour microenvironment in recent years, the nervous system has emerged as a novel and crucial facilitator of cancer growth. In this Review, we describe the foundational, translational, and clinical advances illustrating how nerves contribute to tumour proliferation, stress adaptation, immunomodulation, metastasis, electrical hyperactivity and seizures, and neuropathic pain. Collectively, this expanding knowledge base reveals multiple therapeutic avenues for cancer neuroscience that warrant further exploration in clinical studies. We discuss the available clinical data, including ongoing trials investigating novel agents targeting the tumour-nerve axis, and the therapeutic potential for repurposing existing neuroactive drugs as an anti-cancer approach, particularly in combination with established treatment regimens. Lastly, we discuss the clinical challenges of these treatment strategies and highlight unanswered questions and future directions in the burgeoning field of cancer neuroscience.

Pancreatic ductal adenocarcinoma (PDAC) features abundant perineural invasion (PNI). Intra-tumoral nerves play critical roles in cancer initiation, progression, recurrence, treatment-resistance, metastasis, and mortality for many malignancies but the diverse molecular mechanisms underlying tumor-nerve crosstalk remain largely unknown—hindering the development of therapies targeting this key pathological process. To address this gap, we performed whole transcriptome digital spatial profiling on twelve custom tissue microarrays (n=288 cores) derived from intratumorally-matched regions with and without PNI in primary PDAC specimens (n=31 patients) and independently measured gene expression from cancer cells, fibroblasts, and nerves. We undertook a differential gene expression analysis comparing malignant cells in PNI-present and -absent regions. We quantified the growth kinetics of dorsal root ganglia (DRG) sensory neurons cultured with exogenous candidate proteins, which validated that some candidates augment (e.g., Lgals1) and others inhibit (e.g., Sema3b) neurite outgrowth. Next, we mapped our previously discovered malignant cell programs onto the epithelial segments and observed a significant enrichment of the mesenchymal, basal-like, and neural-like progenitor (NRP) programs versus depletion of the classical program in PNI-present regions. To determine the effects of malignant subtype on nerve outgrowth, we engineered isogenic KrasG12D/+;Trp53FL/FL;Rosa26-dCas9-VPR (KP;dCas9-VPR) organoids to overexpress the master transcription factors (TFs) for each malignant subtype (e.g., Gata6 for classical, Glis3 for NRP). We then performed the DRG neuronal outgrowth assay using conditioned media from each subtype-specific organoid line versus an off-target control and observed that the classical line suppressed neurite outgrowth, the mesenchymal and basal-like lines were neutral, and the NRP line enhanced neurite outgrowth dynamics comparable to the Ngf positive control. Taken together, our findings suggest that the mechanisms underlying nerve recruitment/outgrowth and perineural invasion may be partly decoupled. To further test this hypothesis, we are performing transwell invasion assays comparing KP;dCas9-VPR cancer cell lines that overexpress each of the candidate PNI-associated genes and malignant subtype TFs. We anticipate that this study will transform our understanding of how cancer cells and the peripheral nervous system collaborate and guide prioritization for therapeutic intervention in the burgeoning cancer neuroscience field. Citation Format: William L. Hwang, Jennifer Su, Carina Shiau, Peter L. Wang, Jimmy A. Guo, Nicole A. Lester, Jaimie L. Barth, Hannah I. Hoffman, Prajan Divakar, Jason W. Reeves, Eric Miller, Joseph M. Beechem, Andrew J. Aguirre, Lei Zheng, David T. Ting, Mari Mino-Kenudson, Tyler Jacks. Distinct cancer-intrinsic mechanisms mediate nerve recruitment/outgrowth versus perineural invasion [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2023; Part 1 (Regular and Invited Abstracts); 2023 Apr 14-19; Orlando, FL. Philadelphia (PA): AACR; Cancer Res 2023;83(7_Suppl):Abstract nr 2513.

Intratumoral nerves play important and versatile roles in cancer initiation, progression, recurrence, treatment-resistance, metastasis, morbidity, and mortality for many malignancies but the diverse molecular mechanisms underlying tumor-nerve crosstalk remain largely unknown. One of the differentiating hallmarks of pancreatic ductal adenocarcinoma (PDAC) is an exceptionally high frequency of perineural invasion (PNI), a histopathologic manifestation of tumor-nerve crosstalk whereby cancer cells recruit, migrate towards, and envelop or invade peripheral nerves. Evidence for some neurochemicals/neurotrophins involved in PNI have been uncovered, but most of the underlying work was limited by a lack of cell-type specificity, spatial context, and fragmented focus on individual pathways. To address these shortcomings, we set out to comprehensively identify cell-type specific genes spatially linked to PNI in patient tumors and then dissect the functional roles of these genes through live imaging of dorsal root ganglia (DRG) sensory neurons incubated in conditioned media from cancer cell organoids overexpressing candidate genes via CRISPR activation (CRISPRa). First, we performed whole transcriptome digital spatial profiling (NanoString GeoMx) on twelve custom tissue microarrays (n=288 cores) derived from intratumorally-matched malignant regions with and without PNI in primary resected PDAC specimens (n=31 patients). Differential gene expression (DE) analysis (FDR < 0.001) for PNI demonstrated that for malignant cells there were 271 enriched and 65 depleted genes, and for fibroblasts there were 16 enriched and 27 depleted genes. We further evaluated associations between PNI and expression of malignant subtypes previously identified from single-nucleus RNA-seq applied to 43 primary resected PDAC specimens. We found that malignant cells engaged in PNI were enriched in the mesenchymal, basaloid and neural-like progenitor (NRP) subtypes and depleted in the classical subtype. To test these associations functionally, we generated isogenic murine organoid lines (KrasG12D/+;Trp53FL/FL;R26-dCas9-VPR) overexpressing subtype-driving transcription factors and collected conditioned media. DRG sensory neurons demonstrate enhanced and suppressed growth kinetics when grown in NRP and classical conditioned media, respectively; mesenchymal and basal-like conditioned media do not appear to influence growth kinetics. These results suggest that while mesenchymal, basaloid, and NRP cells likely all play a role in cancer cell invasion of nerves, NRP cells may have an additional role in tumor-nerve tropism. Additional experiments exploring the functional effects of the top enriched and depleted genes from the DE analysis are ongoing. We anticipate that this study will provide a high-resolution understanding of critical intercellular interactions in the PDAC tumor microenvironment that facilitate PNI and tumor-nerve crosstalk more broadly to guide novel therapeutic strategies. Citation Format: William L. Hwang, Jennifer Su, Jimmy A. Guo, Carina Shiau, Jaimie L. Barth, Hannah I. Hoffman, Prajan Divakar, Jason W. Reeves, Eric Miller, Grissel Cervantes-Jaramillo, William Freed-Pastor, Vanessa Funes, Jennifer Y. Wo, Theodore S. Hong, Carlos Fernandez-del Castillo, Lei Zheng, Andrew J. Aguirre, David T. Ting, Mari Mino-Kenudson, Tyler Jacks. Identifying mediators of perineural invasion in pancreatic cancer using spatial transcriptomics [abstract]. In: Proceedings of the AACR Special Conference on Pancreatic Cancer; 2022 Sep 13-16; Boston, MA. Philadelphia (PA): AACR; Cancer Res 2022;82(22 Suppl):Abstract nr C052.

Transcriptional states in pancreatic cancer can stratify patients by response to chemotherapy and clinical outcomes. These include the classical and basal-like states as well as a newly identified neural-like progenitor (NRP) state, which we have previously found to be enriched in primary patient tumors treated with neoadjuvant chemotherapy and radiotherapy. While several transcription factor drivers of classical and basal-like identity have been described, key regulators of the NRP state are unknown. Through in silico approaches, we identified candidate transcription factors of the NRP state, including GLIS3, a Krüppel-like zinc finger protein that mediates neuroendocrine fate during pancreatic development and differentiation of human embryonic stem cells into posterior neural progenitor cells. Our understanding of biologic and clinically-relevant attributes of transcriptional cell states remains limited by state-specific biases in various preclinical models. Existing human cell lines maintained as two-dimensional cultures tend to preferentially represent the basal-like state, whereas human three-dimensional organoid models grown in standard culture conditions best reflect the classical state. These phenotypes are therefore impacted by culture conditions as well as underlying genetic features. Furthermore, most murine pancreatic cancer models do not fully reflect the classical vs. basal-like state heterogeneity observed in humans. To enable systematic study of the classical, basal-like and NRP phenotypes, we developed isogenic KP (KrasG12D/+;Trp53FL/FL) murine organoids with a germline dCas9-VPR system to enable facile overexpression of state-specific transcription factors through CRISPR activation approaches. Quantitative PCR, RNA-sequencing, and proteomics confirmed Gata6, deltaN Trp63, and Glis3 as drivers of classical, basal-like, and NRP identity, respectively. DeltaN Trp63 organoids were further differentiated by loss of luminal morphology. Pairwise comparisons of global transcriptional alterations suggest the greatest similarities between the Gata6- and Glis3-overexpressed models, which is consistent with enhanced associations between classical and NRP states in patient tumors. Finally, although basal-like and NRP states are associated with poorer response to multi-agent chemotherapy, state-specific therapeutic sensitivities to other treatments remain incompletely defined. We therefore performed drug sensitivity assays with a panel of targeted therapies and unveiled state-specific sensitivities. These data were corroborated by drug sensitivity profiling of human patient-derived organoids and cell lines. Taken together, these results suggest a framework for defining cell state-specific vulnerabilities that may aid in stratifying and treating pancreatic cancer patients with new therapies. Citation Format: Jimmy A. Guo, Jennifer Su, Ananya Jambhale, Julien Dilly, Connor J. Hennessey, Carina Shiau, Patrick Yu, Steven Wang, Junning Wang, Laleh Abbassi, James Neiswender, Tate Bertea, Annan Yang, Qijia Yu, Peter Westcott, Jason Schenkel, Daniel Y. Kim, Hannah I. Hoffman, Grissel Cervantes Jaramillo, Giselle A. Uribe, Westley W. Wu, Arnav Mehta, David Ting, Julian A. Pacheco, Amy Deik, Clary Clish, Francisca Vazquez, Brian Wolpin, Aviv Regev, William A. Freed-Pastor, Joseph D. Mancias, Tyler Jacks, William L. Hwang, Andrew J. Aguirre. Systematic dissection of transcriptional states in pancreatic cancer [abstract]. In: Proceedings of the AACR Special Conference on Pancreatic Cancer; 2022 Sep 13-16; Boston, MA. Philadelphia (PA): AACR; Cancer Res 2022;82(22 Suppl):Abstract nr A052.

Most amyotrophic lateral sclerosis (ALS) cases are considered sporadic, without a known genetic basis, and lifestyle factors are suspected to play an etiologic role. We previously observed increased risk of ALS associated with high nail mercury levels as an exposure biomarker and thus hypothesized that mercury exposure via fish consumption patterns increases ALS risk. Lifestyle surveys were obtained from ALS patients (n = 165) and n = 330 age- and sex-matched controls without ALS enrolled in New Hampshire, Vermont, or Ohio, USA. We estimated their annual intake of mercury and omega-3 polyunsaturated fatty acid (PUFA) via self-reported seafood consumption habits, including species and frequency. In our multivariable model, family income showed a significant positive association with ALS risk (p = 0.0003, adjusted for age, sex, family history, education, and race). Neither the estimated annual mercury nor omega-3 PUFA intakes via seafood were associated with ALS risk. ALS incidence is associated with socioeconomic status, however, consistent with a prior international study, this relationship is not linked to mercury intake estimated via fish or seafood consumption patterns.

High tumor mutation burden (TMB) and total mutation count (TMC) can be predictive of better response to immune checkpoint blockade (ICB). Nevertheless, TMB and TMC are limited by variation across cancers and inconsistent definitions due to different profiling methods (targeted vs whole genome sequencing). Our objective was to identify genomic alterations (GAs) associated with ICB response and builds a novel genomic signature predictive of ICB response, independent of TMB/TMC.This was a pan-cancer next generation sequencing (NGS)-association study using January 2014-May 2016 data from AACR Project Genomics Evidence Neo-plasia Information Exchange (GENIE). Participants included 6619 patients with metastatic or un-resectable cancer across 9 cancer types (including 1572 ICB-treated patients). GA data was collected using next-generation sequencing (NGS) assays and downloaded from cbioportal.org. Predictive analyses for ICB response were performed to develop the signature (ImmGA).GAs in 16 genes were associated with improved OS in ICB-treated patients (p 0.005). 13 GAs were associated with an OS benefit in ICB-treated patients (PA novel signature predictive of ICB response (ImmGA) was developed from 13 GAs. Further investigation of the utility of ImmGA for treatment and trial selection is warranted.

Pancreatic ductal adenocarcinoma (PDAC) is a treatment-refractory malignancy in urgent need of a molecular framework for guiding therapeutic strategies. Bulk transcriptomic efforts over the past decade have yielded two broad consensus subtypes: classical pancreatic/epithelial versus basal-like/squamous/quasi-mesenchymal. Although this binary classification enables prognostic stratification, it does not currently inform the administration of treatments uniquely sensitive to either subtype. Furthermore, bulk mRNA studies are challenged by distinguishing contributions from the neoplastic compartment versus other cell types in the microenvironment, which is accentuated in PDAC given that neoplastic cellularity can be low. The application of single-cell transcriptomics to pancreatic tumors has generally lagged behind other cancer types due in part to the difficulty of extracting high-quality RNA from enzymatically degradative tissue, but emerging studies have and will continue to shed light on intratumoral heterogeneity, malignant–stromal interactions, and subtle transcriptional programs previously obscured at the bulk level. In conjunction with insights provided by single-cell/nucleus dissociative techniques, spatially resolved technologies should also facilitate the contextualization of gene programs and inferred cell–cell interactions within the tumor architecture. Finally, given that patients often receive neoadjuvant chemotherapy and/or chemoradiotherapy even in resectable disease, deciphering the gene programs enriched in or induced by cytotoxic therapy will be crucial for developing insights into complementary treatments aimed at eradicating residual cancer cells. Taken together, single-cell and spatial technologies provide an unprecedented opportunity to refine the foundations laid by prior bulk molecular studies and significantly augment precision oncology efforts in pancreatic cancer.

Purpose: Perineural invasion (PNI) is associated with aggressive tumor behavior, recurrence, and metastasis, and can influence the administration of adjuvant treatment. However, standard histopathologic examination has limited sensitivity in detecting PNI and does not provide insights into its mechanistic underpinnings. Experimental Design: A multivariate Cox regression was performed to validate associations between PNI and survival in 2,029 patients across 12 cancer types. Differential expression and gene set enrichment analysis were used to learn PNI-associated programs. Machine learning models were applied to build a PNI gene expression classifier. A blinded re-review of hematoxylin and eosin (H&E) slides by a board-certified pathologist helped determine whether the classifier could improve occult histopathologic detection of PNI. Results: PNI associated with both poor overall survival [HR, 1.73; 95% confidence interval (CI), 1.27–2.36; P < 0.001] and disease-free survival (HR, 1.79; 95% CI, 1.38–2.32; P < 0.001). Neural-like, prosurvival, and invasive programs were enriched in PNI-positive tumors (Padj < 0.001). Although PNI-associated features likely reflect in part the increased presence of nerves, many differentially expressed genes mapped specifically to malignant cells from single-cell atlases. A PNI gene expression classifier was derived using random forest and evaluated as a tool for occult histopathologic detection. On a blinded H&E re-review of sections initially described as PNI negative, more specimens were reannotated as PNI positive in the high classifier score cohort compared with the low-scoring cohort (P = 0.03, Fisher exact test). Conclusions: This study provides salient biological insights regarding PNI and demonstrates a role for gene expression classifiers to augment detection of histopathologic features.

Pancreatic ductal adenocarcinoma (PDAC) remains a treatment-refractory disease. Characterizing PDAC by mRNA profiling remains particularly challenging. Previously identified bulk expression subtypes were influenced by contaminating stroma and have not yet informed clinical management, whereas single cell RNA-seq (scRNA-seq) of fresh tumors under-represented key cell types. Here, we developed a robust single-nucleus RNA-seq (snRNA-seq) technique for frozen archival PDAC specimens and used it to study both untreated tumors and those that received neoadjuvant chemotherapy and radiotherapy (CRT). Gene expression programs learned across untreated malignant cell and fibroblast profiles uncovered a clinically relevant molecular taxonomy with improved prognostic stratification compared to prior classifications. Moreover, in the increasingly-adopted neoadjuvant treatment context, there was a depletion of classical-like phenotypes in malignant cells in favor of basal-like phenotypes associated with TNF-NFkB and interferon signaling as well as the presence of novel acinar and neuroendocrine classical-like states, which may be more resilient to cytotoxic treatment. Spatially-resolved transcriptomics revealed an association between malignant cells expressing these basal-like programs and higher immune infiltration with increased lymphocytic content, whereas those exhibiting classical-like programs were linked to sparser macrophage-predominant microniches, perhaps pointing to susceptibility to distinct therapeutic strategies. Our refined molecular taxonomy and spatial resolution can help advance precision oncology in PDAC through informative stratification in clinical trials and insights into differential therapeutic targeting leveraging the immune system.

610 Background: Pancreatic ductal adenocarcinoma (PDAC) is highly lethal and resistance to chemotherapy and radiotherapy is a major obstacle to improving clinical outcomes. Hence, there is an urgent need to elucidate the gene expression programs, spatial context, and interactions among different cell types in residual disease after neoadjuvant treatment. Methods: We optimized and applied single-nucleus RNA-seq (snRNA-seq) to 43 frozen primary PDAC tumors. Eighteen were treatment-naïve, 14 received FOLFIRINOX followed by radiotherapy with 5-FU or capecitabine (CRT), and 5 were subjected to CRT combined with losartan on protocol (CRTL). We performed unsupervised clustering of single nucleus profiles and then annotated and quantified cell subsets. Malignant and fibroblast gene expression programs were identified by consensus non-negative matrix factorization (cNMF). We mapped our cell type signatures and expression programs onto the tumor architecture using whole-transcriptome digital spatial profiling (DSP) to uncover distinct multicellular spatial neighborhoods and intercellular interactions that compose PDAC and are remodeled by neoadjuvant treatment. Results: Consistent with treatment effect, the proportion of malignant cells was significantly lower in tumors treated with neoadjuvant therapy. Within the immune compartment, CRTL was associated with a higher fraction of CD8+ T cells and Tregs compared to untreated and CRT tumors. Differential expression analysis of CD8+ T cells revealed greater effector function (e.g., IL2, CCL4, CCL5) and reduced quiescence/dysfunction markers (e.g., TIGIT, TCF7, KLF2, LEF1) associated with CRTL. We discovered expression programs across malignant and fibroblast profiles that formed a refined molecular taxonomy, including a novel neuronal-like malignant program enriched in the neoadjuvant groups and associated with the worst prognosis in independent cohorts. Ex vivo treatment of organoids derived from an untreated PDAC with FOLFIRINOX chemotherapy and radiotherapy recapitulated enrichment of the neuronal-like program. Whole-transcriptome DSP revealed three distinct multicellular neighborhoods: classical, squamoid-basaloid, and treatment-enriched. The observed enrichment in post-treatment residual disease of multiple spatially-defined receptor-ligand interactions and a neighborhood featuring colocalization of the neuronal-like malignant program, neurotropic CAF program, and CD8+ T cells may open new opportunities for therapeutic targeting in PDAC. Conclusions: Our work provides a high-resolution molecular framework for understanding the inter- and intra-tumoral heterogeneity of pancreatic cancer, spatial organization into discrete multicellular communities, and treatment-associated reprogramming as a blueprint for exploring novel therapeutic strategies tailored to residual disease.

Targeted therapies for molecularly-defined subtypes have led to immense clinical benefit for many cancer types but have generally not been successful for pancreatic cancer. Given that the mainstay of treatment remains multi-agent chemotherapy with FOLFIRINOX or gemcitabine/nab-paclitaxel, there remains an urgent need to identify novel actionable vulnerabilities for subsets of PDAC patients. Toward this end, we conducted an integrative, genome-scale examination of genetic dependencies and cell surface targets for PDAC by leveraging CRISPR and RNAi screening data from The Cancer Dependency Map Project, genomic data of bulk patient tumors from The Cancer Genome Atlas, and custom single-nucleus RNA-seq of a 43-patient cohort comprised of untreated and treated specimens. Our results re-affirm the prominence of Ras/MAPK signaling and a synthetically-lethal interaction between VPS4A/B, but also reveal recurrent susceptibilities to genes within the fatty acid metabolism, vesicular transport and exocytosis, and nucleobase synthesis pathways that otherwise have minor to moderate depleting effects on the majority of cell lines. Aberrations in frequent tumor suppressor genes and chromosomal arm-level variations appear to modify the strength of dependencies, including that of KRAS, CCND1, and GPX4, and may serve as predictive biomarkers of response. In addition, we leveraged mRNA profiling of bulk primary tumors as well as metastatic organoid models to conduct a genome-wide search for cell surface targets that are highly-expressed in tumors while lowly or not expressed in other toxicity-prone, non-malignant tissues. These putative drug targets do not need to be cancer dependencies and can be compatible with antibody-based therapeutic strategies that leverage alternative modes of cellular toxicity. Our approach identifies MSLN, NECTIN4, TROP2, and other antigens which have previously been shown to be largely tumor-specific but also reveals the expression of multiple putative targets within the CEACAM, claudin, and tetraspanin families. Finally, molecular subtyping efforts over the past decade have yielded classical and basal-like as consensus subtypes with variations therein, but genetic dependencies and cell surface expression patterns unique to either are insufficiently understood. We identified CLDN18, CEACAM5, and CEACAM6 as cell surface antigens for the classical subtype and MSLN, AQP5, and SLC6A14 for basal-like. Dependency on TLK2 and CCND1 is associated with the basal-like and classical subtype, respectively. Taken together, our integrative genomic approach may provide a precision medicine blueprint for stratifying and targeting pancreatic cancer. Citation Format: Jimmy A. Guo, Daniel Zhao, Scott P. Ginebaugh, Steven Wang, Ananya D. Jambhale, Patrick Z. Yu, Westley W. Wu, Peter Chen, Maryann Zhao, Kristen E. Lowder, Kevin S. Kapner, Hannah I. Hoffman, Stephanie W. Cheng, Daniel Y. Kim, Rebecca Boiarsky, Francois Aguet, Brenton Paolella, John M. Krill-Burger, James M. McFarland, Tobiloba Oni, Tyler Jacks, Aviv Regev, Gad Getz, William L. Hwang, Harshabad Singh, Andrew J. Aguirre. Integrative genomic characterization of therapeutic targets for pancreatic cancer [abstract]. In: Proceedings of the AACR Virtual Special Conference on Pancreatic Cancer; 2021 Sep 29-30. Philadelphia (PA): AACR; Cancer Res 2021;81(22 Suppl):Abstract nr PR-006.

A molecular classification of pancreatic ductal adenocarcinoma (PDAC) that informs clinical management remains elusive. Previously identified bulk expression subtypes in the untreated setting were influenced by contaminating stroma whereas single cell RNA-seq (scRNA-seq) of fresh tumors under-represented key cell types. Two consensus subtypes have arisen from these prior efforts: (1) classical-like, and (2) basal-like. Basal-like tumors were associated with worse survival in the metastatic setting but attempts to refine this binary classification have failed to further stratify patient survival. Here, we developed a robust single-nucleus RNA-seq (snRNA-seq) technique for banked frozen PDAC specimens and studied a cohort of untreated resected primary tumors (n ~ 20). Gene expression programs learned across malignant cell and cancer-associated fibroblast (CAF) profiles uncovered a clinically-relevant molecular taxonomy with improved prognostic stratification compared to prior classifications. Digital spatial profiling revealed an association between malignant cells expressing basal-like programs and greater immune infiltration with relatively fewer macrophages, whereas those exhibiting classical-like programs were linked to inflammatory CAFs and macrophage-predominant microniches. Recent clinical trials have supported the increasing adoption of neoadjuvant therapy to aggressively address the risk of micro-metastatic spread and to circumvent concerns of treatment tolerance in the postoperative setting. There is an urgent need to understand how preoperative treatment impacts residual tumor cells and their interactions with other cell types in the tumor microenvironment to identify additional therapeutic vulnerabilities that can be exploited. Towards this end, we performed snRNA-seq on an unmatched cohort of neoadjuvant-treated resected primary tumors (n ~ 25) with most cases involving FOLFIRINOX chemotherapy followed by chemoradiation. Remarkably, the quality of single-nucleus mRNA profiles was comparable between heavily pre-treated and untreated specimens. We identified differentially expressed genes between treated and untreated samples to infer cell-type specific reprogramming in the residual tumor. This analysis revealed that in the neoadjuvant treatment context, there was lower expression of classical-like phenotypes in malignant cells in favor of basal-like phenotypes associated with TNF-NFkB and interferon signaling as well as the presence of novel acinar and neuroendocrine classical-like states. Our refined molecular taxonomy and spatial resolution may help advance precision oncology in PDAC through informative stratification in clinical trials and insights into compartment-specific therapies. Citation Format: William L. Hwang, Karthik A. Jagadeesh, Jimmy A. Guo, Hannah I. Hoffman, Payman Yadollahpour, Jason Reeves, Eugene Drokhlyansky, Nicholas Van Wittenberghe, Samouil Farhi, Denis Schapiro, George Eng, Jason M. Schenkel, William A. Freed-Pastor, Orr Ashenberg, Clifton Rodrigues, Domenic Abbondanza, Toni Delorey, Devan Phillips, Jorge Roldan, Debora Ciprani, Marina Kern, Jaimie L. Barth, Daniel R. Zollinger, Kit Fuhrman, Robin Fropf, Joseph Beechem, Colin Weekes, Cristina R. Ferrone, Jennifer Y. Wo, Theodore S. Hong, Orit Rozenblatt-Rosen, Andrew J. Aguirre, Mari Mino-Kenudson, Carlos Fernandez-del- Castillo, Andrew S. Liss, David T. Ting, Tyler Jacks, Aviv Regev. Multi-compartment reprogramming and spatially-resolved interactions in frozen pancreatic cancer with and without neoadjuvant chemotherapy and radiotherapy at single-cell resolution [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2021; 2021 Apr 10-15 and May 17-21. Philadelphia (PA): AACR; Cancer Res 2021;81(13_Suppl):Abstract nr 94.

Molecular subtyping of pancreatic ductal adenocarcinoma (PDAC) remains in its nascent stages and does not currently inform clinical management or therapeutic development. Previously identified bulk expression subtypes in the untreated setting were influenced by contaminating stroma whereas single cell RNA-seq (scRNA-seq) of fresh tumors under-represented key cell types. Two consensus subtypes have arisen from these prior efforts: (1) classical-pancreatic, encompassing a spectrum of pancreatic lineage precursors, and (2) basal-like/squamous/quasi-mesenchymal, characterized by loss of endodermal identity and aberrations in chromatin modifiers. Basal-like tumors were associated with poorer responses to chemotherapy and worse survival in the metastatic setting but attempts to refine this binary classification have failed to further stratify patient survival. Recent clinical trials have supported the increasing adoption of neoadjuvant therapy to aggressively address the risk of micro-metastatic spread and to circumvent concerns of treatment tolerance in the postoperative setting. There is an urgent need to understand how preoperative treatment reprograms residual tumor cells to identify additional therapeutic vulnerabilities that can be exploited in combination with neoadjuvant CRT. Here, we developed a robust single-nucleus RNA-seq (snRNA-seq) technique for frozen archival PDAC specimens and used it to study both untreated tumors (n = 15) and those that received neoadjuvant CRT (n = 11). Gene expression programs learned across malignant cell and fibroblast profiles uncovered a clinically relevant molecular taxonomy with improved prognostic stratification (median survival: 11.2 months in highest risk group to 44.7 months in lowest risk group) compared to prior classifications. Moreover, in the neoadjuvant treatment context, there was lower expression of classical-like phenotypes in malignant cells in favor of basal-like phenotypes associated with TNF-NFkB and interferon signaling as well as the presence of novel acinar and neuroendocrine classical-like states, which may be more resilient to cytotoxic treatment. These results suggest that differentiated endodermal phenotypes are only prevalent enough to be detected under treatment selection pressure and when observed in treatment-naïve bulk studies, may reflect normal cell contamination. Spatially-resolved transcriptomics revealed an association between malignant cells expressing basal-like programs and higher immune infiltration with increased lymphocytic content, whereas those exhibiting classical-like programs were linked to sparser macrophage-predominant microniches, perhaps pointing to distinct therapeutic susceptibilities. Our refined molecular taxonomy and spatial resolution may help advance precision oncology in PDAC through informative stratification in clinical trials and insights into differential therapeutic targeting leveraging the immune system. Citation Format: William L. Hwang, Karthik A. Jagadeesh, Jimmy A. Guo, Hannah I. Hoffman, Eugene Drokhlyansky, Nicholas Van Wittenberghe, Samouil Farhi, Denis Schapiro, Jason Reeves, Daniel R. Zollinger, George Eng, Jason M. Schenkel, William A. Freed-Pastor, Clifton Rodrigues, Domenic Abbondanza, Debora Ciprani, Jennifer Y. Wo, Theodore S. Hong, Andrew J. Aguirre, Orit Rozenblatt-Rosen, Mari Mino-Kenudson, Carlos Fernandez-del Castillo, Andrew S. Liss, Tyler E. Jacks, Aviv Regev. Single-nucleus and spatial transcriptomics of archival pancreatic ductal adenocarcinoma reveals multi-compartment reprogramming after neoadjuvant treatment [abstract]. In: Proceedings of the AACR Virtual Special Conference on Pancreatic Cancer; 2020 Sep 29-30. Philadelphia (PA): AACR; Cancer Res 2020;80(22 Suppl):Abstract nr PR-007.

3621 Background: Tumor invasion of nerves, blood vessels, and lymphatics are a primary means of local recurrence and escape from the local microenvironment, resulting in metastases and poor clinical outcomes. However, the genetic drivers that are most pertinent to these malignant processes are not well understood, and few therapeutics successfully target perineural invasion (PNI) and lympho-vascular invasion (LVI). Identifying genetic drivers and biomarkers can be valuable for therapeutic targeting and prognostication. Methods: We analyzed surgical pathology reports and bulk RNA-seq data of 1,624 patients across 12 cancer types from The Cancer Genome Atlas (TCGA). Differential gene expression analysis between patients with and without PNI/LVI was performed using DEseq2 in Python while adjusting for age, sex, race, and cancer type. Genes with an adjusted p-value < 0.001 were then used to derive parsimonious signatures using random forest classifier and recursive feature selection algorithms. Results: To assess whether these invasive histological phenotypes have clinical ramifications, we examined outcomes data and found that patients with PNI or LVI have reduced overall (OS) and disease-free survival (DFS) ( p < 0.05) relative to those without. In addition, patients with both PNI and LVI have the lowest DFS from our pan-cancer analysis, suggesting that each may have non-redundant contributions to poor outcomes. From the differential gene expression analysis, we identified a set of 621 and 606 genes that were highly associated with PNI and LVI, respectively (padj < 0.001). Many of these genes such as TEKT5 (padj = 3.18 x 10−64), which is canonically associated with ciliary and flagellar microtubules, and SCRIB (padj = 1.60 x 10−21), which helps establish apico-basal cell polarity, have not been described previously in relevance to PNI and LVI, and warrant further scientific and clinical investigation. These genes were ultimately condensed into a signature that optimizes for both model simplicity and goodness of fit with up to 90% accuracy as determined by trials on both a logistic regression and neural network model. Conclusions: We concluded from a pan-cancer analysis that PNI and LVI are associated with poor outcomes, and we were able to robustly identify sets of genes that characterize each invasive mechanism for further functional investigation.

4633 Background: Pancreatic ductal adenocarcinoma (PDAC) remains a treatment-refractory disease and existing molecular subtypes do not inform clinical decisions. Previously identified bulk transcriptomic subtypes of PDAC were often unintentionally driven by “contaminating” stroma. RNA extraction from pancreatic tissue is difficult and prior single-cell RNA-seq efforts have been limited by suboptimal dissociation/RNA quality and poor performance in the setting of neoadjuvant treatment. We developed a robust single-nucleus RNA-seq (sNuc-seq) technique compatible with frozen archival PDAC specimens. Methods: Single nuclei suspensions were extracted from frozen primary PDAC specimens (n = 27) derived from patients with (borderline)-resectable PDAC who underwent surgical resection with or without neoadjuvant chemoradiotherapy (CRT). Approximately 170,000 nuclei were processed with the 10x Genomics Single Cell 3’ v3 pipeline and gene expression libraries were sequenced (Illumina HiSeq X). Results: Distinct nuclei clusters with gene expression profiles/inferred copy number variation analysis consistent with neoplastic, acinar, ductal, fibroblast, endothelial, endocrine, lymphocyte, and myeloid populations were identified with proportions similar to corresponding multiplexed ion beam imaging. Non-negative matrix factorization revealed intra-tumoral heterogeneity shared across patients. Neoplastic cells featured eight distinct transcriptional topics characterized by developmental (epithelial, mesenchymal, endoderm progenitor, neural progenitor) and environmental (anabolic, catabolic, cycling, hypoxic) programs. CAFs exhibited four different transcriptional topics (activated/desmoplastic, myofibroblast, neurogenic, osteochondral). Differential gene expression and gene set enrichment analyses demonstrated that CRT was associated with an enrichment in myogenic programs in CAFs, activation pathways in immune cells, and type I/II interferons in malignant cells. CRT was also associated with a depletion in developmental programs within malignant cells. Conclusions: We uncovered significant intratumoral heterogeneity and treatment-associated differences in the malignant, fibroblast, and immune compartments of PDAC using sNuc-seq. Deconvolution of clinically-annotated bulk RNA-seq cohorts and characterization of intercellular interactions with receptor-ligand analysis and spatial transcriptomics are ongoing.