Search

Age is among the main risk factors for cancer and any cancer study in adults is faced with an aging tissue and organism. Yet, pre-clinical studies are carried out using young mice and are not able address the impact of aging and associated comorbidities on disease biology and treatment outcomes. Here, we discuss the limitations of current mouse cancer models and suggest strategies for developing novel models to address these major gaps in knowledge and experimental approaches.

Deep learning utilizing convolutional neural networks (CNNs) applied to hematoxylineosin (HE)-stained slides numerically encodes histomorphological tumor features. Tumor heterogeneity is an emerging biomarker in colon cancer that is, captured by these features, whereas microsatellite instability (MSI) is an established biomarker traditionally assessed by immunohistochemistry or polymerase chain reaction.HE-stained slides from The Cancer Genome Atlas (TCGA) colon cohort are passed through the CNN. Resulting imaging features are used to cluster morphologically similar slide regions. Tile-level pairwise similarities are calculated and used to generate a tumor heterogeneity score (THS). Patient-level THS is then correlated with TCGA-reported biomarkers, including MSI-status.HE-stained images from 313 patients generated 534 771 tiles. Deep learning automatically identified and annotated cells by type and clustered morphologically similar slide regions. MSI-high tumors demonstrated significantly higher THS than MSS/MSI-low (p 0.001). THS was higher in MLH1-silent versus non-silent tumors (p 0.001). The sequencing derived MSIsensor score also correlated with THS (r = 0.51, p 0.0001).Deep learning provides spatially resolved visualization of imaging-derived biomarkers and automated quantification of tumor heterogeneity. Our novel THS correlates with MSI-status, indicating that with expanded training sets, translational tools could be developed that predict MSI-status using HE-stained images alone.

Although the route to generate microRNAs (miRNAs) is often depicted as a linear series of sequential and constitutive cleavages, we now appreciate multiple alternative pathways as well as diverse strategies to modulate their processing and function. Here, we identify an unusually profound regulatory role of conserved loop sequences in vertebrate pre-mir-144, which are essential for its cleavage by the Dicer RNase III enzyme in human and zebrafish models. Our data indicate that pre-mir-144 dicing is positively regulated via its terminal loop, and involves the ILF3 complex (NF90 and its partner NF45/ILF2). We provide further evidence that this regulatory switch involves reshaping of the pre-mir-144 apical loop into a structure that is appropriate for Dicer cleavage. In light of our recent findings that mir-144 promotes the nuclear biogenesis of its neighbor mir-451, these data extend the complex hierarchy of nuclear and cytoplasmic regulatory events that can control the maturation of clustered miRNAs.

Supplementary Table 9: Differentially expressed proteins in LDE vs. control treated Ptch;p53 SI-CSC and SD-CSC tumors

Supplementary Table 5: mutational profiles for Fsmo;GFAP-cre SD-CSC tumors

Supplementary figures and Tables 1 & 2

The emergence of treatment resistance significantly reduces the clinical utility of many effective targeted therapies. Although both genetic and epigenetic mechanisms of drug resistance have been reported, whether these mechanisms are stochastically selected in individual tumors or governed by a predictable underlying principle is unknown. Here, we report that the dependence of cancer stem cells (CSC), not bulk tumor cells, on the targeted pathway determines the molecular mechanism of resistance in individual tumors. Using both spontaneous and transplantable mouse models of sonic hedgehog (SHH) medulloblastoma treated with a SHH/Smoothened inhibitor (SMOi), sonidegib/LDE225, we show that genetic-based resistance occurs only in tumors that contain SHH-dependent CSCs. In contrast, SHH medulloblastomas containing SHH-dependent bulk tumor cells but SHH-independent CSCs (SI-CSC) acquire resistance through epigenetic reprogramming. Mechanistically, elevated proteasome activity in SMOi-resistant SI-CSC medulloblastomas alters the tumor cell maturation trajectory through enhanced degradation of specific epigenetic regulators, including histone acetylation machinery components, resulting in global reductions in H3K9Ac, H3K14Ac, H3K56Ac, H4K5Ac, and H4K8Ac marks and gene expression changes. These results provide new insights into how selective pressure on distinct tumor cell populations contributes to different mechanisms of resistance to targeted therapies. This insight provides a new conceptual framework to understand responses and resistance to SMOis and other targeted therapies.Significance:The mechanism by which individual tumors become resistant to targeted therapies is thought to be unpredictable. This study provides novel insights into how selective pressure on cancer stem versus bulk tumor cells drives distinct and predictable mechanisms of resistance to targeted therapies. This finding paves a way for future treatment strategies that incorporate anticipated resistance mechanisms in devising second-line therapies in a personalized manner.

Supplementary Table 8: DEG in LDE vs control in SD-CSC Ptch;p53 tumors

Supplementary Table 6: DEG in LDE vs control treated fSMO;GFAP-cre tumors

Supplementary Table 3: Mutational profiles for Ptch;p53 SD-CSC tumors

Supplementary Table 4: mutational profiles for Ptch;p53 SI-CSC tumors

Supplementary Table 7: DEG in LDE vs. control treated SI-CSC Ptch;p53 tumors

To cure a patient's cancer is to eradicate invasive cells from the ecosystem of the body. However, the ecologic complexity of this challenge is not well understood. Here we show how results from eradications of invasive mammalian species from islands—one of the few contexts in which invasive species have been regularly cleared—inform new research directions for treating cancer. We first summarize the epidemiologic characteristics of island invader eradications and cancer treatments by analyzing recent datasets from the Database of Invasive Island Species Eradications and The Cancer Genome Atlas, detailing the superior successes of island eradication projects. Next, we compare how genetic and environmental factors impact success in each system. These comparisons illuminate a number of promising cancer research and treatment directions, such as heterogeneity engineering as motivated by gene drives and adaptive therapy; multiscale analyses of how population heterogeneity potentiates treatment resistance; and application of ecological data mining techniques to high-throughput cancer data. We anticipate that interdisciplinary comparisons between tumor progression and invasive species would inspire development of novel paradigms to cure cancer.

Table S1: (A) Summary of success rates for six mammalian invasive species under different eradication strategies. 95% confidence interval (CI) was calculated using binomial proportion CI with Jeffreys prior; (B) Summary of success rates (Kaplan-Meier five-year survival rates) for different tumor types under different treatment options. 95% CI was calculated using exponential Greenwood CI for Kaplan-Meier estimator and beta product procedure.

Models that recapitulate the complexity of human tumors are urgently needed to develop more effective cancer therapies. We report a bank of human patient-derived xenografts (PDXs) and matched organoid cultures from tumors that represent the greatest unmet need: endocrine-resistant, treatment-refractory and metastatic breast cancers. We leverage matched PDXs and PDX-derived organoids (PDxO) for drug screening that is feasible and cost-effective with in vivo validation. Moreover, we demonstrate the feasibility of using these models for precision oncology in real time with clinical care in a case of triple-negative breast cancer (TNBC) with early metastatic recurrence. Our results uncovered a Food and Drug Administration (FDA)-approved drug with high efficacy against the models. Treatment with this therapy resulted in a complete response for the individual and a progression-free survival (PFS) period more than three times longer than their previous therapies. This work provides valuable methods and resources for functional precision medicine and drug development for human breast cancer.

The current standard of care for many patients with HER2-positive breast cancer is neoadjuvant chemotherapy in combination with anti-HER2 agents, based on HER2 amplification as detected by in situ hybridization (ISH) or protein immunohistochemistry (IHC). However, hematoxylin & eosin (H&E) tumor stains are more commonly available, and accurate prediction of HER2 status and anti-HER2 treatment response from H&E would reduce costs and increase the speed of treatment selection. Computational algorithms for H&E have been effective in predicting a variety of cancer features and clinical outcomes, including moderate success in predicting HER2 status. In this work, we present a novel convolutional neural network (CNN) approach able to predict HER2 status with increased accuracy over prior methods. We trained a CNN classifier on 188 H&E whole slide images (WSIs) manually annotated for tumor Regions of interest (ROIs) by our pathology team. Our classifier achieved an area under the curve (AUC) of 0.90 in cross-validation of slide-level HER2 status and 0.81 on an independent TCGA test set. Within slides, we observed strong agreement between pathologist annotated ROIs and blinded computational predictions of tumor regions / HER2 status. Moreover, we trained our classifier on pre-treatment samples from 187 HER2+ patients that subsequently received trastuzumab therapy. Our classifier achieved an AUC of 0.80 in a five-fold cross validation. Our work provides an H&E-based algorithm that can predict HER2 status and trastuzumab response in breast cancer at an accuracy that may benefit clinical evaluations.

The supplementary file contains additional methods, analyses, and details of workflows as referenced in the manuscript, including all supplementary figures and tables. A full table of contents and descriptions are provided within.

Purpose:Genomic and transcriptomic alterations during metastasis are considered to affect clinical outcome of colorectal cancers, but detailed clinical implications of metastatic alterations are not fully uncovered. We aimed to investigate the effect of metastatic evolution on in vivo treatment outcome, and identify genomic and transcriptomic alterations associated with drug responsiveness.Experimental Design:We developed and analyzed patient-derived xenograft (PDX) models from 35 patients with colorectal cancer including 5 patients with multiple organ metastases (MOMs). We performed whole-exome, DNA methylation, and RNA sequencing for patient and PDX tumors. With samples from patients with MOMs, we conducted phylogenetic and subclonal analysis and in vivo drug efficacy test on the corresponding PDX models.Results:Phylogenetic analysis using mutation, expression, and DNA methylation data in patients with MOMs showed that mutational alterations were closely connected with transcriptomic and epigenomic changes during the tumor evolution. Subclonal analysis revealed that initial primary tumors with larger number of subclones exhibited more dynamic changes in subclonal architecture according to metastasis, and loco-regional and distant metastases occurred in a parallel or independent fashion. The PDX models from MOMs demonstrated therapeutic heterogeneity for targeted treatment, due to subclonal acquisition of additional mutations or transcriptomic activation of bypass signaling pathway during tumor evolution.Conclusions:This study demonstrated in vivo therapeutic heterogeneity of colorectal cancers using PDX models, and suggests that acquired subclonal alterations in mutations or gene expression profiles during tumor metastatic processes can be associated with the development of drug resistance and therapeutic heterogeneity of colorectal cancers.

Patient-derived xenografts (PDX) are tumor-in-mouse models for cancer. PDX collections, such as the NCI PDXNet, are powerful resources for preclinical therapeutic testing. However, variations in experimental and analysis procedures have limited interpretability. To determine the robustness of PDX studies, the PDXNet tested temozolomide drug response for three prevalidated PDX models (sensitive, resistant, and intermediate) across four blinded PDX Development and Trial Centers using independently selected standard operating procedures. Each PDTC was able to correctly identify the sensitive, resistant, and intermediate models, and statistical evaluations were concordant across all groups. We also developed and benchmarked optimized PDX informatics pipelines, and these yielded robust assessments across xenograft biological replicates. These studies show that PDX drug responses and sequence results are reproducible across diverse experimental protocols. In addition, we share the range of experimental procedures that maintained robustness, as well as standardized cloud-based workflows for PDX exome-sequencing and RNA-sequencing analyses and for evaluating growth.Significance:The PDXNet Consortium shows that PDX drug responses and sequencing results are reproducible across diverse experimental protocols, establishing the potential for multisite preclinical studies to translate into clinical trials.

Supplementary Materials and Methods, Supplementary Figures S1-S10 and Supplementary Tables S1-S6

Three-dimensional (3D) culture models, such as organoids, are flexible systems to interrogate cellular growth and morphology, multicellular spatial architecture, and cell interactions in response to drug treatment. However, new computational methods to segment and analyze 3D models at cellular resolution with sufficiently high throughput are needed to realize these possibilities. Here we reportCellos(Cell and Organoid Segmentation), an accurate, high throughput image analysis pipeline for 3D organoid and nuclear segmentation analysis.Cellossegments organoids in 3D using classical algorithms and segments nuclei using a Stardist-3D convolutional neural network which we trained on a manually annotated dataset of 3,862 cells from 36 organoids confocally imaged at 5 μm z-resolution. To evaluate the capabilities ofCelloswe then analyzed 74,450 organoids with 1.65 million cells, from multiple experiments on triple negative breast cancer organoids containing clonal mixtures with complex cisplatin sensitivities.Celloswas able to accurately distinguish ratios of distinct fluorescently labelled cell populations in organoids, with Celloswas able to recapitulate traditional luminescence-based drug response quantifications by analyzing 3D images, including parallel analysis of multiple cancer clones in the same well. Moreover,Celloswas able to identify organoid and nuclear morphology feature changes associated with treatment. Finally,Cellosenables 3D analysis of cell spatial relationships, which we used to detect ecological affinity between cancer cells beyond what arises from local cell division or organoid composition.Cellosprovides powerful tools to perform high throughput analysis for pharmacological testing and biological investigation of organoids based on 3D imaging.

3D patient tumor avatars (3D-PTAs) hold promise for next-generation precision medicine. Here, we describe the benefits and challenges of 3D-PTA technologies and necessary future steps to realize their potential for clinical decision making. 3D-PTAs require standardization criteria and prospective trials to establish clinical benefits. Innovative trial designs that combine omics and 3D-PTA readouts may lead to more accurate clinical predictors, and an integrated platform that combines diagnostic and therapeutic development will accelerate new treatments for patients with refractory disease.

Development of candidate cancer treatments is a resource-intensive process, with the research community continuing to investigate options beyond static genomic characterization. Toward this goal, we have established the genomic landscapes of 536 patient-derived xenograft (PDX) models across 25 cancer types, together with mutation, copy number, fusion, transcriptomic profiles, and NCI-MATCH arms. Compared with human tumors, PDXs typically have higher purity and fit to investigate dynamic driver events and molecular properties via multiple time points from same case PDXs. Here, we report on dynamic genomic landscapes and pharmacogenomic associations, including associations between activating oncogenic events and drugs, correlations between whole-genome duplications and subclone events, and the potential PDX models for NCI-MATCH trials. Lastly, we provide a web portal having comprehensive pan-cancer PDX genomic profiles and source code to facilitate identification of more druggable events and further insights into PDXs’ recapitulation of human tumors., Patient-derived xenograft models (PDX) have been extensively used to study the molecular and clinical features of cancers. Here the authors present a cohort of 536 PDX models from 25 cancers, as well as their genomic and evolutionary profiles and their suitability for clinical trials.

Personalized medicine holds great promise for improving cancer outcomes, yet there is a large inequity in the demographics of patients from whom genomic data and models, including patient derived xenografts (PDX), are developed and for whom treatments are optimized. In this study we develop a genetic ancestry pipeline for the Cancer Genomics Cloud, which we use to assess the diversity of models currently available in the National Cancer Institute (NCI) supported PDX Development and Trial Centers Research Network (PDXNet). We show that there is an over-representation of models derived from patients of European ancestry, which is consistent with other cancer model resources. We discuss these findings in the context of disparities in cancer incidence and outcomes among demographic groups in the US. For example, for the top cancer health disparities affecting African Americans and Latinos, there is a significant lack of ethnic/race appropriate models needed to advance pre-clinical research and personalized clinical treatment. For stomach and liver tumors, which represent disparities in these two minority populations, there are only three available models derived from patients from such backgrounds. Fortunately, ongoing NCI-funded efforts in minority focused PDXNet centers are actively addressing these gaps. We further discuss these results in the context of power analyses to highlight the immediate need for the development of models from minority populations to address cancer health equity in personalized medicine. Citation Format: Brian J. Sanderson, Paul Lott, Katherine Chiu, Juanita Elizabeth Quino, April Pangia Vang, Michael W. Lloyd, PDXNet Consortium, Anuj Srivastava, Jeffrey H. Chuang, Luis G. Carvajal-Carmona. Development and application of genetic ancestry reconstruction methods to study diversity of patient-derived models in the NCI PDXNet Consortium [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 1946.

Over half of BRAF mutant melanoma patients with response to targeted therapy recur within 15 months. Resistance is postulated to arise from certain clonal populations that enter a slow cycling persister state, evade treatment with relative dormancy, and repopulate the tumor when reactivated. Via longitudinal profiling we define expression states of clonal populations and track their evolutionary progression. BRAF mutant patient derived melanoma xenografts were treated with BRAF/MEK inhibitors through maximum tumor size reduction and re-growth. Spatial transcriptomics (ST) was performed at 5 timepoints across 94 days of treatment, tracking clonal populations with intact tissue structure. Deep learning on H&E-stained images automatically detected histological features that co-localize with expression levels. A novel computational pipeline performed clustering, differential expression and pathway enrichment analysis, copy number variation detection, and pseudotemporal ordering, allowing spatial recreation of clonal phylogenies and cell fate trajectory inference during the treatment course. Three clonal categories were defined; Sensitive: predominant in treatment-naïve samples, Persister: remaining clusters in maximally shrunken specimens, Resistant: re-emergent, fast-proliferating. Analysis of clonal composition identified GAPDH and CCND1 as highly expressed in sensitive clones, suppressed in persisters, and reactivated in resistant clones; TYRP1, DCT, MITF, and NGFR showed an opposite pattern. Persister clones evolved from oxidative phosphorylation toward glycolysis and MAPK regulation. Spatial analysis showed increased glycolysis:oxidative phosphorylation ratio in centrally located sensitive clusters, with more balanced peripheral expression. Resistant clones showed upregulated DUSP6 expression and enrichment in Orexin signaling and MAPK regulation. Imaging- and expression-based clones were highly concordant. Spatial profiling during melanoma treatment identified transient persister and emergent resistant clones, defining expression profiles and phenotypic features. Altered ratio of oxidative to glycolytic metabolic pathways was a hallmark of clonal evolution in both space and time. Specific MAPK pathway re-entry points were identified as candidate resistance mechanisms, suggesting potential therapeutic targets. Deep learning features derived from histological images showed good correlation with ST profiles, providing a promising method for clonal tracking via images alone. This longitudinal experiment mimics a tumor’s clinical course, inducing persistence and resistance via treatment that parallels that seen by patients. Combining ST and imaging techniques, we provide insight into clonal dynamics with novel spatiotemporal resolution, defining an evolutionary roadmap to acquired treatment resistance. Citation Format: Jill Carol Rubinstein, Sergii Domanskyi, Todd B. Sheridan, Brian J. Sanderson, SungHee Park, Jessica Kaster, Haiyin Li, Olga Anczukow, Meenhard Herlyn, Jeffrey H. Chuang. Spatiotemporal profiling defines persister and resistance signatures in targeted treatment of melanoma [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 1164.

High-resolution multiplexed tissue imaging such as imaging mass cytometry (IMC) has enabled quantitative cell characterization with preserved tissue architecture. However, the subcellular resolution information has not been fully leveraged yet, as most analyses have focused on cellular or compartment level features. Here, we have developed a robust computational approach integrating cellular spatial and molecular features in multiplexed images to quantify immune synapses. Our approach enables computational detection of subcellular enrichment of T cell co-receptors at immune synapses, which are key regulators of T cell functions in the tumor microenvironment (TME). We quantified immune synapse strength between T cells and various antigen-presenting cells in tumor regions in one melanoma whole tissue immunofluorescence dataset and two independent melanoma IMC datasets covering 99 patients. We observed that sub-localizations of immune synapse-related molecules are highly correlated, e.g. CD3 with CD8. These co-localizations are cell type-specific, with differential behaviors in CD4+ or CD8+ T subtypes. Intra-tumoral T-cell-myeloid synapses are associated with improved T cell proliferation (p = 2.8E-4). In addition, we also observed that cytotoxic T cell - melanoma synapses in immune rich regions are stronger in immune checkpoint inhibition therapy responders than in non-responding patients (p = 0.013). Our approach enables computational resolution of spatial inter-cellular interactions within the TME, is applicable across imaging modalities, and facilities assessment of biological and clinical significance. Citation Format: Zichao Liu, Victor G. Wang, Jan Martinek, Ali Foroughi pour, Jie Zhou, Karolina Palucka, Jeffrey H. Chuang. Computational analysis of immune synapses in melanoma tumor microenvironment [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 5883.

Patient-derived xenografts (PDXs) model human intra-tumoral heterogeneity in the context of the intact tissue of immunocompromised mice. Histological imaging via hematoxylin and eosin (H&E) staining is performed on PDX samples for routine assessment and, in principle, captures the complex interplay between tumor and stromal cells. Deep learning (DL)-based analysis of large human H&E image repositories has extracted inter-cellular and morphological signals correlated with disease phenotype and therapeutic response. Here, we present an extensive, pan-cancer repository of nearly 1,000 PDX and paired human progenitor H&E images. These images, curated from the PDXNet consortium, are associated with genomic and transcriptomic data, clinical metadata, pathological assessment of cell composition, and, in several cases, detailed pathological annotation of tumor, stroma, and necrotic regions. We demonstrate that DL can be applied to these images to classify tumor regions with an accuracy of 0.87. Further, we show that DL can predict xenograft-transplant lymphoproliferative disorder, the unintended outgrowth of human lymphocytes at the transplantation site, with an accuracy of 0.97. This repository enables PDX-specific investigations of cancer biology through histopathological analysis and contributes important model system data that expand on existing human histology repositories. We expect the PDXNet Image Repository to be valuable for controlled digital pathology analysis, both for the evaluation of technical issues such as stain normalization and for development of novel computational methods based on spatial behaviors within cancer tissues. Citation Format: Brian S. White, Xing Yi Woo, Soner Koc, Todd Sheridan, Steven B. Neuhauser, Shidan Wang, Yvonne A. Evrard, John David Landua, R Jay Mashl, Sherri R. Davies, Bingliang Fang, Maria Gabriela Raso, Kurt W. Evans, Matthew H. Bailey, Yeqing Chen, Min Xiao, Jill Rubinstein, Ali Foroughi pour, Lacey Elizabeth Dobrolecki, Maihi Fujita, Junya Fujimoto, Guanghua Xiao, Ryan C. Fields, Jacqueline L. Mudd, Xiaowei Xu, Melinda G. Hollingshead, Shahanawaz Jiwani, PDXNet consortium, Tiffany A. Wallace, Jeffrey A. Moscow, James H. Doroshow, Nicholas Mitsiades, Salma Kaochar, Chong-xian Pan, Moon S. Chen, Luis G. Carvajal-Carmona, Alana L. Welm, Bryan E. Welm, Ramaswamy Govindan, Shunqiang Li, Michael A. Davies, Jack A. Roth, Funda Meric-Bernstam, Yang Xie, Meenhard Herlyn, Li Ding, Michael T. Lewis, Carol J. Bolt, Dennis A. Dean, Jeffrey H. Chuang. A pan-cancer PDX histology image repository with genomic and pathological annotations for deep learning analysis. [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 5407.

Colorectal cancers are the fourth most commonly diagnosed cancer and the second leading cancer in number of deaths. Many clinical variables, pathological features, and genomic signatures are associated with patient risk, but reliable patient stratification in the clinic remains a challenging task. Here we assess how image, clinical, and genomic features can be combined to predict risk. We first observe that deep learning models based only on whole slide images (WSIs) from The Cancer Genome Atlas accurately separate high risk (OS5years, N=25) patients (AUC=0.81±0.08, 5year survival p-value=2.13e-25, 5year relative risk=5.09±0.05) though such models are less effective at predicting OS for moderate risk (3years

The emergence of treatment resistance significantly reduces the clinical utility of many effective targeted therapies. Although both genetic and epigenetic mechanisms of drug resistance have been reported, whether these mechanisms are stochastically selected in individual tumors or governed by a predictable underlying principle is unknown. Here, we report that the dependence of cancer stem cells (CSC), not bulk tumor cells, on the targeted pathway determines the molecular mechanism of resistance in individual tumors. Using both spontaneous and transplantable mouse models of sonic hedgehog (SHH) medulloblastoma treated with a SHH/Smoothened inhibitor (SMOi), sonidegib/LDE225, we show that genetic-based resistance occurs only in tumors that contain SHH-dependent CSCs. In contrast, SHH medulloblastomas containing SHH-dependent bulk tumor cells but SHH-independent CSCs (SI-CSC) acquire resistance through epigenetic reprogramming. Mechanistically, elevated proteasome activity in SMOi-resistant SI-CSC medulloblastomas alters the tumor cell maturation trajectory through enhanced degradation of specific epigenetic regulators, including histone acetylation machinery components, resulting in global reductions in H3K9Ac, H3K14Ac, H3K56Ac, H4K5Ac, and H4K8Ac marks and gene expression changes. These results provide new insights into how selective pressure on distinct tumor cell populations contributes to different mechanisms of resistance to targeted therapies. This insight provides a new conceptual framework to understand responses and resistance to SMOis and other targeted therapies. Significance: The mechanism by which individual tumors become resistant to targeted therapies is thought to be unpredictable. This study provides novel insights into how selective pressure on cancer stem versus bulk tumor cells drives distinct and predictable mechanisms of resistance to targeted therapies. This finding paves a way for future treatment strategies that incorporate anticipated resistance mechanisms in devising second-line therapies in a personalized manner.