Length-scale study in deep learning prediction for non-small cell lung cancer brain metastasis.

Deep learning-assisted digital pathology has demonstrated the potential to profoundly impact clinical practice, even surpassing human pathologists in performance. However, as deep neural network (DNN) architectures grow in size and complexity, their explainability decreases, posing challenges in interpreting pathology features for broader clinical insights into physiological diseases. To better assess the interpretability of digital microscopic images and guide future microscopic system design, we developed a novel method to study the predictive feature length-scale that underpins a DNN's predictive power. We applied this method to analyze a DNN's capability in predicting brain metastasis from early-stage non-small-cell lung cancer biopsy slides. This study quantifies DNN's attention for brain metastasis prediction, targeting features at both the cellular scale and tissue scale in H&E-stained histological whole slide images. At the cellular scale, the predictive power of DNNs progressively increases with higher resolution and significantly decreases when the resolvable feature length exceeds 5 microns. Additionally, DNN uses more macro-scale features associated with tissue architecture and is optimized when assessing visual fields greater than 41 microns. Our study computes the length-scale requirements for optimal DNN learning on digital whole-slide microscopic images, holding the promise to guide future optical microscope designs in pathology applications and facilitating downstream deep learning analysis.

Impact of Stain Variation and Color Normalization for Prognostic Predictions in Pathology.

In recent years, deep neural networks (DNNs) have demonstrated remarkable performance in pathology applications, potentially even outperforming expert pathologists due to their ability to learn subtle features from large datasets. One complication in preparing digital pathology datasets for DNN tasks is variation in tinctorial qualities. A common way to address this is to perform stain normalization on the images. In this study, we show that a well-trained DNN model trained on one batch of histological slides failed to generalize to another batch prepared at a different time from the same tissue blocks, even when stain normalization methods were applied. This study used sample data from a previously reported DNN that was able to identify patients with early stage non-small cell lung cancer (NSCLC) whose tumors did and did not metastasize, with high accuracy, based on training and then testing of digital images from H&E stained primary tumor tissue sections processed at the same time. In this study we obtained a new series of histologic slides from the adjacent recuts of same tissue blocks processed in the same lab but at a different time. We found that the DNN trained on the either batch of slides/images was unable to generalize and failed to predict progression in the other batch of slides/images (AUCcross-batch = 0.52 - 0.53 compared to AUCsame-batch = 0.74 - 0.81). The failure to generalize did not improve even when the tinctorial difference correction were made through either traditional color-tuning or stain normalization with the help of a Cycle Generative Adversarial Network (CycleGAN) process. This highlights the need to develop an entirely new way to process and collect consistent microscopy images from histologic slides that can be used to both train and allow for the general application of predictive DNN algorithms.

Is Fragmentation of Prostate Core Biopsies Inevitable? A Quality Improvement Initiative.

CONTEXT.­: Core biopsies are standard of care for diagnosis and surveillance of prostate cancer. Fragmentation makes numeric assessment of cancer challenging and increases risk of inaccurate staging with direct implications on management. OBJECTIVE.­: To determine factors responsible for fragmentation at our institution. DESIGN.­: Prostate core biopsies performed at 2 hospital sites during 1 week were prospectively identified. Biopsies were received in multipart formalin jars, either mounted on a nonadherent dressing pad (Telfa, Medtronic Inc) or freely suspended, and grossed by experienced pathologists' assistants. Fragmentation was defined as the difference between number of cores sent by the clinician and number of cores counted by the pathologist on microscopy. RESULTS.­: Forty-six cases (15 benign; 31 malignant) with 535 specimen jars were identified of which 309 of 535 (57.8%) had >1 biopsy core per jar; 230 of 535 (43%) were received mounted on Telfa and 185 of 535 (34.6%) had histologic evidence of adenocarcinoma. Overall fragmentation rate was 157 of 535 (29.3%). Lowest fragmentation rate was seen when 1 core was submitted per jar regardless of mounting method (31 of 226; 14% for single versus 126 of 309; 41% for >1 per jar; P < .001). For 1 Telfa-mounted core, rate of fragmentation was 5 of 18 (27.8%) versus 26 of 203 (12.8%) when unmounted (P = .24). Significant increase in fragmentation of Telfa-mounted cores was seen when there were 3 per jar (32 of 70; 46% mounted fragmented versus 9 of 47; 19% unmounted fragmented specimens; P = .01). CONCLUSIONS.­: Submission of >1 biopsy core per jar and use of Telfa for mounting are associated with increased fragmentation. We recommend limiting submission to 1 core per jar and avoid mounting on Telfa pads.

AI-guided histopathology predicts brain metastasis in lung cancer patients.

Brain metastases can occur in nearly half of patients with early and locally advanced (stage I-III) non-small cell lung cancer (NSCLC). There are no reliable histopathologic or molecular means to identify those who are likely to develop brain metastases. We sought to determine if deep learning (DL) could be applied to routine H&E-stained primary tumor tissue sections from stage I-III NSCLC patients to predict the development of brain metastasis. Diagnostic slides from 158 patients with stage I-III NSCLC followed for at least 5 years for the development of brain metastases (Met+, 65 patients) versus no progression (Met-, 93 patients) were subjected to whole-slide imaging. Three separate iterations were performed by first selecting 118 cases (45 Met+, 73 Met-) to train and validate the DL algorithm, while 40 separate cases (20 Met+, 20 Met-) were used as the test set. The DL algorithm results were compared to a blinded review by four expert pathologists. The DL-based algorithm was able to distinguish the eventual development of brain metastases with an accuracy of 87% (p < 0.0001) compared with an average of 57.3% by the four pathologists and appears to be particularly useful in predicting brain metastases in stage I patients. The DL algorithm appears to focus on a complex set of histologic features. DL-based algorithms using routine H&E-stained slides may identify patients who are likely to develop brain metastases from those who will remain disease free over extended (>5 year) follow-up and may thus be spared systemic therapy. © 2024 The Authors. The Journal of Pathology published by John Wiley & Sons Ltd on behalf of The Pathological Society of Great Britain and Ireland.

The Liquid Biopsy Consortium: Challenges and opportunities for early cancer detection and monitoring.

The emerging field of liquid biopsy stands at the forefront of novel diagnostic strategies for cancer and other diseases. Liquid biopsy allows minimally invasive molecular characterization of cancers for diagnosis, patient stratification to therapy, and longitudinal monitoring. Liquid biopsy strategies include detection and monitoring of circulating tumor cells, cell-free DNA, and extracellular vesicles. In this review, we address the current understanding and the role of existing liquid-biopsy-based modalities in cancer diagnostics and monitoring. We specifically focus on the technical and clinical challenges associated with liquid biopsy and biomarker development being addressed by the Liquid Biopsy Consortium, established through the National Cancer Institute. The Liquid Biopsy Consortium has developed new methods/assays and validated existing methods/technologies to capture and characterize tumor-derived circulating cargo, as well as addressed existing challenges and provided recommendations for advancing biomarker assays.

Extrinsic and intrinsic preanalytical variables affecting liquid biopsy in cancer.

Liquid biopsy, through isolation and analysis of disease-specific analytes, has evolved as a promising tool for safe and minimally invasive diagnosis and monitoring of tumors. It also has tremendous utility as a companion diagnostic allowing detection of biomarkers in a range of cancers (lung, breast, colon, ovarian, brain). However, clinical implementation and validation remains a challenge. Among other stages of development, preanalytical variables are critical in influencing the downstream cellular and molecular analysis of different analytes. Although considerable progress has been made to address these challenges, a comprehensive assessment of the impact on diagnostic parameters and consensus on standardized and optimized protocols is still lacking. Here, we summarize and critically evaluate key variables in the preanalytical stage, including study population selection, choice of biofluid, sample handling and collection, processing, and storage. There is an unmet need to develop and implement comprehensive preanalytical guidelines on the optimal practices and methodologies.

Automatic detection of circulating tumor cells and cancer associated fibroblasts using deep learning.

Circulating tumor cells (CTCs) and cancer-associated fibroblasts (CAFs) from whole blood are emerging as important biomarkers that potentially aid in cancer diagnosis and prognosis. The microfilter technology provides an efficient capture platform for them but is confounded by two challenges. First, uneven microfilter surfaces makes it hard for commercial scanners to obtain images with all cells in-focus. Second, current analysis is labor-intensive with long turnaround time and user-to-user variability. Here we addressed the first challenge through developing a customized imaging system and data pre-processing algorithms. Utilizing cultured cancer and CAF cells captured by microfilters, we showed that images from our custom system are 99.3% in-focus compared to 89.9% from a top-of-the-line commercial scanner. Then we developed a deep-learning-based method to automatically identify tumor cells serving to mimic CTC (mCTC) and CAFs. Our deep learning method achieved precision and recall of 94% (± 0.2%) and 96% (± 0.2%) for mCTC detection, and 93% (± 1.7%) and 84% (± 3.1%) for CAF detection, significantly better than a conventional computer vision method, whose numbers are 92% (± 0.2%) and 78% (± 0.3%) for mCTC and 58% (± 3.9%) and 56% (± 3.5%) for CAF. Our custom imaging system combined with deep learning cell identification method represents an important advance on CTC and CAF analysis.

Circulating Tumor Cells from Enumeration to Analysis: Current Challenges and Future Opportunities.

Circulating tumor cells (CTCs) have been recognized as a major contributor to distant metastasis. Their unique role as metastatic seeds renders them a potential marker in the circulation for early cancer diagnosis and prognosis as well as monitoring of therapeutic response. In the past decade, researchers mainly focused on the development of isolation techniques for improving the recovery rate and purity of CTCs. These developed techniques have significantly increased the detection sensitivity and enumeration accuracy of CTCs. Currently, significant efforts have been made toward comprehensive molecular characterization, ex vivo expansion of CTCs, and understanding the interactions between CTCs and their associated cells (e.g., immune cells and stromal cells) in the circulation. In this review, we briefly summarize existing CTC isolation technologies and specifically focus on advances in downstream analysis of CTCs and their potential applications in precision medicine. We also discuss the current challenges and future opportunities in their clinical utilization.

Prognostic markers in pT3 bladder cancer: A study from the international bladder cancer tissue microarray project.

PURPOSE: We evaluated the prognostic value of 10 putative tumor markers by immunohistochemistry in a large multi-institutional cohort of patients with locally advanced urothelial cancer of the bladder (UCB) with the aim to validate their clinical value and to harmonize protocols for their evaluation. MATERIALS AND METHODS: Primary tumor specimens from 576 patients with pathologic (p)T3 UCB were collected from 24 institutions in North America and Europe. Three replicate 0.6-mm core diameter samples were collected for the construction of a tissue microarray (TMA). Immunohistochemistry (IHC) for 10 previously described tumor markers was performed and scored at 3 laboratories independently according to a standardized protocol. Associations between marker positivity and freedom from recurrence (FFR) or overall survival (OS) were analyzed separately for each individual laboratory using Cox regression analysis. RESULTS: The overall agreement of the IHC scoring among laboratories was poor. Correlation among the 3 laboratories varied across the 10 markers. There was generally a lack of association between the individual markers and FFR or OS. The number of altered cell cycle regulators (p53, Rb, and p21) was associated with increased risk of cancer recurrence (P < 0.032). There was no clear pattern in the relationship between the percentage of markers altered in an 8-marker panel and FFR or OS. CONCLUSIONS: This large international TMA of locally advanced (pT3) UCB suggests that altered expression of p53, Rb, and p21 is associated with worse outcome. However this study also highlights limitations in the reproducibility of IHC even in the most expert hands.

Human-derived osteoblast-like cells and pericyte-like cells induce distinct metastatic phenotypes in primary breast cancer cells.

Approximately 70% of advanced breast cancer patients will develop bone metastases, which accounts for ∼90% of cancer-related mortality. Breast cancer circulating tumor cells (CTCs) establish metastatic tumors in the bone after a close interaction with local bone marrow cells including pericytes and osteoblasts, both related to resident mesenchymal stem/stromal cells (BM-MSCs) progenitors. In vitro recapitulation of the critical cellular players of the bone microenvironment and infiltrating CTCs could provide new insights into their cross-talk during the metastatic cascade, helping in the development of novel therapeutic strategies. Human BM-MSCs were isolated and fractionated according to CD146 presence. CD146+ cells were utilized as pericyte-like cells (PLCs) given the high expression of the marker in perivascular cells, while CD146- cells were induced into an osteogenic phenotype generating osteoblast-like cells (OLCs). Transwell migration assays were performed to establish whether primary breast cancer cells (3384T) were attracted to OLC. Furthermore, proliferation of 3384T breast cancer cells was assessed in the presence of PLC- and OLC-derived conditioned media. Additionally, conditioned media cultures as well as transwell co-cultures of each OLCs and PLCs were performed with 3384T breast cancer cells for gene expression interrogation assessing their induced transcriptional changes with an emphasis on metastatic potential. PLC as well as their conditioned media increased motility and invasion potential of 3384T breast cancer cells, while OLC induced a dormant phenotype, downregulating invasiveness markers related with migration and proliferation. Altogether, these results indicate that PLC distinctively drive 3384T cancer cells to an invasive and migratory phenotype, while OLC induce a quiescence state, thus recapitulating the different phases of the in vivo bone metastatic process. These data show that phenotypic responses from metastasizing cancer cells are influenced by neighboring cells at the bone metastatic niche during the establishment of secondary metastatic tumors.

Untargeted Assessment of Tumor Fractions in Plasma for Monitoring and Prognostication from Metastatic Breast Cancer Patients Undergoing Systemic Treatment.

The aim of this study was to assess the prognostic and predictive value of an untargeted assessment of tumor fractions in the plasma of metastatic breast cancer patients and to compare circulating tumor DNA (ctDNA) with circulating tumor cells (CTC) and conventional tumor markers. In metastatic breast cancer patients (n = 29), tumor fractions in plasma were assessed using the untargeted mFAST-SeqS method from 127 serial blood samples. Resulting z-scores for the ctDNA were compared to tumor fractions established with the recently published ichorCNA algorithm and associated with the clinical outcome. We observed a close correlation between mFAST-SeqS z-scores and ichorCNA ctDNA quantifications. Patients with mFAST-SeqS z-scores above three (34.5%) showed significantly worse overall survival (p = 0.014) and progression-free survival (p = 0.018) compared to patients with lower values. Elevated z-score values were clearly associated with radiologically proven progression. The baseline CTC count, carcinoembryonic antigen (CEA), and cancer antigen (CA)15-5 had no prognostic impact on the outcome of patients in the analyzed cohort. This proof of principle study demonstrates the prognostic impact of ctDNA levels detected with mFAST-SeqS as a very fast and cost-effective means to assess the ctDNA fraction without prior knowledge of the genetic landscape of the tumor. Furthermore, mFAST-SeqS-based ctDNA levels provided an early means of measuring treatment response.

Circulating Tumor Cells: Strategies for Capture, Analyses, and Propagation.

Circulating tumor cells (CTCs) play a central role in tumor dissemination and metastases, which are ultimately responsible for most cancer deaths. Technologies that allow for identification and enumeration of rare CTC from cancer patients' blood have already established CTC as an important clinical biomarker for cancer diagnosis and prognosis. Indeed, current efforts to robustly characterize CTC as well as the associated cells of the tumor microenvironment such as circulating cancer associated fibroblasts (cCAF), are poised to unmask key insights into the metastatic process. Ultimately, the clinical utility of CTC will be fully realized once CTC can be reliably cultured and proliferated as a biospecimen for precision management of cancer patients, and for discovery of novel therapeutics. In this review, we highlight the latest CTC capture and analyses technologies, and discuss in vitro strategies for culturing and propagating CTC.

Multigene methylation analysis of enriched circulating tumor cells associates with poor progression-free survival in metastatic breast cancer patients.

Blood-based biomarkers such as circulating tumor cells (CTCs) provide dynamic real-time assessment of molecular tumor characteristics beyond the primary tumor. The aim of this study was to evaluate the feasibility of a size-based microfilter to assess multigene methylation analysis of enriched CTCs in a prospective proof-of principle study. We examined the quantitative methylation status of nine genes (AKR1B1, BMP6, CST6, HOXB4, HIST1H3C, ITIH5, NEUROD1, RASSF1, SOX17) in enriched CTCs from metastatic breast cancer patients. Feasibility and clinical performance testing were assessed in a test set consisting of 37 patients and 25 healthy controls. With established cut-off values from the healthy control group, methylation of enriched CTCs was detected in at least one gene in 18/37 patients (48.6%), while 97.8% of all control samples were unmethylated. Patients with CTCs unmethylated for CST6, ITIH5, or RASSF1 showed significantly longer PFS compared to patients with corresponding enriched methylated CTCs. This proof-of-principle study shows the feasibility of a size-based microfilter to enrich and analyze multigene methylation profile of CTCs from metastatic breast cancer patients. For the first time, we report that multigene methylation analysis of enriched CTCs provides prognostic information in metastatic breast cancer patients.

Frequency and clinical impact of preoperative circulating tumor cells in resectable non-metastatic lung adenocarcinomas.

OBJECTIVES: Despite successful surgery, 30-50% of patients with resectable non-small cell lung cancer develop tumor recurrence within 5 years of surgery. MATERIALS AND METHODS: In this prospective study, we performed CTC enumerations in 40 patients with non-metastatic lung adenocarcinoma (NMLA) using a size-based microfilter. Additionally, cfDNA isolated from plasma was analyzed in 35 out of 40 patients. RESULTS: CTCs were identified in 15 out of 40 patients (37.5%) with a range of 1-44 cells, whereas mutated cfDNA was only detected in 3 out of 35 patients (8.6%). Disease-free survival (DFS) was significantly associated with CTC positivity (log-rank p=0.025), grading (log-rank p=0.019), tumor stage (log-rank p=0.025) and lymph node status (log-rank p=0.029). Multivariate analysis, including tumor stage and grading, showed that CTC positivity (p=0.006), grading (0.039) and tumor stage (p=0.022) were independently associated with DFS. CONCLUSION: Our study found that microfilter-based CTC enumeration in NMLA patients is an independent predictor of worse DFS. The used NGS-based cfDNA characterization had limited sensitivity to be clinically informative in our study cohort. CTC assessment before surgery can thus identify NMLA patients at high risk of disease recurrence.

Opioid Prescribing Practices and Training Needs of Québec Family Physicians for Chronic Noncancer Pain.

AIM: To examine medical practices and training needs of Québec family physicians with respect to pain management and opioid prescription for chronic noncancer pain (CNCP). METHODOLOGY: An online survey was carried out in 2016. RESULTS: Of 636 respondents (43.0% men; 54.3% ≥ 50 years old), 15.2% and 70.9% felt very or somewhat confident that they could properly prescribe opioids for CNCP. Concerns related to abuse (72.5% strongly/somewhat agree), dependence (73.2%), and lack of support (75.4%) were the main barriers reported. Only 19.7% always/often screened their patients for risks of abuse and dependence using a screening tool. About two-thirds of participants (65.7%) had recently (last five years) taken part in continuing education programs on opioid use for CNCP and 73.4% on CNCP management. Patient evaluation and differential diagnoses of chronic pain syndromes were rated as a top priority for further training. CONCLUSIONS: This study provides insights into Québec family physicians' concerns, practices, and needs with respect to the management of CNCP. Physicians' difficulties around the application of strategies to mitigate the problem of opioid abuse and addiction are worrying. The need to better train physicians in the field of pain and addiction cannot be emphasized enough.

Fourier Ptychographic Microscopy for Rapid, High-Resolution Imaging of Circulating Tumor Cells Enriched by Microfiltration.

Examining the hematogenous compartment for evidence of metastasis has increased significantly within the oncology research community in recent years, due to the development of technologies aimed at the enrichment of circulating tumor cells (CTCs), the subpopulation of primary tumor cells that gain access to the circulatory system and are responsible for colonization at distant sites. In contrast to other technologies, filtration-based CTC enrichment, which exploits differences in size between larger tumor cells and surrounding smaller, non-tumor blood cells, has the potential to improve CTC characterization through isolation of tumor cell populations with greater molecular heterogeneity. However, microscopic analysis of uneven filtration surfaces containing CTCs is laborious, time-consuming, and inconsistent, preventing widespread use of filtration-based enrichment technologies. Here, integrated with a microfiltration-based CTC and rare cell enrichment device we have previously described, we present a protocol for Fourier Ptychographic Microscopy (FPM), a method that, unlike many automated imaging platforms, produces high-speed, high-resolution images that can be digitally refocused, allowing users to observe objects of interest present on multiple focal planes within the same image frame. The development of a cost-effective and high-throughput CTC analysis system for filtration-based enrichment technologies could have profound clinical implications for improved CTC detection and analysis.

Understanding PSA and its derivatives in prediction of tumor volume: Addressing health disparities in prostate cancer risk stratification.

OBJECTIVES: To address health disparities in risk stratification of U.S. Hispanic/Latino men by characterizing influences of prostate weight, body mass index, and race/ethnicity on the correlation of PSA derivatives with Gleason score 6 (Grade Group 1) tumor volume in a diverse cohort. RESULTS: Using published PSA density and PSA mass density cutoff values, men with higher body mass indices and prostate weights were less likely to have a tumor volume <0.5 cm3. Variability across race/ethnicity was found in the univariable analysis for all PSA derivatives when predicting for tumor volume. In receiver operator characteristic analysis, area under the curve values for all PSA derivatives varied across race/ethnicity with lower optimal cutoff values for Hispanic/Latino (PSA=2.79, PSA density=0.06, PSA mass=0.37, PSA mass density=0.011) and Non-Hispanic Black (PSA=3.75, PSA density=0.07, PSA mass=0.46, PSA mass density=0.008) compared to Non-Hispanic White men (PSA=4.20, PSA density=0.11 PSA mass=0.53, PSA mass density=0.014). MATERIALS AND METHODS: We retrospectively analyzed 589 patients with low-risk prostate cancer at radical prostatectomy. Pre-operative PSA, patient height, body weight, and prostate weight were used to calculate all PSA derivatives. Receiver operating characteristic curves were constructed for each PSA derivative per racial/ethnic group to establish optimal cutoff values predicting for tumor volume ≥0.5 cm3. CONCLUSIONS: Increasing prostate weight and body mass index negatively influence PSA derivatives for predicting tumor volume. PSA derivatives' ability to predict tumor volume varies significantly across race/ethnicity. Hispanic/Latino and Non-Hispanic Black men have lower optimal cutoff values for all PSA derivatives, which may impact risk assessment for prostate cancer.

Do Black NonHispanic Men Produce Less Prostate Specific Antigen in Benign Prostate Tissue or Cancer Compared to White NonHispanic Men with Gleason Score 6 (Grade Group 1) Prostate Cancer?

PURPOSE: We evaluated prostate specific antigen production by benign prostate tissue and Gleason score 3+3=6 (Grade Group 1) prostate cancer in black and white nonHispanic men. MATERIALS AND METHODS: We used Gleason score 3+3=6 (Grade Group 1) cases to assess prostate specific antigen production by benign prostate tissue in cases with low volume cancer that did not influence prostate specific antigen and in those with high volume cancer in which gland weight did not influence prostate specific antigen. We then created age, prostate specific antigen and prostate weight adjusted cohorts to demonstrate tumor volume per 1 ng/ml prostate specific antigen and 1 µg prostate specific antigen mass. Prostate specific antigen density and prostate specific antigen mass density were used to adjust for prostate weight. RESULTS: Comparison of 58 black and 301 white men with low volume cancer demonstrated equal prostate specific antigen production by benign prostate tissue. Comparison of 30 black and 75 white men with high volume cancer indicated that prostate specific antigen was being driven by cancer volume, with lower prostate specific antigen production in black men. In the cohort of 54 black and 134 white men matched by age, prostate specific antigen and prostate weight, tumor volume per 1 ng/ml prostate specific antigen or 1 µg prostate specific antigen mass adjusted for prostate weight was 25% and 26% higher in black men, respectively. CONCLUSIONS: Benign prostate tissue produces equal amounts of prostate specific antigen in black and white men. Gleason score 3+3=6 (Grade Group 1) prostate cancer produces less prostate specific antigen in black men. These data should be considered for lowering prostate specific antigen and its derivatives in determining biopsy thresholds and for adjusting values for active surveillance criteria in black men.

Interactions between Adipocytes and Breast Cancer Cells Stimulate Cytokine Production and Drive Src/Sox2/miR-302b-Mediated Malignant Progression.

Consequences of the obesity epidemic on cancer morbidity and mortality are not fully appreciated. Obesity is a risk factor for many cancers, but the mechanisms by which it contributes to cancer development and patient outcome have yet to be fully elucidated. Here, we examined the effects of coculturing human-derived adipocytes with established and primary breast cancer cells on tumorigenic potential. We found that the interaction between adipocytes and cancer cells increased the secretion of proinflammatory cytokines. Prolonged culture of cancer cells with adipocytes or cytokines increased the proportion of mammosphere-forming cells and of cells expressing stem-like markers in vitro. Furthermore, contact with immature adipocytes increased the abundance of cancer cells with tumor-forming and metastatic potential in vivo. Mechanistic investigations demonstrated that cancer cells cultured with immature adipocytes or cytokines activated Src, thus promoting Sox2, c-Myc, and Nanog upregulation. Moreover, Sox2-dependent induction of miR-302b further stimulated cMYC and SOX2 expression and potentiated the cytokine-induced cancer stem cell-like properties. Finally, we found that Src inhibitors decreased cytokine production after coculture, indicating that Src is not only activated by adipocyte or cytokine exposures, but is also required to sustain cytokine induction. These data support a model in which cancer cell invasion into local fat would establish feed-forward loops to activate Src, maintain proinflammatory cytokine production, and increase tumor-initiating cell abundance and metastatic progression. Collectively, our findings reveal new insights underlying increased breast cancer mortality in obese individuals and provide a novel preclinical rationale to test the efficacy of Src inhibitors for breast cancer treatment.