Yue and colleagues present a landmark anatomical atlas of human protein abundance across non-malignant tissues and cancers. We argue that precision oncology needs functional proteomic maps that add signalling activity, post-translational modifications, pathology-guided compartment resolution, metastatic and longitudinal sampling, clinical annotation and population diversity to make proteomics therapeutically actionable.

Malignant pleural mesothelioma (MPM) remains a rare but highly aggressive malignancy with limited treatment options and poor prognosis. For nearly twenty years, platinum–pemetrexed chemotherapy has persisted as the unchanged standard treatment; although recent progress in immunotherapy has modestly disrupted this therapeutic plateau, survival outcomes remain disappointingly limited. This review aims to provide a comprehensive overview of the epigenetic landscape of MPM, focusing particularly on the oncogenic and therapeutic implications of enhancer of zeste homolog 2 (EZH2), and to discuss its potential as a target for novel therapeutic strategies and combination regimens. Epigenetic dysregulation has emerged as a central driver of mesothelioma pathogenesis. EZH2, the catalytic component of the polycomb repressive complex 2 (PRC2), mediates histone H3K27 trimethylation, silencing tumor suppressor genes and promoting malignant transformation. In addition to its canonical role, EZH2 has non-canonical oncogenic effects that modulate transcription, apoptosis, DNA repair, and immune evasion. High EZH2 expression correlates with BAP1 loss, which enhances chromatin remodeling defects and disease aggressiveness. Preclinical and early clinical data demonstrate that EZH2 inhibitors—including tazemetostat, valemetostat, GSK126, EPZ011989, tulmimetostat, and novel PROTAC-based degraders such as MS1943—can suppress tumor progression, modulate the tumor immune microenvironment, and restore therapeutic sensitivity. Furthermore, combination approaches integrating EZH2 inhibition with chemotherapy or immune checkpoint blockade show synergistic potential in overcoming resistance. EZH2 represents a pivotal epigenetic regulator and a promising therapeutic target in MPM. Further understanding the dual canonical and non-canonical roles of EZH2 in tumor biology will be key to optimizing targeted and combinatorial treatment strategies. Future research should focus on translating EZH2 inhibition into clinical benefit, identifying predictive biomarkers of response, and exploring rational combinations with chemotherapy, targeted drugs, or immunotherapy to improve survival outcomes in mesothelioma patients.

BackgroundCaused by the parasite Trypanosoma cruzi, Chagas disease affects an estimated 7 million people globally. Diagnosis of Chagas disease in infants is urgently needed, as early detection allows for more effective treatment and reduced mortality. However, current diagnostics are inappropriate for effective detection in infants due to differences in the mechanism of disease in infants and the infant immune system, as well as lack of diagnostic sensitivity and loss to follow up. Studying peripheral biomarkers in urine can leverage physiological concentration in the bladder to increase yield of proteins secreted by pathogen, infected cells, or antigen processed by immune cells residing in different body sites.Principal findingsWe analyzed the urine of a cohort of infants who were congenitally infected with Chagas disease, using a method including affinity enrichment, mass spectrometry, and bioinformatics analysis to characterize the T. cruzi secreted peptidome. We identified 198 peptides specific for T. cruzi and analyzed them in light of their potential for diagnostic utility. Our protocol revealed that peptides of the hyper-mutating mucin-associated surface protein and trans-sialidase protein families could be identified in patient urine and can serve as diagnostic markers of disease. We developed antibodies against conserved regions of each protein and validated that these antibodies could be used to differentiate the urine of Chagas disease patients (N = 16 cases) from healthy controls (N = 19). By utilizing affinity enrichment sample preprocessing and anti-trans-sialidase and anti-MASP antibodies in tandem, we differentiated cases from controls with 87.5% sensitivity and 94.7% specificity.Conclusions/significanceOur work suggests that it is possible to detect Trypanosoma cruzi infection directly from a noninvasively collected fluid such as urine. A direct test in urine with this success rate would be well suited for rapid diagnosis in low-resource areas. Further studies to validate this approach are warranted.

Digital 3D mammography is currently the most utilized screening tool used to identify early-stage breast cancer. Mammograms are scored using the BI-RADS categories (Breast Imaging Reporting and Data System). Categories IV and V mammograms are associated with a significant risk of breast cancer, mandating a biopsy and a pathologic diagnosis. Image guided percutaneous needle biopsy diagnoses cancer in about 25% of these BIRADS IV, V abnormalities emphasizing the lack of specificity and sensitivity of 3d Digital Mammography in diagnosing early-stage breast cancer. The poor positive predictive score subjects 75% of these patients with benign radiographic abnormalities to the potential anxiety, cost and harms of an unnecessary tissue biopsy. We studied 150 patients with BIRADS IV, and V mammographic abnormalities with serum, plasma and saliva obtained just prior to image guided biopsy. Hydrogel affinity nanoparticles were used to harvest and concentrate low abundance proteins, and the proteins were then identified by mass spectrometry. Candidate peptides were identified, and these markers were verified in blinded confirmation. We developed and validated a statistical model to differentiate between cancer and non-cancer cases using a combination of proteomic and microbial biomarkers. Biomarkers were identified through bivariate analysis to find those with significant differences in abundance between cancer cases and controls. Additionally, lasso feature selection was employed to pinpoint the biomarkers most strongly associated with the cancer outcome. The model was constructed by comparing three distinct models using maximum likelihood estimates and Akaike information criterion, with the most parsimonious model selected for final analysis. K-fold cross-validation was used to ensure the model’s robustness, and a receiver operating characteristic (ROC) curve analysis yielded an area under the curve (AUC) of 0.82, indicating strong overall predictive accuracy. To determine risk categories, we identified thresholds using spline analysis and calibration models. Splines were applied to the predicted probabilities to detect significant changes in cancer likelihood, and calibration curves were used to fine-tune the threshold. This process identified a 0.5 probability cut-off, which was used to categorize predictions into low and high likelihood of cancer. Confusion matrices were then employed to evaluate model performance at this threshold. The model demonstrated high sensitivity (92%), effectively identifying true positive cases, but had a lower specificity (45%), indicating a higher rate of false positives. Despite challenges such as handling factor levels and missing values, the model, which incorporated nine significant biomarkers (seven proteomes and two microbes), proved valuable for identifying high-risk patients. The model’s high sensitivity potentially makes it a useful tool to guide the clinical decision to biopsy a mammographic abnormality or follow it clinically with additional interval follow-up imaging. An accurate risk marker can potentially reduce unnecessary invasive biopsies. To enhance clinical utility, future work should focus on improving specificity to reduce false positives and validating the model with larger, more diverse datasets to increase its generalizability and practical application. Citation Format: Kurayi Mahachi, Wendy Pelton, Angela Toepp, Alessandra Luchini, Lance Liotta, Richard Hoefer. Improving the Specificity and Sensitivity of Digital 3D Mammography using serum, plasma and saliva derived proteomes and mathematical modeling [abstract]. In: Proceedings of the San Antonio Breast Cancer Symposium 2024; 2024 Dec 10-13; San Antonio, TX. Philadelphia (PA): AACR; Clin Cancer Res 2025;31(12 Suppl):Abstract nr P1-05-25.

Background: Borreliosis infection and post-treatment Lyme disease syndrome are associated with nonspecific, multisystem symptoms, and there is currently no effective direct test for Lyme disease. The seroprevalence of Borrelia miyamotoi among individuals with suspected tick-borne illnesses and persistent symptoms is not well documented. Although bacterial extracellular vesicles (BEVs) have been suggested as potential sources of disease biomarkers, evidence supporting their role in Borrelia infection is lacking. Methods: Patients (N=66) with symptoms and with clinical suspicion of a tick-borne illness were enrolled in the study and followed longitudinally over the course of 1 year (min 1 visit, max 4 visits). The General Symptom Questionnaire (GSQ-30) was administered to the patients at each visit to measure symptom burden. Lyme and Borrelia miyamotoi serologies were conducted at LabCorp or Quest, under the standard of care. A mass spectrometry proteomic assay measuring Borrelia-specific peptides was conducted in the urine of the patients. Results: Among the analyzed patients, 65% had positive Lyme serology and 33% had positive B. miyamotoi serology with partial overlap (20%). Urinary Borrelia-specific peptides correlated with the GSQ-30 score after correcting for age, Lyme serology, and B. miyamotoi serology (coefficient = 0.57, t-test p-value < 0.00001). Urinary extracellular vesicles had positive reactivity in a p66 immunoassay, a known marker of Borrelia BEVs. Urinary peptides included glycerophosphodiester phosphodiesterase, variable small and large outer membrane proteins, the multicopy lipoprotein family, and Borrelia direct repeat proteins, which are also known markers of Borrelia BEVs. Conclusions: The B. miyamotoi seroprevalence (33%) in a symptomatic population with clinical suspicion of a tick-borne illness was higher than the seroprevalence estimated from previous studies. As it provides an objective measurement correlated with a standardized symptom burden score, urinary Borrelia peptide detection offers a promising avenue for future assay development. This proteomic analysis supported the presence of Borrelia BEVs in the urine of symptomatic patients with suspected tick-borne illnesses.

Mitophagy is a critically important survival mechanism in which toxic, aged, or defective mitochondria are segregated into mitophagosomes, which shuttle the damaged mitochondrial segments to the lysosome and proteasome for destruction. Cancer cells rely on mitophagy under conditions of high oxidative stress or increased energy demand. Oxidative stress can generate a large volume of damaged mitochondria, overwhelming lysosomal removal. Accumulated damaged mitochondria are toxic and their proper removal is crucial for maintaining mitochondrial health. We propose a new cancer cell mechanism for survival that is activated when the demand for segregating and eliminating damaged mitochondria exceeds the capacity of the lysosome or proteasome. Specifically, we show that tumor cells subjected to oxidative stress by carbonyl cyanide-3-chlorophenylhdrazone (CCCP) eliminate damaged mitochondria segments by bypassing the lysosome to export them outside the cell via extracellular vesicles (EVs), a process termed “secretory mitophagy”. PINK1, the initiator of mitophagy, remains associated with the damaged mitochondria that exported in EVs. Using several types of cancer cells, we show that tumor cells treated with CCCP can be induced to switch over to secretory mitophagy by treatment with Bafilomycin A1, which blocks the fusion of mitophagosomes with lysosomes. Under these conditions, an increased number of PINK1 + EVs are exported. This is associated with greater cell survival by a given CCCP dose, enhanced mitochondrial ATP production, and reduced mitochondrial oxidative damage (membrane depolarization). Our data supports the hypothesis that secretory mitophagy is a previously unexplored process by which cancer cells adapt to survive therapeutic or hypoxic stress. Ultimately, our findings may inform new prevention strategies targeting pre-malignant lesions and therapeutic approaches designed to sensitize tumor cells to oxidative stress-inducing therapies.

Abstract Lung cancer carcinogenesis is the outcome of a field of premalignant changes that occur in the bronchial tree leading to the overt emergence of the malignant lung cancer (field cancerization). A tissue or blood proteomic signature of these premalignant changes represent a novel concept of risk biomarkers that can predict the probability of future emergence of Lung Cancer. In this study, we analyzed two independent cohorts of non-small cell lung cancer (NSCLC) patients. One cohort consisted of 18 patients who donated blood before and after tumor resection surgery and matched tissue samples (UAMS Medical Center). The second set derived from a longitudinal case-control study of plasma sample collected during the physician health study that were donated at different times prior to the diagnosis of Non-Small Cell Lung Cancer (N=272). We used affinity-enriched mass spectrometry analysis and statistical learning techniques to identify a set of common markers that discriminated plasma cases and controls and were conserved in different bronchial tissue compartments including tumor and adjacent tissue. The tumor microenvironment adjacent to and surrounding cancer cells undergoes to progressive stages of carcinogenesis during field cancerization. We identified 22 proteins involved in cell adhesion, protein folding, immune response, angiogenesis, and cancer proliferation. This is a novel class of functional risk biomarkers to determine the probability of NSCLC onset. In conclusion, our findings shed light on a new class of functional biomarkers that underly tumor pathogenesis mechanisms and can provide hints for new strategies of tumor prevention therapy or lifestyle modification. Citation Format: Rayan Ibrahim Alhammad, Ngoc Bao Vuong, Weidong Zhou, Donald J Johann Jr, Barbara Birkaya, Lance Liotta, Alessandra Luchini. Field cancerization proteomic analysis provides a novel class of functional risk biomarkers for non small cell lung cancer [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2024; Part 1 (Regular Abstracts); 2024 Apr 5-10; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2024;84(6_Suppl):Abstract nr 2448.

As the economic burden associated with vision loss and ocular damage continues to rise, there is a need to explore novel treatment strategies. Extracellular vesicles (EVs) are enriched with various biological cargo, and there is abundant literature supporting the reparative and immunomodulatory properties of stem cell EVs across a broad range of pathologies. However, one area that requires further attention is the reparative effects of stem cell EVs in the context of ocular damage. Additionally, most of the literature focuses on EVs isolated from primary stem cells; the use of EVs isolated from human telomerase reverse transcriptase (hTERT)-immortalized stem cells has not been thoroughly examined. Using our large-scale EV-manufacturing platform, we reproducibly manufactured EVs from hTERT-immortalized mesenchymal stem cells (MSCs) and employed various methods to characterize and profile their associated cargo. We also utilized well-established cell-based assays to compare the effects of these EVs on both healthy and damaged retinal pigment epithelial cells. To the best of our knowledge, this is the first study to establish proof of concept for reproducible, large-scale manufacturing of hTERT-immortalized MSC EVs and to investigate their potential reparative properties against damaged retinal cells. The results from our studies confirm that hTERT-immortalized MSC EVs exert reparative effects in vitro that are similar to those observed in primary MSC EVs. Therefore, hTERT-immortalized MSCs may represent a more consistent and reproducible platform than primary MSCs for generating EVs with therapeutic potential.

Plant-based adhesives, such as those made from wheat, have been prominently used for books and paper-based objects and are also used as conservation adhesives. Starch paste originates from starch granules, whereas flour paste encompasses the entire wheat endosperm proteome, offering strong adhesive properties due to gluten proteins. From a conservation perspective, understanding the precise nature of the adhesive is vital as the longevity, resilience, and reaction to environmental changes can differ substantially between starch- and flour-based pastes. We devised a proteomics method to discern the protein content of these pastes. Protocols involved extracting soluble proteins using 0.5 M NaCl and 30 mM Tris-HCl solutions and then targeting insoluble proteins, such as gliadins and glutenins, with a buffer containing 7 M urea, 2 M thiourea, 4% CHAPS, 40 mM Tris, and 75 mM DTT. Flour paste's proteome is diverse (1942 proteins across 759 groups), contrasting with starch paste's predominant starch-associated protein makeup (218 proteins in 58 groups). Transformation into pastes reduces proteomes' complexity. Testing on historical bookbindings confirmed the use of flour-based glue, which is rich in gluten and serpins. High levels of deamidation were detected, particularly for glutamine residues, which can impact the solubility and stability of the glue over time. The mass spectrometry proteomics data have been deposited to the ProteomeXchange, Consortium (http://proteomecentral.proteomexchange.org) via the MassIVE partner repository with the data set identifier MSV000093372 (ftp://MSV000093372@massive.ucsd.edu).

Human immunodeficiency virus type 1 (HIV-1) infection can result in HIV-associated neurocognitive disorder (HAND), a spectrum of disorders characterized by neurological impairment and chronic inflammation. Combined antiretroviral therapy (cART) has elicited a marked reduction in the number of individuals diagnosed with HAND. However, there is continual, low-level viral transcription due to the lack of a transcription inhibitor in cART regimens, which results in the accumulation of viral products within infected cells. To alleviate stress, infected cells can release accumulated products, such as TAR RNA, in extracellular vesicles (EVs), which can contribute to pathogenesis in neighboring cells. Here, we demonstrate that cART can contribute to autophagy deregulation in infected cells and increased EV release. The impact of EVs released from HIV-1 infected myeloid cells was found to contribute to CNS pathogenesis, potentially through EV-mediated TLR3 (Toll-like receptor 3) activation, suggesting the need for therapeutics to target this mechanism. Three HIV-1 TAR-binding compounds, 103FA, 111FA, and Ral HCl, were identified that recognize TAR RNA and reduce TLR activation. These data indicate that packaging of viral products into EVs, potentially exacerbated by antiretroviral therapeutics, may induce chronic inflammation of the CNS observed in cART-treated patients, and novel therapeutic strategies may be exploited to mitigate morbidity.

Abstract Tumor cellular heterogeneity is a complex problem in cancer molecular diagnostics and personalized therapy. The tumor is a product of the different types and interactions of host and immune cells. Investigators have created a variety of methods to procure separate subpopulations of the tumor microenvironment for individual analysis. One of the most successful methods of this is laser capture microdissection (LCM). This method captures specific subpopulations of cells under direct microscopic visualization. This LCM technology has successfully been used for over 20 years, revealing a variety of insights into cancer pathogenesis and mechanisms of therapeutic response employed in numerous clinical trials. Nevertheless, there are two drawbacks to the current systems. For UV cutting (365 nm), the tissue is significantly damaged by the UV light energy. Infrared (808 nm) laser systems do not damage the tissue however, this method has low power and poor resolution. Our first hypothesis to be tested is that a new class of near UV 405 nm laser will have small capture areas approaching single cells while simultaneously preventing cellular damage. The second weakness of LCM is the requirement for an open-faced tissue section without a coverslip. Moreover, when visualizing the tissue of interest, the identification of regions of interest is obscured due to the refractive index mismatch between the laser microdissection cap and the tissue surface. We hypothesize that using a volatile organic compound with a refractive index similar to glass would elucidate the tissue region of interest to allow proper visualization. We successfully developed and tested a near 405 nm wavelength LCM system that utilizes a “liquid” coverslip to visualize the sample of interest with higher clarity, resolution, and contrast for immunohistochemistry, immuno-fluorescence, blood smear slides, and conventional hematoxylin-eosin tissue staining. Overall, these critical improvements LCM technology now allow for single-cell tissue capture, system automation, tissue visualization, and opens the door for artificial intelligence-assisted spatial profiling. Citation Format: Thomas Philipson, Marissa Howard, Kevin Johnson, Amanda Still, Furkat Yunusov, Emanuel Petricoin, Virginia Espina, Noel Gonzalez, Alan Carpino, Lance Liotta. Optimization of digital pathology through laser capture microdissection with a 405 nm laser [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2024; Part 1 (Regular Abstracts); 2024 Apr 5-10; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2024;84(6_Suppl):Abstract nr 3768.

Despite of massive emergence of molecular targeting drugs, the mainstay of advanced gastric cancer (GC) therapy is DNA-damaging drugs. Using a reverse-phase protein array-based proteogenomic analysis of a panel of 8 GC cell lines, we identified genetic alterations and signaling pathways, potentially associated with resistance to DNA-damaging drugs, including 5-fluorouracil (5FU), cisplatin, and etoposide. Resistance to cisplatin and etoposide, but not 5FU, was negatively associated with global copy number loss, vimentin expression, and caspase activity, which are considered hallmarks of previously established EMT subtype. The segregation of 19,392 protein expression time courses by sensitive and resistant cell lines for the drugs tested revealed that 5FU-resistant cell lines had lower changes in global protein dynamics, suggesting their robust protein level regulation, than their sensitive counterparts, whereas the cell lines that are resistant to other drugs showed increased protein dynamics in response to each drug. Despite faint global protein dynamics, 5FU-resistant cell lines showed increased signal transducer and activator of transcription 1 phosphorylation and PD-L1 expression in response to 5FU. In publicly available cohort data, expression of signal transducer and activator of transcription 1 and NFκB target genes induced by proinflammatory cytokines was associated with prolonged survival in GC. In our validation cohort, total lymphocyte count, rather than PD-L1 positivity, predicted a better relapse-free survival rate in GC patients with 5FU-based adjuvant chemotherapy than those with surgery alone. Moreover, total lymphocyte count+ patients who had no survival benefit from adjuvant chemotherapy were discriminated by expression of IκBα, a potent negative regulator of NFκB. Collectively, our results suggest that 5FU resistance observed in cell lines may be overcome by host immunity or by combination therapy with immune checkpoint blockade.

Despite success in the treatment of some blood cancers and melanoma, positive response to immunotherapies remains disappointingly low in the treatment of solid tumors. The context of the molecular crosstalk within the tumor microenvironment can result in dysfunctional immune cell activation, leading to tumor tolerance and progression. Although modulating these protein–protein interactions (PPIs) is vital for appropriate immune cell activation and recognition, targeting nonenzymatic PPIs has proven to be fraught with challenges. To address this, a synthetic, multivalent molecular modality comprised of small interfering peptides precisely hybridized to a semirigid DNA scaffold is introduced. Herein, a prototype of this modality that targets the IL‐33/ST2 signaling axis, which is associated with tumor tolerance and immunotherapy treatment failure is described. Using peptides that mimic the specific high‐energy “hotspot” residues with which the IL‐33/ST2 coreceptor, IL‐1RAcP, interacts with the initial binary complex, this platform is shown to effectively bind IL‐33/ST2 with a KD of 110 nm. Additionally, this molecule effectively abrogates signal transduction in cell models at high nanomolar concentrations and is exquisitely selective for this complex over structurally similar PPIs within the same cytokine superfamily.

The combination of advanced mass spectrometry and enrichment-based sample preparation methods has enhanced analytical capabilities in clinical proteomics. In this chapter, we describe a method of proteome analysis to identify Borrelia-derived peptides in urine that includes a sample affinity enrichment method coupled with liquid chromatography tandem mass spectrometry analysis and a bioinformatic peptide authentication algorithm.

Abstract Tumor metabolic reprogramming is a hallmark of cancer progression, survival, and therapeutic resistance. A targetable class of cancer metabolic adaptation exploits mitophagy known to be linked to the cancer phenotype. Mitophagy selectively eliminates dysfunctional mitochondria by targeting them, via autophagosome shuttling, to the lysosome for degradation. Cancer cell mitophagy is triggered by elevated oxidative stress and mitochondrial DNA damage caused by hypoxia, radiotherapy, and immunotherapy. A high mitophagy demand can overwhelm the lysosome capacity resulting in the accumulation of damaged mitochondria and can suppress biogenesis of healthy mitochondria. We hypothesize that the newly discovered secretory mitophagy mechanism exports damaged mitochondrial fission-released segments to reduce the overload pressure on the lysosomal system, and thereby sustains cancer cell survival in the face of therapeutic mitochondrial stress. We have discovered a form of secretory mitophagy occurring in vivo in a growing solid tumor. Our molecular analysis of the full repertoire of extracellular vesicles, EVs, shed into the resident tumor interstitial fluid, IF, in vivo yielded a rich set of information about the functional state of mitochondria within the tumor cells, and the host cells. A set of proteins required for sequential steps of fission-induced mitophagy preferentially populated the CD81 positive IF EVs, including PINK1, DRP1 and FIS1. The export of cellular mitochondria proteins to CD81 positive EVs was confirmed by density gradient isolation from the bulk EV isolate followed by anti-CD81 immunoprecipitation, Mito Tracker export into CD81 positive EVs, ultrastructural characterization and Exoview analyzer. Further, we stimulated mitochondrial oxidative stress and blocked the fusion of the mitophagosome with the lysosome, which markedly stimulated the export of the secretory mitophagy unit. The outcome is new understanding of the importance of secretory mitophagy that can constitute a therapeutic target, and a new clinically relevant means of monitoring the in vivo state of mitophagic flux within the tumor microenvironment by means of mitophagy derived EVs. Citation Format: Purva V. Gade, Marissa Howard, Angela Rojas, Layla Hasanzadah, Lance Liotta, Fatah Kashanchi. Monitoring metabolic plasticity in the tumor microenvironment in vivo by mitophagy generated extracellular vesicles [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2024; Part 1 (Regular Abstracts); 2024 Apr 5-10; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2024;84(6_Suppl):Abstract nr 7063.

Towards developing quantitative models of anti-tumor activities of macrophages, we evaluated the effects of cytokines, tumor exosomes, and polarization states of macrophages in a tumor microenvironemnt using a system of differential equations. We modeled the non-linear dynamics of macrophage polarization states (M0/M1/M2), tumor cell killing by macrophages, and evasion of macrophage mediated killing by tumor originated extracellular vesicle decoys. Solving these coupled differential equations using numerical approaches, showed that the rate of macrophage polarization into the M1 state is the critical determinant of anti-tumor activity mediated by M1 polarized macrophages. To determine what metabolomic factors correlate with the polarization of naive macrophage into anti-tumor M1 or pro-tumor M2 phenotypes, we performed LC/MS-based untargeted metabolomic analysis. Statistical analysis using Python-Scikit-learn was performed on the metabo-lomic data from naive, M1 or M2 polarized murine macrophages followed by multiple feature selection methods. Application of ensemble machine learning methods to both secreted and cell associated metabolites revealed novel molecules of fatty acid metabolism to be the main mediators of polarization. Integration of ensemble machine learning feature-ranking tools into our analysis of metabolomic data identified new potential targets in macrophage metabolism for enhancing anti-tumor activities. ### Competing Interest Statement The authors have declared no competing interest.

Flow cytometry MFI and % positive cells data for fluorescently labelled trastuzumab and pertuzumab in SKBR3, HCC1954, T47D and MCF-7 treated with TKIs