Mapping three-dimensional intratumor proteomic heterogeneity in uterine serous carcinoma by multiregion microsampling.

BACKGROUND: Although uterine serous carcinoma (USC) represents a small proportion of all uterine cancer cases, patients with this aggressive subtype typically have high rates of chemotherapy resistance and disease recurrence that collectively result in a disproportionately high death rate. The goal of this study was to provide a deeper view of the tumor microenvironment of this poorly characterized uterine cancer variant through multi-region microsampling and quantitative proteomics. METHODS: Tumor epithelium, tumor-involved stroma, and whole "bulk" tissue were harvested by laser microdissection (LMD) from spatially resolved levels from nine USC patient tumor specimens and underwent proteomic analysis by mass spectrometry and reverse phase protein arrays, as well as transcriptomic analysis by RNA-sequencing for one patient's tumor. RESULTS: LMD enriched cell subpopulations demonstrated varying degrees of relatedness, indicating substantial intratumor heterogeneity emphasizing the necessity for enrichment of cellular subpopulations prior to molecular analysis. Known prognostic biomarkers were quantified with stable levels in both LMD enriched tumor and stroma, which were shown to be highly variable in bulk tissue. These USC data were further used in a comparative analysis with a data generated from another serous gynecologic malignancy, high grade serous ovarian carcinoma, and have been added to our publicly available data analysis tool, the Heterogeneity Analysis Portal ( https://lmdomics.org/ ). CONCLUSIONS: Here we identified extensive three-dimensional heterogeneity within the USC tumor microenvironment, with disease-relevant biomarkers present in both the tumor and the stroma. These data underscore the critical need for upfront enrichment of cellular subpopulations from tissue specimens for spatial proteogenomic analysis.

Signaling dynamics in coexisting monoclonal cell subpopulations unveil mechanisms of resistance to anti-cancer compounds.

BACKGROUND: Tumor heterogeneity is a main contributor of resistance to anti-cancer targeted agents though it has proven difficult to study. Unfortunately, model systems to functionally characterize and mechanistically study dynamic responses to treatment across coexisting subpopulations of cancer cells remain a missing need in oncology. METHODS: Using single cell cloning and expansion techniques, we established monoclonal cell subpopulations (MCPs) from a commercially available epidermal growth factor receptor (EGFR)-mutant non-small cell lung cancer cell line. We then used this model sensitivity to the EGFR inhibitor osimertinib across coexisting cell populations within the same tumor. Pathway-centered signaling dynamics associated with response to treatment and morphological characteristics of the MCPs were assessed using Reverse Phase Protein Microarray. Signaling nodes differentially activated in MCPs less sensitive to treatment were then pharmacologically inhibited to identify target signaling proteins putatively implicated in promoting drug resistance. RESULTS: MCPs demonstrated highly heterogeneous sensitivities to osimertinib. Cell viability after treatment increased > 20% compared to the parental line in selected MCPs, whereas viability decreased by 75% in other MCPs. Reduced treatment response was detected in MCPs with higher proliferation rates, EGFR L858R expression, activation of EGFR binding partners and downstream signaling molecules, and expression of epithelial-to-mesenchymal transition markers. Levels of activation of EGFR binding partners and MCPs' proliferation rates were also associated with response to c-MET and IGFR inhibitors. CONCLUSIONS: MCPs represent a suitable model system to characterize heterogeneous biomolecular behaviors in preclinical studies and identify and functionally test biological mechanisms associated with resistance to targeted therapeutics.

The KRAS-regulated kinome identifies WEE1 and ERK coinhibition as a potential therapeutic strategy in KRAS-mutant pancreatic cancer.

Oncogenic KRAS drives cancer growth by activating diverse signaling networks, not all of which have been fully delineated. We set out to establish a system-wide profile of the KRAS-regulated kinase signaling network (kinome) in KRAS-mutant pancreatic ductal adenocarcinoma (PDAC). We knocked down KRAS expression in a panel of six cell lines and then applied multiplexed inhibitor bead/MS to monitor changes in kinase activity and/or expression. We hypothesized that depletion of KRAS would result in downregulation of kinases required for KRAS-mediated transformation and in upregulation of other kinases that could potentially compensate for the deleterious consequences of the loss of KRAS. We identified 15 upregulated and 13 downregulated kinases in common across the panel of cell lines. In agreement with our hypothesis, all 15 of the upregulated kinases have established roles as cancer drivers (e.g., SRC, TGF-ß1, ILK), and pharmacological inhibition of one of these upregulated kinases, DDR1, suppressed PDAC growth. Interestingly, 11 of the 13 downregulated kinases have established driver roles in cell cycle progression, particularly in mitosis (e.g., WEE1, Aurora A, PLK1). Consistent with a crucial role for the downregulated kinases in promoting KRAS-driven proliferation, we found that pharmacological inhibition of WEE1 also suppressed PDAC growth. The unexpected paradoxical activation of ERK upon WEE1 inhibition led us to inhibit both WEE1 and ERK concurrently, which caused further potent growth suppression and enhanced apoptotic death compared with WEE1 inhibition alone. We conclude that system-wide delineation of the KRAS-regulated kinome can identify potential therapeutic targets for KRAS-mutant pancreatic cancer.

Heterogeneous Off-Target Effects of Ultra-Low Dose Dimethyl Sulfoxide (DMSO) on Targetable Signaling Events in Lung Cancer In Vitro Models.

Targetable alterations in cancer offer novel opportunities to the drug discovery process. However, pre-clinical testing often requires solubilization of these drugs in cosolvents like dimethyl sulfoxide (DMSO). Using a panel of cell lines commonly used for in vitro drug screening and pre-clinical testing, we explored the DMSO off-target effects on functional signaling networks, drug targets, and downstream substrates. Eight Non-Small Cell Lung Cancer (NSCLC) cell lines were incubated with three concentrations of DMSO (0.0008%, 0.002%, and 0.004% v/v) over time. Expression and activation levels of 187 proteins, of which 137 were kinases and downstream substrates, were captured using the Reverse Phase Protein Array (RPPA). The DMSO effect was heterogeneous across cell lines and varied based on concentration, exposure time, and cell line. Of the 187 proteins measured, all were statistically different in at least one comparison at the highest DMSO concentration, followed by 99.5% and 98.9% at lower concentrations. Only 46% of the proteins were found to be statistically different in more than 5 cell lines, indicating heterogeneous response across models. These cell line specific alterations modulate response to in vitro drug screening. Ultra-low DMSO concentrations have broad and heterogeneous effects on targetable signaling proteins. Off-target effects need to be carefully evaluated in pre-clinical drug screening and testing.

The impact of ultraviolet- and infrared-based laser microdissection technology on phosphoprotein detection in the laser microdissection-reverse phase protein array workflow.

Reversible protein phosphorylation represents a key mechanism by which signals are transduced in eukaryotic cells. Dysregulated phosphorylation is also a hallmark of carcinogenesis and represents key drug targets in the precision medicine space. Thus, methods that preserve phosphoprotein integrity in the context of clinical tissue analyses are crucially important in cancer research. Here we investigated the impact of UV laser microdissection (UV LMD) and IR laser capture microdissection (IR LCM) on phosphoprotein abundance of key cancer signaling protein targets assessed by reverse-phase protein microarray (RPPA). Tumor epithelial cells from consecutive thin sections obtained from four high-grade serous ovarian cancers were harvested using either UV LMD or IR LCM methods. Phosphoprotein abundances for ten phosphoproteins that represent important drug targets were assessed by RPPA and revealed no significant differences in phosphoprotein integrity from those obtained using higher-energy UV versus the lower-energy IR laser methods.

Utilization of Proteomic Technologies for Precision Oncology Applications.

Genomic analysis of tumor specimens has revealed that cancer is fundamentally a proteomic disease at the functional level: driven by genomically defined derangements, but selected for in the proteins that are encoded and the aberrant activation of signaling and biochemical networks. This activation is measured by posttranslational modifications such as phosphorylation and other modifications that modulate cellular signaling, and these events cannot be effectively measured by genomic analysis alone. Moreover, these signaling networks by and large represent the targets for many FDA-approved and experimental molecularly targeted therapeutics. Consequently, it is important that we consider new classification schemas for oncology based not on tumor site of origin or histology under the microscope but on the functional protein signaling architecture. There are numerous proteomic technologies that could be discussed from a purely technological standpoint, but this chapter will concentrate on an overview of the main proteomic technologies available for conducting protein pathway activation analysis of clinical specimens such as multiplex immunoassays, phospho-specific flow cytometry, reverse phase protein microarrays, quantitative immunohistochemistry, and mass spectrometry. This chapter will focus on the application of these technologies to cancer-based clinical studies evaluating prognostic/predictive markers or for stratifying patients to personalized treatments.

An exploratory study examining how nano-liquid chromatography-mass spectrometry and phosphoproteomics can differentiate patients with advanced fibrosis and higher percentage collagen in non-alcoholic fatty liver disease.

BACKGROUND: Non-alcoholic steatohepatitis (NASH) is among the leading causes of liver disease worldwide. It is increasingly recognized that the phenotype of NASH may involve a number of different pathways, of which each could become important therapeutic targets. The aim of this study is to use high resolution mass spectrometry (MS) and phosphoproteomics techniques to assess the serum proteome and hepatic phosphoproteome in subjects with NASH-related fibrosis. METHODS: Sixty-seven biopsy-proven NAFLD subjects with frozen sera and liver tissue were included. Reverse phase protein microarray was used to quantify the phosphorylation of key signaling proteins in liver and nano-liquid chromatography (LC)-MS was used to sequence target biomarkers in the serum. An image analysis algorithm was used to quantify the percentage of collagen (% collagen) using computer-assisted morphometry. Using multiple regression models, serum proteomes and phosphorylated hepatic proteins that were independently (p ≤ 0.05) associated with advanced fibrosis (stage ≥ 2) and higher % collagen were assessed. RESULTS: Phosphorylated signaling pathways in the liver revealed that apoptosis signal-regulating kinase 1, mitogen-activated protein kinase (ASK1-MAPK pathway involving ASK1 S38 (p < 0.02) and p38 MAPK (p = 0.0002)) activated by the inflammatory cytokine interleukin (IL-10) (p < 0.001), were independently associated with higher % collagen. LC-MS data revealed that serum alpha-2 macroglobulin (α2M) (p = 0.0004) and coagulation factor V (p = 0.0127) were independently associated with higher % hepatic collagen. CONCLUSIONS: Simultaneous profiling of serum proteome and hepatic phosphoproteome reveals that the activation of ASK1 S38, p38 MAPK in the liver, and serum α2M and coagulation factor V are independently associated with hepatic collagen deposition in patients with NASH. These data suggest the role of these pathways in the pathogenesis of NASH-related fibrosis as a potential therapeutic target.

Protein network construction using reverse phase protein array data.

In this paper, we introduce a novel computational method for constructing protein networks based on reverse phase protein array (RPPA) data to identify complex patterns in protein signaling. The method is applied to phosphoproteomic profiles of basal expression and activation/phosphorylation of 76 key signaling proteins in three breast cancer cell lines (MCF7, LCC1, and LCC9). Temporal RPPA data are acquired at 48h, 96h, and 144h after knocking down four genes in separate experiments. These genes are selected from a previous study as important determinants for breast cancer survival. Interaction networks are constructed by analyzing the expression levels of protein pairs using a multivariate analysis of variance model. A new scoring criterion is introduced to determine relevant protein pairs. Through a network topology based analysis, we search for wiring patterns to identify key proteins that are associated with significant changes in expression levels across various experimental conditions.

Kinase-driven metabolic signalling as a predictor of response to carboplatin-paclitaxel adjuvant treatment in advanced ovarian cancers.

BACKGROUND: The biological mechanisms underlying early- and advanced-stage epithelial ovarian cancers (EOCs) are still poorly understood. This study explored kinase-driven metabolic signalling in early and advanced EOCs, and its role in tumour progression and response to carboplatin-paclitaxel treatment. METHODS: Tumour epithelia were isolated from two independent sets of primary EOC (n=72 and 30 for the discovery and the validation sets, respectively) via laser capture microdissection. Reverse phase protein microarrays were used to broadly profile the kinase-driven metabolic signalling of EOC with particular emphasis on the LBK1-AMPK and AKT-mTOR axes. Signalling activation was compared between early and advanced lesions, and carboplatin-paclitaxel-sensitive and -resistant tumours. RESULTS: Advanced EOCs were characterised by a heterogeneous kinase-driven metabolic signature and decreased phosphorylation of the AMPK-AKT-mTOR axis compared to early EOC (P<0.05 for AMPKα T172, AMPKα1 S485, AMPKß1 S108, AKT S473 and T308, mTOR S2448, p70S6 S371, 4EBP1 S65, GSK-3 α/ß S21/9, FOXO1 T24/FOXO3 T32, and FOXO1 S256). Advanced tumours with low relative activation of the metabolic signature and increased FOXO1 T24/FOXO3 T32 phosphorylation (P=0.041) were associated with carboplatin-paclitaxel resistance. CONCLUSIONS: If validated in a larger cohort of patients, the decreased AMPK-AKT-mTOR activation and phosphorylation of FOXO1 T24/FOXO3 T32 may help identify carboplatin-paclitaxel-resistant EOC patients.

Functional characterization of epithelial ovarian cancer histotypes by drug target based protein signaling activation mapping: implications for personalized cancer therapy.

Epithelial ovarian carcinoma (EOC) is a deadly disease, with a 5-year survival of 30%. The aim of the study was to perform broad-scale protein signaling activation mapping to evaluate if EOC can be redefined based on activated protein signaling network architecture rather than histology. Tumor cells were isolated using laser capture microdissection (LCM) from 72 EOCs. Tumors were classified as serous (n = 38), endometrioid (n = 13), mixed (n = 8), clear cell (CCC; n = 7), and others (n = 6). LCM tumor cells were lysed and subjected to reverse-phase protein microarray to measure the expression/activation level of 117 protein drug targets. Unsupervised hierarchical clustering analysis was utilized to explore the overall signaling network. ANOVA was used to detect significant differences among the groups (p < 0.05). Regardless of histology, unsupervised analysis revealed five pathway-driven clusters. When the EOC histotypes were compared by ANOVA, only CCC showed a distinct signaling network, with activation of EGFR, Syk, HER2/ErbB2, and SHP2 (p = 0.0007, p = 0.0021, p < 0.0001, and p = 0.0410, respectively). The histological classification of EOC fails to adequately describe the underpinning protein signaling network. Nevertheless, CCC presents unique signaling characteristics compared to the other histotypes. EOC may need to be characterized by functional signaling activation mapping rather than pure histology.

A pilot study utilizing multi-omic molecular profiling to find potential targets and select individualized treatments for patients with previously treated metastatic breast cancer.

The primary objective was to determine if multi-omic molecular profiling (MMP) informed selection of approved cancer treatments could change the clinical course of disease for patients with previously treated metastatic breast cancer (MBC) (i.e., produce a growth modulation index (GMI) ≥1.3). GMI was calculated as the ratio of progression free survival on MMP-selected therapy/time to progression on last prior treatment. To meet the primary objective at least 35 % of the subjects should demonstrate a GMI ≥1.3. Secondary endpoints included determining the response rate (according to RECIST 1.1), the percent of patients with non-progression at 4 months, and overall survival in patients whose therapy is selected by molecular profiling and proteomic analysis. Eligible patients had MBC, with ≥3 prior lines of therapy. A multi-omic based approach was performed incorporating multiplexed immunohistochemistry, c-DNA microarray, and phosphoprotein pathway activation mapping by reverse phase protein array. MMP was performed on fresh core biopsies; results were generated and sent to a Treatment Selection Committee (TSC) for review and treatment selection. Three sites enrolled 28 patients, of which 25 were evaluable. The median range of prior treatment was 7 (range 3-12). The MMP analysis and treatment recommendation were delivered within a median of 15.5 days from biopsy (range 12-23). The TSC selected MMP-rationalized treatment in 100 % (25/25) of cases. None of the MMP-based therapies were the same as what the clinician would have selected if the MMP had not been performed. GMI ≥1.3 was reported in 11/25 (44 %) patients. Partial responses were noted in 5/25 (20 %), stable disease in 8/25 (32 %) and 9/25 (36 %) had no progression at 4 months. This pilot study demonstrates the feasibility of finding possible treatments for patients with previously treated MBC using a multiplexed MMP-rationalized treatment recommendation. This MMP approach merits further investigation.

Stratification of clear cell renal cell carcinoma by signaling pathway analysis.

Investigation of cell signaling pathways in 16 clear cell renal cell carcinomas to identify groups based on commonly shared phosphorylation-driven signaling networks. Using laser capture microdissection and reverse-phase protein arrays, we profiled 75 key nodes spanning signaling pathways important in tumorigenesis. Analysis revealed significantly different (P < 0.05) signaling levels for 27 nodes between two groups of samples, designated A (4 samples; high EGFR, RET, and RASGFR1 levels, converging to activate AKT/mTOR) and B (12 samples; high ERK1/2 and STAT phosphorylation). Group B was further partitioned into groups C (7 samples; elevated expression of LC3B) and D (5 samples; activation of Src and STAT). Network analysis indicated that group A was characterized by signaling pathways related to cell cycle and proliferation, and group B by pathways related to cell death and survival. Homogeneous clear cell renal cell carcinomas could be stratified into at least two major functional groups.

Enhancing precision in colorectal cancer surgery: development of an LGR5-targeting RSPO1 peptide mimetic as a contrast agent for intraoperative fluorescence molecular imaging.

Colorectal cancer (CRC) is the third most common cancer worldwide. In the United States alone, CRC was responsible for approximately 52,550 deaths in 2023, with an estimated 153,020 new cases. CRC presents with synchronous peritoneal spread in 5-10% of patients, and up to 20-50% of patients with recurrent disease will develop metachronous colorectal cancer peritoneal metastatic (CRC-PM) disease. Eradication of the tumor, tumor margins and microscopic residual disease is paramount, as microscopic residual disease is associated with local recurrences, with 5-year survival rates of less than 35%. The success of resection and reduction of residual disease depends on the accuracy with which cancer cells and normal tissue can be intra-operatively distinguished. Fluorescence Molecular Imaging (IFMI) and tumor-targeted contrast agents represent a promising approach for intraoperative detection and surgical intervention. Proper target selection, the development of scalable imaging agents and enhanced real-time tumor and tumor microenvironment imaging are critical to enabling enhanced surgical resection. LGR5 (leucine-rich repeat-containing G-protein-coupled receptor 5), a colonic crypt stem cell marker and the receptor for the R-spondins (RSPO) in the Wnt signaling pathway, is also expressed on colorectal cancer stem cells (CSC) and on CRC tumors and metastases, suggesting it could be a useful target for imaging of CRC. However, there are numerous diverging reports on the role of LGR5 in CRC therapy and outcomes. Herein, we report on the synthesis and validation of a 37 amino acid RSPO1-mimetic peptide, termed RC18, that was specifically designed to access the R-spondin binding site of LGR5 to potentially be used for interoperative imaging of CRC-PM. The receptor-binding capabilities of the RC18 indicate that direct interactions with LGR5 neither significantly increased LGR5 signaling nor blocked RSPO1 binding and signal transduction, suggesting that the RSPO1-mimetic is functionally inert, making it an attractive contrast agent for intraoperative CRC-PM imaging.

Targeting HER2/HER3 co-mutations in metastatic breast cancer: Case reports of exceptional responders to trastuzumab and pertuzumab therapy.

BACKGROUND: Overexpression of HER2 plays an important role in cancer progression and is the target of multiple therapies in HER2-positive breast cancer. Recent studies have also highlighted the presence of activating mutations in HER2, and HER3 that are predicted to enhance HER2 downstream pathway activation in a HER2-dependent manner. METHODS: In this report, we present two exceptional responses in hormone receptor-positive, HER2-nonamplified, HER2/HER3 co-mutated metastatic breast cancer patients who were treated with the anti-HER2-directed monoclonal antibodies, trastuzumab and pertuzumab. RESULTS: Both patients acheived exceptional responses to treatment, suggesting that combined trastuzumab, pertuzumab, and endocrine therapy could be a highly effective therapy for these patients and our observations could help prioritize trastuzumab deruxtecan as an early therapeutic choice for patients whose cancers have activating mutations in HER2.

Mediator kinase inhibition impedes transcriptional plasticity and prevents resistance to ERK/MAPK-targeted therapy in KRAS-mutant cancers.

Acquired resistance remains a major challenge for therapies targeting oncogene activated pathways. KRAS is the most frequently mutated oncogene in human cancers, yet strategies targeting its downstream signaling kinases have failed to produce durable treatment responses. Here, we developed multiple models of acquired resistance to dual-mechanism ERK/MAPK inhibitors across KRAS-mutant pancreatic, colorectal, and lung cancers, and then probed the long-term events enabling survival against this class of drugs. These studies revealed that resistance emerges secondary to large-scale transcriptional adaptations that are diverse and cell line-specific. Transcriptional reprogramming extends beyond the well-established early response, and instead represents a dynamic, evolved process that is refined to attain a stably resistant phenotype. Mechanistic and translational studies reveal that resistance to dual-mechanism ERK/MAPK inhibition is broadly susceptible to manipulation of the epigenetic machinery, and that Mediator kinase, in particular, can be co-targeted at a bottleneck point to prevent diverse, cell line-specific resistance programs.

Determining the ERK-regulated phosphoproteome driving KRAS-mutant cancer.

To delineate the mechanisms by which the ERK1 and ERK2 mitogen-activated protein kinases support mutant KRAS-driven cancer growth, we determined the ERK-dependent phosphoproteome in KRAS-mutant pancreatic cancer. We determined that ERK1 and ERK2 share near-identical signaling and transforming outputs and that the KRAS-regulated phosphoproteome is driven nearly completely by ERK. We identified 4666 ERK-dependent phosphosites on 2123 proteins, of which 79 and 66%, respectively, were not previously associated with ERK, substantially expanding the depth and breadth of ERK-dependent phosphorylation events and revealing a considerably more complex function for ERK in cancer. We established that ERK controls a highly dynamic and complex phosphoproteome that converges on cyclin-dependent kinase regulation and RAS homolog guanosine triphosphatase function (RHO GTPase). Our findings establish the most comprehensive molecular portrait and mechanisms by which ERK drives KRAS-dependent pancreatic cancer growth.

Proteogenomic analysis of enriched HGSOC tumor epithelium identifies prognostic signatures and therapeutic vulnerabilities.

We performed a deep proteogenomic analysis of bulk tumor and laser microdissection enriched tumor cell populations from high-grade serous ovarian cancer (HGSOC) tissue specimens spanning a broad spectrum of purity. We identified patients with longer progression-free survival had increased immune-related signatures and validated proteins correlating with tumor-infiltrating lymphocytes in 65 tumors from an independent cohort of HGSOC patients, as well as with overall survival in an additional 126 HGSOC patient cohort. We identified that homologous recombination deficient (HRD) tumors are enriched in pathways associated with metabolism and oxidative phosphorylation that we validated in independent patient cohorts. We further identified that polycomb complex protein BMI-1 is elevated in HR proficient (HRP) tumors, that elevated BMI-1 correlates with poor overall survival in HRP but not HRD HGSOC patients, and that HRP HGSOC cells are uniquely sensitive to BMI-1 inhibition.

Obesity as a risk factor for triple-negative breast cancers: a systematic review and meta-analysis.

Triple-negative breast cancer (TNBC) is a subtype of breast tumor with unique characteristics in terms of clinical-pathological presentation, prognosis, and response to therapy. Epidemiological investigations focusing on the identification of risk factors involved in the onset and progression of TNBCs have identified unique demographic, anthropometric, and reproductive characteristics involved in the etiopathogenesis of this subtype of breast tumors. This systematic review and meta-analysis evaluates the association between TNBCs and obesity and menopause status. Eligible articles were identified through three databases and secondary reference analysis. The search was conducted from the first record to February 2012. Eleven original articles meeting a priori established inclusion criteria were incorporated in the quantitative analysis. Case-case and case-control comparisons were performed. In addition, a case-case comparison was conducted before and after stratification according to menopausal status. Based on the level of between-study heterogeneity, pooled odds ratio (OR) and 95 % confidence interval were calculated using fixed or random models. The case-case comparison showed a significant association between TNBC and obesity (OR: 1.20; 95 % CI: 1.03-1.40). These results were confirmed by the case-control comparison (OR: 1.24; 95 % CI: 1.06-1.46). Once stratification based on menopausal status was applied to the case-case analysis, significant results were observed only in the pre-menopausal group (OR: 1.43; 95 % CI: 1.23-1.65). According to this analysis, obese women are at a greater risk of presenting with a TNBC than non-obese women, and menopause status may be a mitigating factor. If validated, these findings should be taken into consideration for the development of targeted preventive programs.

Automated Classification of Lung Cancer Subtypes Using Deep Learning and CT-Scan Based Radiomic Analysis.

Artificial intelligence and emerging data science techniques are being leveraged to interpret medical image scans. Traditional image analysis relies on visual interpretation by a trained radiologist, which is time-consuming and can, to some degree, be subjective. The development of reliable, automated diagnostic tools is a key goal of radiomics, a fast-growing research field which combines medical imaging with personalized medicine. Radiomic studies have demonstrated potential for accurate lung cancer diagnoses and prognostications. The practice of delineating the tumor region of interest, known as segmentation, is a key bottleneck in the development of generalized classification models. In this study, the incremental multiple resolution residual network (iMRRN), a publicly available and trained deep learning segmentation model, was applied to automatically segment CT images collected from 355 lung cancer patients included in the dataset "Lung-PET-CT-Dx", obtained from The Cancer Imaging Archive (TCIA), an open-access source for radiological images. We report a failure rate of 4.35% when using the iMRRN to segment tumor lesions within plain CT images in the lung cancer CT dataset. Seven classification algorithms were trained on the extracted radiomic features and tested for their ability to classify different lung cancer subtypes. Over-sampling was used to handle unbalanced data. Chi-square tests revealed the higher order texture features to be the most predictive when classifying lung cancers by subtype. The support vector machine showed the highest accuracy, 92.7% (0.97 AUC), when classifying three histological subtypes of lung cancer: adenocarcinoma, small cell carcinoma, and squamous cell carcinoma. The results demonstrate the potential of AI-based computer-aided diagnostic tools to automatically diagnose subtypes of lung cancer by coupling deep learning image segmentation with supervised classification. Our study demonstrated the integrated application of existing AI techniques in the non-invasive and effective diagnosis of lung cancer subtypes, and also shed light on several practical issues concerning the application of AI in biomedicine.