Anticancer drug clustering in lung cancer based on gene expression profiles and sensitivity database

Applications of Machine Learning (ML) and Mathematical Modeling (MM) in Healthcare with Special Focus on Cancer Prognosis and Anticancer Therapy: Current Status and Challenges

The use of data-driven high-throughput analytical techniques, which has given rise to computational oncology, is undisputed. The widespread use of machine learning (ML) and mathematical modeling (MM)-based techniques is widely acknowledged. These two approaches have fueled the advancement in cancer research and eventually led to the uptake of telemedicine in cancer care. For diagnostic, prognostic, and treatment purposes concerning different types of cancer research, vast databases of varied information with manifold dimensions are required, and indeed, all this information can only be managed by an automated system developed utilizing ML and MM. In addition, MM is being used to probe the relationship between the pharmacokinetics and pharmacodynamics (PK/PD interactions) of anti-cancer substances to improve cancer treatment, and also to refine the quality of existing treatment models by being incorporated at all steps of research and development related to cancer and in routine patient care. This review will serve as a consolidation of the advancement and benefits of ML and MM techniques with a special focus on the area of cancer prognosis and anticancer therapy, leading to the identification of challenges (data quantity, ethical consideration, and data privacy) which are yet to be fully addressed in current studies.

An Introduction to Machine Learning

In the last few years, machine learning (ML) and artificial intelligence have seen a new wave of publicity fueled by the huge and ever-increasing amount of data and computational power as well as the discovery of improved learning algorithms. However, the idea of a computer learning some abstract concept from data and applying them to yet unseen situations is not new and has been around at least since the 1950s. Many of these basic principles are very familiar to the pharmacometrics and clinical pharmacology community. In this paper, we want to introduce the foundational ideas of ML to this community such that readers obtain the essential tools they need to understand publications on the topic. Although we will not go into the very details and theoretical background, we aim to point readers to relevant literature and put applications of ML in molecular biology as well as the fields of pharmacometrics and clinical pharmacology into perspective.

A transfer learning approach via procrustes analysis and mean shift for cancer drug sensitivity prediction

Transfer learning (TL) algorithms aim to improve the prediction performance in a target task (e.g. the prediction of cisplatin sensitivity in triple-negative breast cancer patients) via transferring knowledge from auxiliary data of a related task (e.g. the prediction of docetaxel sensitivity in breast cancer patients), where the distribution and even the feature space of the data pertaining to the tasks can be different. In real-world applications, we sometimes have a limited training set in a target task while we have auxiliary data from a related task. To obtain a better prediction performance in the target task, supervised learning requires a sufficiently large training set in the target task to perform well in predicting future test examples of the target task. In this paper, we propose a TL approach for cancer drug sensitivity prediction, where our approach combines three techniques. First, we shift the representation of a subset of examples from auxiliary data of a related task to a representation closer to a target training set of a target task. Second, we align the shifted representation of the selected examples of the auxiliary data to the target training set to obtain examples with representation aligned to the target training set. Third, we train machine learning algorithms using both the target training set and the aligned examples. We evaluate the performance of our approach against baseline approaches using the Area Under the receiver operating characteristic (ROC) Curve (AUC) on real clinical trial datasets pertaining to multiple myeloma, nonsmall cell lung cancer, triple-negative breast cancer, and breast cancer. Experimental results show that our approach is better than the baseline approaches in terms of performance and statistical significance.

Gemcitabine-Induced TIMP1 Attenuates Therapy Response and Promotes Tumor Growth and Liver Metastasis in Pancreatic Cancer

Gemcitabine constitutes one of the backbones for chemotherapy treatment in pancreatic ductal adenocarcinoma (PDAC), but patients often respond poorly to this agent. Molecular markers downstream of gemcitabine treatment in preclinical models may provide an insight into resistance mechanisms. Using cytokine arrays, we identified potential secretory biomarkers of gemcitabine resistance (response) in the transgenic KRasG12D; Trp53R172H; Pdx-1 Cre (KPC) mouse model of PDAC. We verified the oncogenic role of the cytokine tissue inhibitor of matrix metalloproteinases 1 (TIMP1) in primary pancreatic tumors and metastases using both in vitro techniques and animal models. We identified potential pathways affected downstream of TIMP1 using the Illumina Human H12 array. Our findings were validated in both primary and metastatic models of pancreatic cancer. Gemcitabine increased inflammatory cytokines including TIMP1 in the KPC mouse model. TIMP1 was upregulated in patients with pancreatic intraepithelial neoplasias grade 3 and PDAC lesions relative to matched normal pancreatic tissue. In addition, TIMP1 played a role in tumor clonogenic survival and vascular density, while TIMP1 inhibition resensitized tumors to gemcitabine and radiotherapy. We observed a linear relationship between TIMP-1 expression, liver metastatic burden, and infiltration by CD11b+Gr1+ myeloid cells and CD4+CD25+FOXP3+ Tregs, whereas the presence of tumor cells was required for immune cell infiltration. Overall, our results identify TIMP1 upregulation as a resistance mechanism to gemcitabine and provide a rationale for combining chemo/radiotherapy with TIMP1 inhibitors in PDAC. Cancer Res; 77(21); 5952-62. ©2017 AACR.

Recursive Random Lasso (RRLasso) for Identifying Anti-Cancer Drug Targets

Uncovering driver genes is crucial for understanding heterogeneity in cancer. L1-type regularization approaches have been widely used for uncovering cancer driver genes based on genome-scale data. Although the existing methods have been widely applied in the field of bioinformatics, they possess several drawbacks: subset size limitations, erroneous estimation results, multicollinearity, and heavy time consumption. We introduce a novel statistical strategy, called a Recursive Random Lasso (RRLasso), for high dimensional genomic data analysis and investigation of driver genes. For time-effective analysis, we consider a recursive bootstrap procedure in line with the random lasso. Furthermore, we introduce a parametric statistical test for driver gene selection based on bootstrap regression modeling results. The proposed RRLasso is not only rapid but performs well for high dimensional genomic data analysis. Monte Carlo simulations and analysis of the "Sanger Genomics of Drug Sensitivity in Cancer dataset from the Cancer Genome Project" show that the proposed RRLasso is an effective tool for high dimensional genomic data analysis. The proposed methods provide reliable and biologically relevant results for cancer driver gene selection.

MiR-379/411 cluster regulates IL-18 and contributes to drug resistance in malignant pleural mesothelioma

Malignant pleural mesothelioma (MPM) is a rapidly fatal malignancy that is increasing in incidence in Japan. In this study, we performed gene and microRNA (miRNA) expression profiling to identify novel therapeutic targets in MPM cells. Based on relative sensitivities to pemetrexed (PEM) and the histone deacetylase (HDAC) inhibitor, vorinostat (SAHA), 211H cells were determined to be the only sensitive MPM cell line out of the 6 tested. On the same series of cell lines, we performed whole genome transcriptomic profiling via DNA microarrays and pathway analysis of the derived data. Of particular note, IL-18 gene expression levels were significantly higher in the cell lines that were either drug resistant or displayed intermediate sensitivity, compared to the sensitive 211H cell line. Pathway analysis revealed IL-18 as an important gene associated with drug sensitivity of MPM cells. A relationship between IL-18 overexpression and drug resistance was also observed following targeted assessment of 10 cytokine genes using quantitative RT-PCR. miRNA expression profiles were evaluated in the MPM cell line panel in order to discern the mechanism of IL-18 induction in the drug-resistant lines. We found that miR-379 and miR-411 belonged to the same cluster of miRNAs located on chromosome 14q32 that commonly target the IL-18 gene. Luciferase reporter assays revealed that miR-379 and miR-411 directly target the IL-18 gene. Introduction of miR-379 plus miR-411, as well as IL-18 silencing, significantly suppressed the invasive capacity of MESO1 cells in vitro. Furthermore, the use of either PEM or SAHA together with miR-379 plus miR-411 mimics mediated increased sensitivity to these drugs in MESO1 cells. These results suggest that the miR-379/411 cluster may provide new therapeutic opportunities for advanced MPM patients, depending on the nature of IL-18 gene expression.

H-Prune through GSK-3β interaction sustains canonical WNT/β-catenin signaling enhancing cancer progression in NSCLC

H-Prune hydrolyzes short-chain polyphosphates (PPase activity) together with an hitherto cAMP-phosphodiesterase (PDE), the latest influencing different human cancers by its overexpression. H-Prune promotes cell migration in cooperation with glycogen synthase kinase-3 (Gsk-3β). Gsk-3β is a negative regulator of canonical WNT/β-catenin signaling. Here, we investigate the role of Gsk-3β/h-Prune complex in the regulation of WNT/β-catenin signaling, demonstrating the h-Prune capability to activate WNT signaling also in a paracrine manner, through Wnt3a secretion. In vivo study demonstrates that h-Prune silencing inhibits lung metastasis formation, increasing mouse survival. We assessed h-Prune levels in peripheral blood of lung cancer patients using ELISA assay, showing that h-Prune is an early diagnostic marker for lung cancer. Our study dissects out the mechanism of action of h-Prune in tumorigenic cells and also sheds light on the identification of a new therapeutic target in non-small-cell lung cancer.

An integrative analysis of the tumorigenic role of TAZ in human non-small cell lung cancer

PURPOSE

TAZ, also known as WWTR1, has recently been suggested as an oncogene in non-small cell lung cancer (NSCLC). We investigated the clinical relevance of TAZ expression and its functional role in NSCLC tumorigenesis.

EXPERIMENTAL DESIGN

We characterized TAZ at the DNA (n=192), mRNA (n=196), and protein levels (n=345) in an NSCLC patient cohort. Gene expression analysis was complemented by a meta-analysis of public datasets (n=1,382). The effects of TAZ on cell proliferation and cell cycle were analyzed in cell cultures and on tumor growth in mice. TAZ-dependent microarray-based expression profiles in NSCLC cells were combined with molecular profiles in human NSCLC tissues for in silico analysis.

RESULTS

Higher TAZ mRNA and protein levels were associated with shorter patient survival. Transduction of TAZ enhanced cell proliferation and tumorigenesis in bronchial epithelial cells, whereas TAZ silencing suppressed cell proliferation and induced cell cycle arrest in NSCLC cells. Microarray and cell culture experiments showed that ErbB ligands (amphiregulin, epiregulin, and neuregulin 1) are downstream targets of TAZ. Our in silico analysis revealed a TAZ signature that substantiated the clinical impact of TAZ and confirmed its relationship to the epidermal growth factor receptor signaling pathway.

CONCLUSION

TAZ expression defines a clinically distinct subgroup of patients with NSCLC. ErbB ligands are suggested to mediate the effects of TAZ on lung cancer progression. Our findings emphasize the tumorigenic role of TAZ and may serve as the basis for new treatment strategies.

Significance of osteopontin in the sensitivity of malignant pleural mesothelioma to pemetrexed

Pemetrexed (PEM) is currently recommended as one of the standard anticancer drugs for malignant pleural mesothelioma (MPM). However, the mechanism of the sensitivity of MPM to PEM remains unclear. We analyzed the antitumor effects of PEM in six MPM cell lines by MTS assay. To identify genes associated with drug sensitivity, we conducted gene expression profiling on the same set of cell lines using GeneChips and pathway analysis. Three cell lines were sensitive to PEM. A total fo 18 transcripts and 14 genes identified by GeneChips were significantly correlated with sensitivity to PEM. Pathway analysis revealed that osteopontin (SPP1/OPN) was an important target in PEM sensitivity. Overexpression of SPP1/OPN was observed in the sensitive cells by quantitative PCR and western blot analysis. Introduction of SPP1/OPN by lentiviral vector significantly enhanced the invasion activities of MPM cells. PEM treatment with SPP1/OPN knockdown inhibited the PEM-induced cell growth-inhibitory effect in PEM-sensitive cells. Expression of SPP1/OPN and AKT phosphorylation significantly decreased after PEM treatment of the PEM-sensitive cells. High immunohistochemical expression of SPP1/OPN was observed in two of three MPM patients who had a partial response to PEM-based chemotherapy. PEM has antitumor effects in MPM cells dependent on SPP1/OPN overexpression resulting in AKT activation. Our results suggest that SPP1 may be used as a single predictive biomarker of the effectiveness of PEM treatment in MPM.

Gene expression profiling reveals GC and CEACAM1 as new tools in the diagnosis of lung carcinoids

Background:

Classification of lung carcinoids into typical and atypical is a diagnostic challenge since no immunohistochemical tools are available to support pathologists in distinguishing between the two subtypes. A differential diagnosis is essential for clinicians to correctly discuss therapy, prognosis and follow-up with patients. Indeed, the distinction between the two typical and atypical subtypes on biopsies/cytological specimens is still unfeasible and sometimes limited also after radical surgeries. By comparing the gene expression profile of typical (TC) and atypical carcinoids (AC), we intended to find genes specifically expressed in one of the two subtypes that could be used as diagnostic markers.

Methods:

Expression profiling, with Affymetrix arrays, was performed on six typical and seven atypical samples. Data were validated on an independent cohort of 29 tumours, by means of quantitative PCR and immunohistochemistry (IHC).

Results:

High-throughput gene expression profiling was successfully used to identify a gene signature specific for atypical lung carcinoids. Among the 273 upregulated genes in the atypical vs typical subtype, GC (vitamin D-binding protein) and CEACAM1 (carcinoembryonic antigen family member) emerged as potent diagnostic markers. Quantitative PCR and IHC on a validation set of 17 ACs and 12 TCs confirmed their reproducibility and feasibility.

Conclusions:

GC and CEACAM1 can distinguish between TC and AC, defining an IHC assay potentially useful for routine cytological and histochemical diagnostic procedures. The high sensitivity and reproducibility of this new diagnostic algorithm strongly support a further validation on a wider sample size.

MiR-134/487b/655 cluster regulates TGF-β-induced epithelial-mesenchymal transition and drug resistance to gefitinib by targeting MAGI2 in lung adenocarcinoma cells

Epithelial-mesenchymal transition (EMT) has recently been recognized as a key element of cell invasion, migration, metastasis, and drug resistance in several types of cancer, including non-small cell lung cancer (NSCLC). Our aim was to clarify microRNA (miRNA)-related mechanisms underlying EMT followed by acquired resistance to epidermal growth factor receptor tyrosine kinase inhibitor (EGFR-TKI) in NSCLC. miRNA expression profiles were examined before and after transforming growth factor β1 (TGF-β1) exposure in four human adenocarcinoma cell lines with or without EMT. Correlation between expressions of EMT-related miRNAs and resistance to EGFR-TKI gefitinib was evaluated. miRNA array and real-time quantitative reverse transcription PCR (qRT-PCR) revealed that TGF-β1 significantly induced overexpression of miR-134, miR-487b, and miR-655, which belong to the same cluster located on chromosome 14q32, in lung adenocarcinoma cells with EMT. MAGI2 (membrane-associated guanylate kinase, WW, and PDZ domain-containing protein 2), a predicted target of these miRNAs and a scaffold protein required for PTEN, was diminished in A549 cells with EMT after the TGF-β1 stimulation. Overexpression of miR-134 and miR-487b promoted the EMT phenomenon and affected the drug resistance to gefitinib, whereas knockdown of these miRNAs inhibited the EMT process and reversed TGF-β1-induced resistance to gefitinib. Our study demonstrated that the miR-134/487b/655 cluster contributed to the TGF-β1-induced EMT phenomenon and affected the resistance to gefitinib by directly targeting MAGI2, in which suppression subsequently caused loss of PTEN stability in lung cancer cells. The miR-134/miR-487b/miR-655 cluster may be a new therapeutic target in patients with advanced lung adenocarcinoma, depending on the EMT phenomenon.

High cofilin-1 levels correlate with cisplatin resistance in lung adenocarcinomas

High cofilin-1 levels have been shown to be an accurate prognostic biomarker in non-small cell lung cancer (NSCLC) and a predictive factor in drug resistance. Herein we explore the role of cofilin-1 in cis-diamminedichloroplatinum(II) (cisplatin) resistance. We evaluated cofilin-1 levels in intrinsically cisplatin-resistant A549 (ICR-A549) cells and determined the cisplatin toxicity in A549 cells transiently transfected and overexpressing CFL1 plasmid. Moreover, expression levels (activity) of the CFL1 gene network were analyzed in a cisplatin-resistant human lung adenocarcinoma cell panel. ICR-A549 cells, selected by challenging parental cells with 10-fold drug GI50 value, presented a sixfold increase in cisplatin GI50 value and an increased cofilin-1 immunocontent (P < 0.01). In addition, cells transfected with cofilin-1 became more resistant to cisplatin (P < 0.01). High activity of the CFL1 gene network was found in a cisplatin-resistant adenocarcinoma cell panel (P < 0.01). In vitro evidences suggest that cofilin-1 is a biological predictor of cisplatin resistance, supporting new treatment initiatives based on cofilin-1 levels to guide chemotherapeutic interventions in NSCLC patients.

Genome-wide analysis of three-way interplay among gene expression, cancer cell invasion and anti-cancer compound sensitivity

BACKGROUND

Chemosensitivity and tumor metastasis are two primary issues in cancer management. Cancer cells often exhibit a wide range of sensitivity to anti-cancer compounds. To gain insight on the genetic mechanism of drug sensitivity, one powerful approach is to employ the panel of 60 human cancer cell lines developed by the National Cancer Institute (NCI). Cancer cells also show a broad range of invasion ability. However, a genome-wide portrait on the contributing molecular factors to invasion heterogeneity is lacking.

METHODS

Our lab performed an invasion assay on the NCI-60 panel. We identified invasion-associated (IA) genes by correlating our invasion profiling data with the Affymetrix gene expression data on NCI-60. We then employed the recently released chemosensitivity data of 99 anti-cancer drugs of known mechanism to investigate the gene-drug correlation, focusing on the IA genes. Afterwards, we collected data from four independent drug-testing experiments to validate our findings on compound response prediction. Finally, we obtained published clinical and molecular data from two recent adjuvant chemotherapy cohorts, one on lung cancer and one on breast cancer, to test the performance of our gene signature for patient outcome prediction.

RESULTS

First, we found 633 IA genes from the invasion-gene expression correlation study. Then, for each of the 99 drugs, we obtained a subset of IA genes whose expression levels correlated with drug-sensitivity profiles. We identified a set of eight genes (EGFR, ITGA3, MYLK, RAI14, AHNAK, GLS, IL32 and NNMT) showing significant gene-drug correlation with paclitaxel, docetaxel, erlotinib, everolimus and dasatinib. This eight-gene signature (derived from NCI-60) for chemosensitivity prediction was validated by a total of 107 independent drug tests on 78 tumor cell lines, most of which were outside of the NCI-60 panel. The eight-gene signature predicted relapse-free survival for the lung and breast cancer patients (log-rank P = 0.0263; 0.00021). Multivariate Cox regression yielded a hazard ratio of our signature of 5.33 (95% CI = 1.76 to 16.1) and 1.81 (95% CI = 1.19 to 2.76) respectively. The eight-gene signature features the cancer hallmark epidermal growth factor receptor (EGFR) and genes involved in cell adhesion, migration, invasion, tumor growth and progression.

CONCLUSIONS

Our study sheds light on the intricate three-way interplay among gene expression, invasion and compound-sensitivity. We report the finding of a unique signature that predicts chemotherapy survival for both lung and breast cancer. Augmenting the NCI-60 model with in vitro characterization of important phenotype-like invasion potential is a cost-effective approach to power the genomic chemosensitivity analysis.

Cancer association study of aminoacyl-tRNA synthetase signaling network in glioblastoma

Aminoacyl-tRNA synthetases (ARSs) and ARS-interacting multifunctional proteins (AIMPs) exhibit remarkable functional versatility beyond their catalytic activities in protein synthesis. Their non-canonical functions have been pathologically linked to cancers. Here we described our integrative genome-wide analysis of ARSs to show cancer-associated activities in glioblastoma multiforme (GBM), the most aggressive malignant primary brain tumor. We first selected 23 ARS/AIMPs (together referred to as ARSN), 124 cancer-associated druggable target genes (DTGs) and 404 protein-protein interactors (PPIs) of ARSs using NCI’s cancer gene index. 254 GBM affymetrix microarray data in The Cancer Genome Atlas (TCGA) were used to identify the probe sets whose expression were most strongly correlated with survival (Kaplan-Meier plots versus survival times, log-rank t-test <0.05). The analysis identified 122 probe sets as survival signatures, including 5 of ARSN (VARS, QARS, CARS, NARS, FARS), and 115 of DTGs and PPIs (PARD3, RXRB, ATP5C1, HSP90AA1, CD44, THRA, TRAF2, KRT10, MED12, etc). Of note, 61 survival-related probes were differentially expressed in three different prognosis subgroups in GBM patients and showed correlation with established prognosis markers such as age and phenotypic molecular signatures. CARS and FARS also showed significantly higher association with different molecular networks in GBM patients. Taken together, our findings demonstrate evidence for an ARSN biology-dominant contribution in the biology of GBM.

The p44/wdr77-dependent cellular proliferation process during lung development is reactivated in lung cancer

During lung development, cells proliferate for a defined length of time before they begin to differentiate. Factors that control this proliferative process and how this growth process is related to lung cancer are currently unknown. Here, we found that the WD40-containing protein (p44/wdr77) was expressed in growing epithelial cells at the early stages of lung development. In contrast, p44/wdr77 expression was diminished in fully differentiated epithelial cells in the adult lung. Loss of p44/wdr77 gene expression led to cell growth arrest and differentiation. Re-expression of p44/wdr77 caused terminally differentiated cells to re-enter the cell cycle. Our findings suggest that p44/wdr77 is essential and sufficient for proliferation of lung epithelial cells. P44/Wdr77 was re-expressed in lung cancer, and silencing p44/wdr77 expression strongly inhibited growth of lung adenocarcinoma cells in tissue culture and abolished growth of lung adenocarcinoma tumor xenografts in mice. The growth arrest induced by loss of p44/wdr77 expression was partially through the p21-Rb signaling. Our results suggest that p44/wdr77 controls cellular proliferation during lung development, and this growth process is reactivated during lung tumorigenesis.

Enzastaurin has anti-tumour effects in lung cancers with overexpressed JAK pathway molecules

BACKGROUND

Enzastaurin, an oral serine-threonine kinase inhibitor, was initially developed as an ATP-competitive selective inhibitor against protein kinase Cβ. However, the mechanism by which enzastaurin contributes to tumourigenesis remains unclear.

METHODS

We analysed the anti-tumour effects of enzastaurin in 22 lung cancer cell lines to ascertain the potential for enzastaurin-based treatment of lung cancer. To identify molecules or signalling pathways associated with this sensitivity, we conducted a gene, receptor tyrosine kinases phosphorylation and microRNA expression profiling study on the same set of cell lines.

RESULTS

We identified eight genes by pathway analysis of molecules having gene-drug sensitivity correlation, and used them to build a support vector machine algorithm model by which sensitive cell lines were distinguished from resistant cell lines. Pathway analysis revealed that the JAK/STAT signalling pathway was one of the main ones involved in sensitivity to enzastaurin. Overexpression of JAK1 was observed in the sensitive cells by western blotting. Simultaneous administration of enzastaurin and JAK inhibitor inhibited enzastaurin-induced cell growth-inhibitory effect. Furthermore, lentiviral-mediated JAK1-overexpressing cells were more sensitive to enzastaurin than control cells.

CONCLUSION

Our results suggested that the JAK1 pathway may be used as a single predictive biomarker for enzastaurin treatment. The anti-tumour effect of enzastaurin should be evaluated in lung cancer with overexpressed JAK pathway molecules.

Improved Feature Selection by Incorporating Gene Similarity into the LASSO

Personalized medicine is customizing treatments to a patient’s genetic profile and has the potential to revolutionize medical practice. An important process used in personalized medicine is gene expression profiling. Analyzing gene expression profiles is difficult, because there are usually few patients and thousands of genes, leading to the curse of dimensionality. To combat this problem, researchers suggest using prior knowledge to enhance feature selection for supervised learning algorithms. The authors propose an enhancement to the LASSO, a shrinkage and selection technique that induces parameter sparsity by penalizing a model’s objective function. Their enhancement gives preference to the selection of genes that are involved in similar biological processes. The authors’ modified LASSO selects similar genes by penalizing interaction terms between genes. They devise a coordinate descent algorithm to minimize the corresponding objective function. To evaluate their method, the authors created simulation data where they compared their model to the standard LASSO model and an interaction LASSO model. The authors’ model outperformed both the standard and interaction LASSO models in terms of detecting important genes and gene interactions for a reasonable number of training samples. They also demonstrated the performance of their method on a real gene expression data set from lung cancer cell lines.

Adaptive elastic-net sparse principal component analysis for pathway association testing

Pathway or gene set analysis has become an increasingly popular approach for analyzing high-throughput biological experiments such as microarray gene expression studies. The purpose of pathway analysis is to identify differentially expressed pathways associated with outcomes. Important challenges in pathway analysis are selecting a subset of genes contributing most to association with clinical phenotypes and conducting statistical tests of association for the pathways efficiently. We propose a two-stage analysis strategy: (1) extract latent variables representing activities within each pathway using a dimension reduction approach based on adaptive elastic-net sparse principal component analysis; (2) integrate the latent variables with the regression modeling framework to analyze studies with different types of outcomes such as binary, continuous or survival outcomes. Our proposed approach is computationally efficient. For each pathway, because the latent variables are estimated in an unsupervised fashion without using disease outcome information, in the sample label permutation testing procedure, the latent variables only need to be calculated once rather than for each permutation resample. Using both simulated and real datasets, we show our approach performed favorably when compared with five other currently available pathway testing methods.

Expression profiling in progressive stages of fumarate-hydratase deficiency: the contribution of metabolic changes to tumorigenesis

Hereditary leiomyomatosis and renal cell carcinoma (HLRCC) is caused by mutations in the Krebs cycle enzyme fumarate hydratase (FH). It has been proposed that "pseudohypoxic" stabilization of hypoxia-inducible factor-α (HIF-α) by fumarate accumulation contributes to tumorigenesis in HLRCC. We hypothesized that an additional direct consequence of FH deficiency is the establishment of a biosynthetic milieu. To investigate this hypothesis, we isolated primary mouse embryonic fibroblast (MEF) lines from Fh1-deficient mice. As predicted, these MEFs upregulated Hif-1α and HIF target genes directly as a result of FH deficiency. In addition, detailed metabolic assessment of these MEFs confirmed their dependence on glycolysis, and an elevated rate of lactate efflux, associated with the upregulation of glycolytic enzymes known to be associated with tumorigenesis. Correspondingly, Fh1-deficient benign murine renal cysts and an advanced human HLRCC-related renal cell carcinoma manifested a prominent and progressive increase in the expression of HIF-α target genes and in genes known to be relevant to tumorigenesis and metastasis. In accord with our hypothesis, in a variety of different FH-deficient tissues, including a novel murine model of Fh1-deficient smooth muscle, we show a striking and progressive upregulation of a tumorigenic metabolic profile, as manifested by increased PKM2 and LDHA protein. Based on the models assessed herein, we infer that that FH deficiency compels cells to adopt an early, reversible, and progressive protumorigenic metabolic milieu that is reminiscent of that driving the Warburg effect. Targets identified in these novel and diverse FH-deficient models represent excellent potential candidates for further mechanistic investigation and therapeutic metabolic manipulation in tumors.

HSP27 modulates epithelial to mesenchymal transition of lung cancer cells in a Smad-independent manner

Epithelial to mesenchymal transition (EMT) is induced by transforming growth factor-β1 (TGF-β1) and is a crucial event for cancer cells to acquire invasive and metastatic phenotypes. However, the signals that induce EMT in cancer cells have yet to be adequately defined. In this study, a proteomic investigation was performed to understand the signaling pathway of the EMT of lung cancer using two-dimensional difference gel electrophoresis (2D-DIGE) and mass spectrometry. The protein expression profiles of A549 were compared to those of A549 cells treated with TGF-β1. Of more than 2,000 protein spots shown by 2D-DIGE, 53 were found to be up- or down-regulated upon induction with TGF-β1. In the 53 protein spots, the protein level of heat shock protein (HSP) 27 was found to increase significantly. HSP27 protein was higher in two different lung cancer cell lines, demonstrating the EMT phenomenon with TGF-β1. Notably, the silencing of HSP27 enhanced spindle integration, resulting in an additive effect with TGF-β1-induced EMT. Furthermore, the TGF-β1-induced HSP27 increase was not affected by the suppression of Smad2 and Smad3 in A549 cells. These results suggest that HSP27 was involved in TGF-β1-induced EMT in a Smad-independent manner in lung cancer cells and may provide an effective clinical strategy in lung cancer patients whose tumors are dependent on TGF-β1-induced EMT.

Tumor and host factors that may limit efficacy of chemotherapy in non-small cell and small cell lung cancer

While chemotherapy provides useful palliation, advanced lung cancer remains incurable since those tumors that are initially sensitive to therapy rapidly develop acquired resistance. Resistance may arise from impaired drug delivery, extracellular factors, decreased drug uptake into tumor cells, increased drug efflux, drug inactivation by detoxifying factors, decreased drug activation or binding to target, altered target, increased damage repair, tolerance of damage, decreased proapoptotic factors, increased antiapoptotic factors, or altered cell cycling or transcription factors. Factors for which there is now substantial clinical evidence of a link to small cell lung cancer (SCLC) resistance to chemotherapy include MRP (for platinum-based combination chemotherapy) and MDR1/P-gp (for non-platinum agents). SPECT MIBI and Tc-TF scanning appears to predict chemotherapy benefit in SCLC. In non-small cell lung cancer (NSCLC), the strongest clinical evidence is for taxane resistance with elevated expression or mutation of class III beta-tubulin (and possibly alpha tubulin), platinum resistance and expression of ERCC1 or BCRP, gemcitabine resistance and RRM1 expression, and resistance to several agents and COX-2 expression (although COX-2 inhibitors have had minimal impact on drug efficacy clinically). Tumors expressing high BRCA1 may have increased resistance to platinums but increased sensitivity to taxanes. Limited early clinical data suggest that chemotherapy resistance in NSCLC may also be increased with decreased expression of cyclin B1 or of Eg5, or with increased expression of ICAM, matrilysin, osteopontin, DDH, survivin, PCDGF, caveolin-1, p21WAF1/CIP1, or 14-3-3sigma, and that IGF-1R inhibitors may increase efficacy of chemotherapy, particularly in squamous cell carcinomas. Equivocal data (with some positive studies but other negative studies) suggest that NSCLC tumors with some EGFR mutations may have increased sensitivity to chemotherapy, while K-ras mutations and expression of GST-pi, RB or p27kip1 may possibly confer resistance. While limited clinical data suggest that p53 mutations are associated with resistance to platinum-based therapies in NSCLC, data on p53 IHC positivity are equivocal. To date, resistance-modulating strategies have generally not proven clinically useful in lung cancer, although small randomized trials suggest a modest benefit of verapamil and related agents in NSCLC.

Development and validation of predictive indices for a continuous outcome using gene expression profiles

There have been relatively few publications using linear regression models to predict a continuous response based on microarray expression profiles. Standard linear regression methods are problematic when the number of predictor variables exceeds the number of cases. We have evaluated three linear regression algorithms that can be used for the prediction of a continuous response based on high dimensional gene expression data. The three algorithms are the least angle regression (LAR), the least absolute shrinkage and selection operator (LASSO), and the averaged linear regression method (ALM). All methods are tested using simulations based on a real gene expression dataset and analyses of two sets of real gene expression data and using an unbiased complete cross validation approach. Our results show that the LASSO algorithm often provides a model with somewhat lower prediction error than the LAR method, but both of them perform more efficiently than the ALM predictor. We have developed a plug-in for BRB-ArrayTools that implements the LAR and the LASSO algorithms with complete cross-validation.

New biomarkers for lung cancer

IMPORTANCE OF THE FIELD

Despite many efforts to improve early detection, lung cancer remains the leading cause of cancer deaths. Stage is the main determinant of prognosis and the basis for deciding treatment options. Screening tests for lung cancer have not been successful so far.

AREAS COVERED IN THE REVIEW

The article reviews the available literature related to biomarkers in use at present and those that could be used for early diagnosis, staging, prognosis, response to therapy and prediction of recurrence. The single biomarkers are analysed, divided according to the technological methods used and the locations of sampling.

WHAT THE READER WILL GAIN

The reader will gain knowledge on biomarkers in use and those now under study. The reader will also gain insights into the difficulties pertaining to the development of biomarkers, results reproducibility and clinical application.

TAKE HOME MESSAGE

Although some markers seem to be promising, at present there is no consensus on the proven value of their clinical use in lung cancer. The future lies probably in a panel of biomarkers instead of individual assays, or in predictive models derived from the integration of clinical variables and gene expression profiles.

Anticancer drug clustering based on proteomic profiles and a sensitivity database in a lung cancer cell line panel

Previously, we performed a molecular pharmacological study that applied a combination of DNA microarray-based gene expression profiling and drug sensitivity tests in vitro with a view to designing an improved chemotherapeutic strategy for advanced lung cancer. Utilizing recent key technological advances in proteomics, particularly antibody array-based methodologies, the current study aimed to examine the benefit of protein expression profiling in an analogous molecular pharmacological context. We performed protein expression analysis in a panel of lung cancer cell lines via an antibody array approach. Using a modified NCI program, we related cell line-specific proteomic profiles to the previously determined cytotoxic activity of a selection of commonly used anticancer agents, namely docetaxel, paclitaxel, gemcitabine, vinorelbine, 5-fluorouracil (5-FU), SN38, cisplatin (CDDP) and carboplatin (CBDCA). In addition, we compared these results with those obtained from our prior DNA microarray-based transcriptomic study. In our expression-drug correlation analysis using antibody array, gemcitabine consistently belonged to an isolated cluster. Docetaxel, paclitaxel, 5-FU, SN38, CBDCA and CDDP were gathered together into one large cluster. These results coincided with those generated by the prior transcriptomic study. Various genes were commonly listed that differentiated gemcitabine from the others. The identified factors associated with drug sensitivities were different between both analyses. Our proteomic profiling data provided confirmation of the previous transcript expression-drug sensitivity correlation analysis. These results suggest that chemotherapy regimens that include gemcitabine should be evaluated in second-line chemotherapy in cases where the first-line chemotherapy did not include this drug. Protein expression-drug sensitivity correlations in lung cancer cells in vitro may provide useful information in determining the most appropriate therapeutic options for lung cancer patients.

Clinical impact of high-throughput gene expression studies in lung cancer

Lung cancer is the leading cause of cancer death in the United States and the world. The high mortality rate results, in part, from the lack of effective tools for early detection and the inability to identify subsets of patients who would benefit from adjuvant chemotherapy or targeted therapies. The development of high-throughput genome-wide technologies for measuring gene expression, such as microarrays, have the potential to impact the mortality rate of lung cancer patients by improving diagnosis, prognosis, and treatment. This review will highlight recent studies using high-throughput gene expression technologies that have led to clinically relevant insights into lung cancer. The hope is that diagnostic and prognostic biomarkers that have been developed as part of this work will soon be ready for wide-spread clinical application and will have a dramatic impact on the evaluation of patients with suspect lung cancer, leading to effective personalized treatment regimens.

Sprouty 2 regulates DNA damage-induced apoptosis in Ras-transformed human fibroblasts

We have reported that expression of Sprouty 2 (Spry2) is necessary for tumor formation by HRas(V12)-transformed fibroblasts. We now report on the role of Spry2 in the inhibition of UV(254 nm) radiation-induced apoptosis in HRas(V12)-transformed human fibroblasts. Silencing Spry2 in this context resulted in increased apoptosis, associated with decreased Akt activation and decreased phosphorylation of HDM2 at Ser-166, which has been shown to stabilize HDM2. As a consequence, when cells with silenced Spry2 were UV-irradiated, they exhibited diminished levels of HDM2 and elevated levels of p53. In agreement with these findings, overexpression of Spry2 in the parental non-transformed fibroblasts led to increased Akt activation and to the stabilization of HDM2. It also led to diminished expression of p53 and decreased apoptosis following UV irradiation. Silencing Spry2 in HRas-transformed cells decreased Rac1 activation, but independent expression of Spry2 in the non-transformed parental cells had no effect on Rac1, suggesting a specific involvement in the activation of Rac1 by Ras. Silencing Spry2 in HRas(V12)-transformed cells resulted in diminished interaction between HRas and Tiam1, a Rac1-specific nucleotide exchange factor. Expression of constitutively active Rac1 in cells with silenced Spry2 partly reversed the effect of Spry2 down-regulation. Furthermore, loss of Spry2 expression in HRas(V12)-transformed cells augmented the cytotoxicity of the DNA-damaging, chemotherapeutic agent cisplatin, a process that was also reversed by active Rac1. Together, these data show that Spry2 inhibits apoptosis in response to DNA damage by regulating Akt, HDM2, and p53, by a process mediated partly by Rac1.

Gene expression patterns within cell lines are predictive of chemosensitivity

Background

The NCI has undertaken a twenty-year project to characterize compound sensitivity patterns in a selected set of sixty tumor derived cell lines. Previous studies have explored the relationship between compound sensitivity patterns to gene expression, protein expression, and DNA copy number for these same cell lines. A strong correlation between the pattern of expression of a biomarker and sensitivity to a compound could suggest a clinically interesting biological relationship between the two.

Results

We isolated RNA's and measured expression of 40000 genes using cDNA microarrays from the fifty-nine publicly available cell lines. Analysis of this data set in comparison with published gene expression data sets demonstrates a high degree of reproducibility in expression level measurements even using completely independent RNA preparations and array technologies. Using the fifty-nine cell lines for discovery and an additional seven cell lines for which extensive compound sensitivity data were available as a test set, we determined that gene-compound pairs with a correlation coefficient above 0.6 had a false discovery rate of approximately 5%. Large scale features of the gene expression and chemosensitivity data, such as tissue of origin and other physiological factors, did not seem to explain the majority of correlations between gene and compound patterns.

Conclusion

A comparison of gene expression and compound sensitivity in panels of cell lines was demonstrated to have a relatively high validation and low false discovery rate supporting the use of this approach and datasets for identifying candidate biomarkers and targeted biologically active compounds.

Frequent overexpression of the genes FXR1, CLAPM1 and EIF4G located on amplicon 3q26-27 in squamous cell carcinoma of the lung

Previously, we reported gene amplification at chromosome 3q26-27 in more than one third of squamous cell carcinomas of the lung. Frequent amplification of eukaryotic translation initiation factor 4G on 3q27.1 indicated a possible role of this amplification in translation initiation. The analysis of 61 squamous cell lung carcinomas shows that the percentage of carcinomas with a 3q27.1 amplification increases in higher malignant tumors. Non-invasive (T1) and minimal-invasive (T2) tumor stages showed similar percentages of amplified and non-amplified tumors, whereas locally-invasive (T3) tumors revealed a statistically significant (p < 0.05) increased percentage of amplified tumors. Microarrays were used to analyze the expression pattern of genes mapping in the amplified domain and its flanking regions (3q25-28) as well as the expression of genes directly or indirectly associated with translation initiation in squamous cell carcinoma, large cell carcinoma, adenocarcinoma and small cell carcinoma. Three genes, namely FXR1, CLAPM1 and EIF4G, are most frequently overexpressed in the center of the amplified domain in squamous cell carcinomas. The eukaryotic translation initiation factors 4A1, 2B and 4B as well as the poly(A)-binding protein PABPC1 where found to be overexpressed in all lung cancer entities. We found, however, no overexpression of eIF4E. Our results contribute to the understanding of the frequent amplification processes in squamous cell carcinomas of the lung and to the understanding of the translation initiation that appears not to require eIF4E in lung carcinogenesis.

Metagene projection for cross-platform, cross-species characterization of global transcriptional states

The high dimensionality of global transcription profiles, the expression level of 20,000 genes in a much small number of samples, presents challenges that affect the sensitivity and general applicability of analysis results. In principle, it would be better to describe the data in terms of a small number of metagenes, positive linear combinations of genes, which could reduce noise while still capturing the invariant biological features of the data. Here, we describe how to accomplish such a reduction in dimension by a metagene projection methodology, which can greatly reduce the number of features used to characterize microarray data. We show, in applications to the analysis of leukemia and lung cancer data sets, how this approach can help assess and interpret similarities and differences between independent data sets, enable cross-platform and cross-species analysis, improve clustering and class prediction, and provide a computational means to detect and remove sample contamination.

In vitro Simulation Study of Individualized Chemotherapy in Lung Cancer

The primary aim of this in vitro simulation study was to evaluate the utility of gene expression profile analysis in predicting the effect of varying drug combinations for the treatment of lung cancer. Using 10 human cancer cell lines, we focused our gene expression analysis on a cohort of candidate sensitivity-prediction factors, previously reported using cDNA filter arrays, with a view to predicting the ability of a set of anti-cancer drugs commonly used to treat lung cancer, namely cisplatin, 5-fluorouracil (5FU), SN38, docetaxel, gemcitabine, and vinorelbine. Altered expression of genes for glutathione-S-transferase-pi, uridine phosphorylase, O-6-methylguanine-DNA methyltransferase, and multidrug resistance 1 was identified in lung cancer cell lines. Drug sensitivity testing, in the form of methylthiotetrazol analysis, was performed using these six anti-cancer drugs against the panel of 10 lung cancer cell lines. We compared the predicted chemosensitivity based on the gene expression pattern of 19 well-known sensitivity-related genes with the cytotoxic activity of each of these anti-cancer drugs. Molecular profiling data predicted resistance to CDDP in LK-2 cells, 5FU in LK-2, PC7, A549, NCI-N231, Lu135 cells, irinitecan in PC9 cells, and VNR in PC7 cells. However, the prediction efficacy (number of predicted inactive drugs by gene expression analysis/number of inactive drugs by methylthiotetrazol assay) was 21.6% (8 of 37). No false-positive findings in relation to sensitivity-related genes were obtained on the basis of this molecular analysis. Thus, prediction of sensitivity to lung cancer by molecular analysis appears possible. With elucidation of additional drug sensitivity factors, selection of appropriate anticancer drugs by gene expression profiling may make it possible to increase the response rate in lung cancer chemotherapy.

Single-agent gemcitabine for biliary tract cancers. Study outcomes and systematic review of the literature

OBJECTIVE

The aim of this study was to investigate the outcomes of gemcitabine-treated patients with inoperable biliary tract cancers.

METHODS

We conducted a retrospective study of consecutively treated 22 inoperable biliary tract cancer patients with gemcitabine (500-1,000 mg/m(2) on days 1, 8, 15 every 4 weeks) as first-line, and 17 patients as second- or third-line treatment.

RESULTS

The response rate of patients treated with gemcitabine as first-line and second- or third-line treatment was 5.3 and 28.5%, respectively. The median overall survival time in the first-, and second- or third-line treatment groups was 8.3 and 17.0 months, and the 1-year survival rate was 44.0 and 50.9%, respectively. The present study also suggests the possibility that the prognosis of patients with high levels of C-reactive protein and total bilirubin, or a low level of albumin might be worse.

CONCLUSIONS

Our results indicate that the treatment of inoperable biliary tract cancers with gemcitabine is feasible. There was no difference in the response rate and overall survival between biliary tract cancer patients in the first- and second- or third-line treatment groups. We also present the systematic review of literature of the recent treatment results of biliary tract cancers treated with gemcitabine.