Pei-Chun Chen

Identification of a Novel Biomarker, SEMA5A, for Non–Small Cell Lung Carcinoma in Nonsmoking Women

Background: Although cigarette smoking is the major risk factor for lung cancer, only 7% of female lung cancer patients in Taiwan have a history of smoking. The genetic mechanisms of carcinogenesis in nonsmokers are unclear, but semaphorins have been suggested to play a role as lung tumor suppressors. This report is a comprehensive analysis of the molecular signature of nonsmoking female lung cancer patients in Taiwan, with a particular focus on the semaphorin gene family. Methods: Sixty pairs of tumor and adjacent normal lung tissue specimens were analyzed by using Affymetrix U133plus2.0 expression arrays. Differentially expressed genes in tumor tissues were identified by a paired t test and validated by reverse transcriptase-PCR and immunohistochemistry. Functional analysis was conducted by using Ingenuity Pathway Analysis as well as gene set enrichment analysis and sigPathway algorithms. Kaplan-Meier survival analyses were used to evaluate the association of SEMA5A expression and clinical outcome. Results: We identified 687 differentially expressed genes in non–small cell lung carcinoma (NSCLC). Many of these genes, most notably the semaphorin family, were participants in the axon guidance signaling pathway. The downregulation of SEMA5A in tumor tissue, both at the transcriptional and translational levels, was associated with poor survival among nonsmoking women with NSCLC. Conclusions: In summary, several semaphorin gene family members were identified as potential therapeutic targets, and SEMA5A may be useful as a prognostic biomarker for NSCLC, which may also be gender specific in Taiwanese patients. Impact: A novel biomarker for NSCLC is identified. Cancer Epidemiol Biomarkers Prev; 19(10); 2590–7. ©2010 AACR.

Integrated Analyses of Copy Number Variations and Gene Expression in Lung Adenocarcinoma

Numerous efforts have been made to elucidate the etiology and improve the treatment of lung cancer, but the overall five-year survival rate is still only 15%. Identification of prognostic biomarkers for lung cancer using gene expression microarrays poses a major challenge in that very few overlapping genes have been reported among different studies. To address this issue, we have performed concurrent genome-wide analyses of copy number variation and gene expression to identify genes reproducibly associated with tumorigenesis and survival in non-smoking female lung adenocarcinoma. The genomic landscape of frequent copy number variable regions (CNVRs) in at least 30% of samples was revealed, and their aberration patterns were highly similar to several studies reported previously. Further statistical analysis for genes located in the CNVRs identified 475 genes differentially expressed between tumor and normal tissues (p<10−5). We demonstrated the reproducibility of these genes in another lung cancer study (pu200a=u200a0.0034, Fishers exact test), and showed the concordance between copy number variations and gene expression changes by elevated Pearson correlation coefficients. Pathway analysis revealed two major dysregulated functions in lung tumorigenesis: survival regulation via AKT signaling and cytoskeleton reorganization. Further validation of these enriched pathways using three independent cohorts demonstrated effective prediction of survival. In conclusion, by integrating gene expression profiles and copy number variations, we identified genes/pathways that may serve as prognostic biomarkers for lung tumorigenesis.

A new regularized least squares support vector regression for gene selection.

BackgroundSelection of influential genes with microarray data often faces the difficulties of a large number of genes and a relatively small group of subjects. In addition to the curse of dimensionality, many gene selection methods weight the contribution from each individual subject equally. This equal-contribution assumption cannot account for the possible dependence among subjects who associate similarly to the disease, and may restrict the selection of influential genes.ResultsA novel approach to gene selection is proposed based on kernel similarities and kernel weights. We do not assume uniformity for subject contribution. Weights are calculated via regularized least squares support vector regression (RLS-SVR) of class levels on kernel similarities and are used to weight subject contribution. The cumulative sum of weighted expression levels are next ranked to select responsible genes. These procedures also work for multiclass classification. We demonstrate this algorithm on acute leukemia, colon cancer, small, round blue cell tumors of childhood, breast cancer, and lung cancer studies, using kernel Fisher discriminant analysis and support vector machines as classifiers. Other procedures are compared as well.ConclusionThis approach is easy to implement and fast in computation for both binary and multiclass problems. The gene set provided by the RLS-SVR weight-based approach contains a less number of genes, and achieves a higher accuracy than other procedures.

Identification of prognostic genes for recurrent risk prediction in triple negative breast cancer patients in Taiwan.

Discrepancies in the prognosis of triple negative breast cancer exist between Caucasian and Asian populations. Yet, the gene signature of triple negative breast cancer specifically for Asians has not become available. Therefore, the purpose of this study is to construct a prediction model for recurrence of triple negative breast cancer in Taiwanese patients. Whole genome expression profiling of breast cancers from 185 patients in Taiwan from 1995 to 2008 was performed, and the results were compared to the previously published literature to detect differences between Asian and Western patients. Pathway analysis and Cox proportional hazard models were applied to construct a prediction model for the recurrence of triple negative breast cancer. Hierarchical cluster analysis showed that triple negative breast cancers from different races were in separate sub-clusters but grouped in a bigger cluster. Two pathways, cAMP-mediated signaling and ephrin receptor signaling, were significantly associated with the recurrence of triple negative breast cancer. After using stepwise model selection from the combination of the initial filtered genes, we developed a prediction model based on the genes SLC22A23, PRKAG3, DPEP3, MORC2, GRB7, and FAM43A. The model had 91.7% accuracy, 81.8% sensitivity, and 94.6% specificity under leave-one-out support vector regression. In this study, we identified pathways related to triple negative breast cancer and developed a model to predict its recurrence. These results could be used for assisting with clinical prognosis and warrant further investigation into the possibility of targeted therapy of triple negative breast cancer in Taiwanese patients.

Bayesian random effects for interrater and test–retest reliability with nested clinical observations

OBJECTIVEnThe assessment of inter- and intrarater reliability usually involves more than one level of nesting structures in the collected data, where repeated observations are made by multiple raters. Most approaches, however, are not designed to accommodate both inter- and intrarater reliability jointly, not to mention further difficulties arising when modeling with dichotomous responses. The multiple sources of dependence because of nesting structures and the existence of covariates can result in complexity in inference.nnnSTUDY DESIGN AND SETTINGnWe first establish the equivalence between correlation and kappa under common positive correlation models for multiple raters and then apply a Bayesian generalized linear mixed-effects model to interpret simultaneously both types of reproducibility through different annotations of similarity. In addition to marginal correlations, the correlated random effects among raters are adopted to infer similarity between raters, whereas the correlation for random time effects may contribute to test-retest reliability.nnnRESULTSnThis model accounts for individual covariates and random effects because of subjects, raters, and time, and it covers a wide variety of data structures and types. An application of endodontic radiographic examinations is illustrated.nnnCONCLUSIONnThis Bayesian hierarchical correlation model offers a wide applicability, flexibility, and feasibility in modeling inter- and intrarater reliability together.

A simple Bayesian mixture model with a hybrid procedure for genome-wide association studies

Genome-wide association studies often face the undesirable result of either failing to detect any influential markers at all because of a stringent level for testing error corrections or encountering difficulty in quantifying the importance of markers by their P-values. Advocates of estimation procedures prefer to estimate the proportion of association rather than test significance to avoid overinterpretation. Here, we adopt a Bayesian hierarchical mixture model to estimate directly the proportion of influential markers, and then proceed to a selection procedure based on the Bayes factor (BF). This mixture model is able to accommodate different sources of dependence in the data through only a few parameters. Specifically, we focus on a standardized risk measure of unit variance so that fewer parameters are involved in inference. The expected value of this measure follows a mixture distribution with a mixing probability of association, and it is robust to minor allele frequencies. Furthermore, to select promising markers, we use the magnitude of the BF to represent the strength of evidence in support of the association between markers and disease. We demonstrate this procedure both with simulations and with SNP data from studies on rheumatoid arthritis, coronary artery disease, and Crohns disease obtained from the Wellcome Trust Case–Control Consortium. This Bayesian procedure outperforms other existing methods in terms of accuracy, power, and computational efficiency. The R code that implements this method is available at http://homepage.ntu.edu.tw/~ckhsiao/Bmix/Bmix.htm.

Use of germline polymorphisms in predicting concurrent chemoradiotherapy response in esophageal cancer.

PURPOSEnTo identify germline polymorphisms to predict concurrent chemoradiation therapy (CCRT) response in esophageal cancer patients.nnnMATERIALS AND METHODSnA total of 139 esophageal cancer patients treated with CCRT (cisplatin-based chemotherapy combined with 40 Gy of irradiation) and subsequent esophagectomy were recruited at the National Taiwan University Hospital between 1997 and 2008. After excluding confounding factors (i.e., females and patients aged ≥70 years), 116 patients were enrolled to identify single nucleotide polymorphisms (SNPs) associated with specific CCRT responses. Genotyping arrays and mass spectrometry were used sequentially to determine germline polymorphisms from blood samples. These polymorphisms remain stable throughout disease progression, unlike somatic mutations from tumor tissues. Two-stage design and additive genetic models were adopted in this study.nnnRESULTSnFrom the 26 SNPs identified in the first stage, 2 SNPs were found to be significantly associated with CCRT response in the second stage. Single nucleotide polymorphism rs16863886, located between SGPP2 and FARSB on chromosome 2q36.1, was significantly associated with a 3.93-fold increase in pathologic complete response to CCRT (95% confidence interval 1.62-10.30) under additive models. Single nucleotide polymorphism rs4954256, located in ZRANB3 on chromosome 2q21.3, was associated with a 3.93-fold increase in pathologic complete response to CCRT (95% confidence interval 1.57-10.87). The predictive accuracy for CCRT response was 71.59% with these two SNPs combined.nnnCONCLUSIONSnThis is the first study to identify germline polymorphisms with a high accuracy for predicting CCRT response in the treatment of esophageal cancer.

Impact of universal health coverage on urban-rural inequity in psychiatric service utilisation for patients with first admission for psychosis: a 10-year nationwide population-based study in Taiwan.

Objective To examine the disparities in psychiatric service utilisation over a 10-year period for patients with first admission for psychosis in relation to urban–rural residence following the implementation of universal health coverage in Taiwan. Design Population-based retrospective cohort study. Setting Taiwans National Health Insurance Research Database, which has a population coverage rate of over 99% and contains all medical claim records of a nationwide cohort of patients with at least one psychiatric admission between 1996 and 2007. Participants 69u2005690 patients aged 15–59u2005years with first admission between 1998 and 2007 for any psychotic disorder. Main exposure measure Patients’ urban–rural residence at first admissions. Main outcome measures Absolute and relative inequality indexes of the following quality indicators after discharge from the first admission: all-cause psychiatric readmission at 2 and 4u2005years, dropout of psychiatric outpatient service at 30u2005days, and emergency department (ED) treat-and-release encounter at 30u2005days. Results Between 1998 and 2007, the 4-year readmission rate decreased from 65% to 58%, the 30-day dropout rate decreased from 18% to 15%, and the 30-day ED encounter rate increased from 8% to 10%. Risk of readmission has significantly decreased in rural and urban patients, but at a slower speed for the rural patients (p=0.026). The adjusted HR of readmission in rural versus urban patients has increased from 1.00 (95% CI 0.96 to 1.04) in 1998–2000 to 1.08 (95% CI 1.03 to 1.12) in 2005–2007, indicating a mild widening of the urban–rural gap. Urban–rural differences in 30-day dropout and ED encounter rates have been stationary over time. Conclusions The universal health coverage in Taiwan did not narrow urban–rural inequity of psychiatric service utilisation in patients with psychosis. Therefore, other policy interventions on resource allocation, service delivery and quality of care are needed to improve the outcome of rural-dwelling patients with psychosis.

SNP rs10248565 in HDAC9 as a novel genomic aberration biomarker of lung adenocarcinoma in non-smoking women

BackgroundNumerous efforts have been made to elucidate the etiology and improve the treatment of lung cancer, but the overall five-year survival rate is still only 15%. Although cigarette smoking is the primary risk factor for lung cancer, only 7% of female lung cancer patients in Taiwan have a history of smoking. Since cancer results from progressive accumulation of genetic aberrations, genomic rearrangements may be early events in carcinogenesis.ResultsIn order to identify biomarkers of early-stage adenocarcinoma, the genome-wide DNA aberrations of 60 pairs of lung adenocarcinoma and adjacent normal lung tissue in non-smoking women were examined using Affymetrix Genome-Wide Human SNP 6.0 arrays. Common copy number variation (CNV) regions were identified by ≥30% of patients with copy number beyond 2u2009±u20090.5 of copy numbers for each single nucleotide polymorphism (SNP) and at least 100 continuous SNP variant loci. SNPs associated with lung adenocarcinoma were identified by McNemar’s test. Loss of heterozygosity (LOH) SNPs were identified in ≥18% of patients with LOH in the locus. Aberration of SNP rs10248565 at HDAC9 in chromosome 7p21.1 was identified from concurrent analyses of CNVs, SNPs, and LOH.ConclusionThe results elucidate the genetic etiology of lung adenocarcinoma by demonstrating that SNP rs10248565 may be a potential biomarker of cancer susceptibility.

Time trends in first admission rates for schizophrenia and other psychotic disorders in Taiwan, 1998–2007: a 10-year population-based cohort study

PurposeTo examine the trend in annual first admission rates for psychotic disorders as a whole as well as individual psychotic disorders in Taiwan from 1998 to 2007, and influences of age, sex, and geographic region on the trend.MethodUsing the inpatient claims records in the National Health Insurance Research Database, we estimated the yearly first admission rates for schizophrenia and other psychotic disorders, including voluntary (1998–2007) and involuntary (2004–2007) admissions. Both narrow and broad definitions of psychotic disorders were examined.ResultsWhile involuntary first admission rates were stable, a crescendo–decrescendo change in voluntary first admission rates for psychotic disorders was observed, peaking in 2001. The increase from 1998 to 2001 was closely associated with health insurance expansion. Before 2001, the voluntary first admission rates in males aged 15–24 were underestimated as military personnel records were not included in the database. From 2001 to 2007, voluntary first admissions for psychotic disorders decreased 38%; the decrease could not be accounted for by the mild diagnostic shifts away from schizophrenia to affective psychosis or substance-induced psychosis. During the entire observation period, first admission rates for schizophrenia decreased 48%, while affective psychosis increased 84%. Gender disparities in the first admission rates gradually diminished, but geographic disparities persisted.ConclusionsFirst admission rates for psychosis significantly reduced in Taiwan between 1998 and 2007, mainly driven by the reduced hospitalization risk for schizophrenia. Special attention should be paid to the increased hospitalization for other types of psychotic disorders (especially affective psychosis) and the unresolved geographic disparities.