Robust gene expression signature from formalin-fixed paraffin-embedded samples predicts prognosis of non-small-cell lung cancer patients - PubMed (original) (raw)

. 2011 Sep 1;17(17):5705-14.

doi: 10.1158/1078-0432.CCR-11-0196. Epub 2011 Jul 8.

Guanghua Xiao, Kevin R Coombes, Carmen Behrens, Luisa M Solis, Gabriela Raso, Luc Girard, Heidi S Erickson, Jack Roth, John V Heymach, Cesar Moran, Kathy Danenberg, John D Minna, Ignacio I Wistuba

Affiliations

Robust gene expression signature from formalin-fixed paraffin-embedded samples predicts prognosis of non-small-cell lung cancer patients

Yang Xie et al. Clin Cancer Res. 2011.

Abstract

Purpose: The requirement of frozen tissues for microarray experiments limits the clinical usage of genome-wide expression profiling by using microarray technology. The goal of this study is to test the feasibility of developing lung cancer prognosis gene signatures by using genome-wide expression profiling of formalin-fixed paraffin-embedded (FFPE) samples, which are widely available and provide a valuable rich source for studying the association of molecular changes in cancer and associated clinical outcomes.

Experimental design: We randomly selected 100 Non-Small-Cell lung cancer (NSCLC) FFPE samples with annotated clinical information from the UT-Lung SPORE Tissue Bank. We microdissected tumor area from FFPE specimens and used Affymetrix U133 plus 2.0 arrays to attain gene expression data. After strict quality control and analysis procedures, a supervised principal component analysis was used to develop a robust prognosis signature for NSCLC. Three independent published microarray datasets were used to validate the prognosis model.

Results: This study showed that the robust gene signature derived from genome-wide expression profiling of FFPE samples is strongly associated with lung cancer clinical outcomes and can be used to refine the prognosis for stage I lung cancer patients, and the prognostic signature is independent of clinical variables. This signature was validated in several independent studies and was refined to a 59-gene lung cancer prognosis signature.

Conclusions: We conclude that genome-wide profiling of FFPE lung cancer samples can identify a set of genes whose expression level provides prognostic information across different platforms and studies, which will allow its application in clinical settings.

©2011 AACR.

PubMed Disclaimer

Figures

Figure 1

Figure 1

(a) Flow chart of the derivation and validation of the robust gene signature from formalin-fixed and paraffin-embedded samples collected from M.D. Anderson UT-Lung Cancer SPORE tissue bank. (b) Flow chart of the derivation and validation of 59-gene prognosis signature.

Figure 2

Figure 2

Microarray analysis of the gene-expression profiles from formalin-fixed and paraffin-embedded (FFPE) lung tumor samples. (a) Unsupervised cluster analysis of the 55 FFPE lung cancer patient cohort using the expression profile of 1400 robust genes that pass the microarray quality control criterion. Vertical and horizontal axes represent robust genes and lung cancer patient clusters, respectively. (b) Kaplan-Meier plot showing the association of the expression of robust genes with patient survival _P_-values were obtained using the log-rank test. Red color represents sample Cluster I and black color represents sample Cluster II defined by unsupervised clustering algorithm using robust gene profiling data. indicates censored samples. Gene set enrichment analysis found that the ER negative signature derived from breast cancer patients is enriched in group 1 defined by RGS expression (c), and the ER positive signature derived from breast cancer patients is enriched in group 2 defined by RGS expression (d). The y axis shows running enrichment scores for the specific gene set on the 1400 pre-ranked genes. The x axis shows the rank in the ordered dataset. The vertical lines represent the locations of the genes that are in the specific gene set.

Figure 3

Figure 3

Kaplan-Meier plots showing the predictive power of the robust gene signatures. 55 FFPE tumor samples from M. D. Anderson Cancer Center were randomly divided into training (25 samples) and testing (30 samples) sets (a). Independent validation of the robust gene signature in the 442-frozen-sample cohort from multi-institute consortium. The microarray data sets were divided into two groups, one for the training and the other for the testing cohort according to the original paper (b). The training data is 55 FFPE tumor samples and the testing data set is 442-frozen-sample cohort from multi-institute consortium. The testing was done for all patients (c), stage I patients (e), stage II patients (f) and stage III patients (g) separately. The training data is the consortium dataset with 442 forzen samples and the testing data is 55 FFPE samples from M.D. Anderson Cancer Center (d). P values were obtained by the log-rank test. Red and black lines represent predicted high- and low-risk groups, respectively. · indicates censored samples.

Figure 4

Figure 4

Comparison of individual gene effect across FFPE samples from M, D. Anderson Cancer Center and 442 frozen samples from consortium. (a) Venn-diagram of genes associated with overall survival (P<0.05 in univariate Cox regression models). It shows 59 genes are significantly associated with survival in both FFPE data and consortium data. (b) The hazard ratios from univariate Cox regression models for the 59 genes common in both sets are consistent between FFPE set and consortium set. (c) Regulatory gene and protein interaction networks defined by the 59 predictors. Computational molecular interaction network prediction based on genes and proteins associated with the significant pathways in the Ingenuity Pathways Knowledge Base (IPKB) by Ingenuity Pathways Analysis (IPA). Interactions between the different nodes are given as solid (direct interaction) and dashed (indirect interaction) lines (edges) with various colors for the different interaction types. This network received the highest score by IPA and is mostly centered on the transcription factors HNF4A and HNF1A, and ONECUT1. The shaded genes are the genes belonging to 59-gene signature.

Figure 5

Figure 5

Kaplan-Meier plots showing the predictive power of the 59-gene signature for two independent validation sets. The training data is 55 FFPE tumor samples from M.D. Anderson Cancer Center and the testing data set is frozen samples from lung cancer patients from Bhattacharjee et al dataset (a), the stage I patients from Bhattacharjee et al dataset (b), frozen samples from lung cancer patients from Bild et al dataset (c), and the stage I patients from Bild et al dataset (d). P values were obtained by the log-rank test. Red and black lines represent predicted high- and low-risk groups, respectively. indicates censored samples.

References

    1. Jemal A, Siegel R, Ward E, Hao Y, Xu J, Thun MJ. Cancer statistics, 2009. CA Cancer J Clin. 2009 Jul-Aug;59(4):225–49. - PubMed
    1. Tsuboi M, Ohira T, Saji H, Miyajima K, Kajiwara N, Uchida O, et al. The present status of postoperative adjuvant chemotherapy for completely resected non-small cell lung cancer. Ann Thorac Cardiovasc Surg. 2007 Apr;13(2):73–7. - PubMed
    1. Douillard JY, Rosell R, De Lena M, Carpagnano F, Ramlau R, Gonzales-Larriba JL, et al. Adjuvant vinorelbine plus cisplatin versus observation in patients with completely resected stage IB-IIIA non-small-cell lung cancer (Adjuvant Navelbine International Trialist Association [ANITA]): a randomised controlled trial. Lancet Oncol. 2006 Sep;7(9):719–27. - PubMed
    1. Kato H, Ichinose Y, Ohta M, Hata E, Tsubota N, Tada H, et al. A randomized trial of adjuvant chemotherapy with uracil-tegafur for adenocarcinoma of the lung. N Engl J Med. 2004 Apr 22;350(17):1713–21. - PubMed
    1. Arriagada R, Bergman B, Dunant A, Le Chevalier T, Pignon JP, Vansteenkiste J. Cisplatin-based adjuvant chemotherapy in patients with completely resected non-small-cell lung cancer. N Engl J Med. 2004 Jan 22;350(4):351–60. - PubMed

Publication types

MeSH terms

Substances

Grants and funding

LinkOut - more resources