OL Oncology Letters 1792-1074 1792-1082 D.A. Spandidos 10.3892/ol.2017.5691 OL-0-0-5691 Articles Gene signatures associated with drug resistance to irinotecan and oxaliplatin predict a poor prognosis in patients with colorectal cancer SunXinrong 1 * WangXiang 1 * FengWenming 1 GuoHuihui 1 TangChengwu 1 LuYongliang 2 XiangXiaobin 3 BaoYing 1 Department of Gastrointestinal Surgery, The First Affiliated Hospital, Huzhou Teachers College, The First People's Hospital of Huzhou, Huzhou, Zhejiang 313000, P.R. China Department of Medicine, Huzhou Teachers College, Huzhou, Zhejiang 313000, P.R. China Department of Sports Medicine, Guangdong Orthopedic Hospital, Guangzhou, Guangdong 510045, P.R. China Correspondence to: Dr Ying Bao, Department of Gastrointestinal Surgery, The First Affiliated Hospital, Huzhou Teachers College, The First People's Hospital of Huzhou, 158 Guangchanghou Road, Huzhou, Zhejiang 313000, P.R. China, E-mail: baoying2222@126.com

Contributed equally

 04 2017 07 02 2017 13 4 2089 2096 30082015 30112016 Copyright: © Sun et al. 2017 This is an open access article distributed under the terms of the Creative Commons Attribution-NonCommercial-NoDerivs License, which permits use and distribution in any medium, provided the original work is properly cited, the use is non-commercial and no modifications or adaptations are made.

The identification of novel survival predictors may help to improve the appropriate management of colorectal cancer (CRC). In the present study, two gene sets associated with irinotecan or oxaliplatin resistance in CRC cell lines were first identified and subsequently applied to the clinical CRC microarray dataset GSE14333. Subsequently, a 60-gene irinotecan resistance-associated signature and a 13-gene oxaliplatin resistance-associated signature were established, which were able to classify CRC patients into high- and low-risk subgroups with varied clinical outcomes [irinotecan-resistance gene signature: hazard ratio (HR)=0.4607, 95% confidence interval (CI)=0.3369–0.6300, P<0.0001; oxaliplatin-resistance gene signature: HR=0.6119, 95% CI=0.4547–0.8233, P=0.0008]. The performance of these two gene expression signatures in predicting outcome risk were also validated in two other independent CRC gene expression microarray datasets, GSE17536 (irinotecan-resistance gene signature: HR=0.5318, 95% CI=0.3359–0.8419, P=0.0079; oxaliplatin-resistance gene signature: HR=0.5383, 95% CI=0.3400–0.8521, P=0.0114) and GSE17537 (irinotecan-resistance gene signature: HR=0.2827, 95% CI=0.1173–0.6813, P=0.0088; oxaliplatin-resistance gene signature: HR=0.2378, 95% CI=0.09773–0.5784, P=0.0023). Furthermore, the combination of these two gene classifiers demonstrated a superior performance in CRC prognosis prediction than either used individually. Therefore, this study proposed novel gene classifier models for CRC prognosis prediction, which may be potentially useful to inform treatment decisions for patients with CRC in clinical settings.

 colorectal cancer prognosis gene signature microarray chemotherapy resistance Introduction

Regimens based on the anti-metabolite drug 5-fluorouracil (5-FU) combined with the topoisomerase I inhibitor irinotecan or the DNA-binding agent oxaliplatin, which is also known as oxaliplatin with 5-FU and folinic acid chemotherapy (FOLFOX), are typically used as the initial chemotherapeutic treatment for colorectal cancer (CRC) (1). Although an objective response to chemotherapeutic regimens significantly increases the survival of patients with CRC, a relatively high proportion (50–70%) of innate and acquired resistance is still a major clinical problem faced by these individuals (2). Therefore, predicting the risk of chemotherapy resistance and a poor prognosis is of important clinical significance for patients with CRC.

Common clinicopathological variables, including differentiation and nerve or vessel invasion, demonstrate a relatively weak predictive power for discriminating among CRC patients with varied risks of clinical outcomes (3). Although several molecular characteristics, including chromosomal instability, microsatellite instability and the CpG island methylator phenotype, have been used as prognosis predictors for CRC (4,5), these genetic aberrations are not directly responsible for the responses or resistance to chemotherapeutic drugs.

Based on mRNA expression patterns, a series of gene signatures have been developed for CRC, which are able to divide CRC patients into subclasses that present distinct prognostic profiles (6–8). Considering the association between chemotherapy efficiency and CRC prognosis, the present study hypothesized that drug resistance-associated genes may be used to build gene signature models to predict the clinical outcomes for patients with CRC.

In the present study, using the previously published CRC mRNA microarray expression datasets deposited in the Gene Expression Omnibus (GEO) database, two gene signatures associated with acquired resistance to irinotecan or oxaliplatin were established. These individual and combined gene signatures demonstrated a predictive power to stratify patients with CRC into good or poor survival groups.

 Materials and methods <sec> <title>Datasets

Four gene expression microarray datasets for CRC deposited in the GEO database (https://www.ncbi.nlm.nih.gov/geo/) were used for data mining in the present study. Gene expression data for GSE42387 was obtained from three human colon cancer cell lines (HCT116, HT29 and LoVo) and their sub-cell lines with acquired resistance to irinotecan (active metabolite SN-38) or oxaliplatin (9). GSE42387 was used to identify irinotecan or oxaliplatin resistance-associated genes.

Three clinical CRC tissue gene expression microarray datasets were selected for the gene signature training and validation. Gene expression data of 226 CRC tissues with complete follow-up data from GSE14333 were used as a training dataset to establish a drug-resistance gene signature associated with clinical outcome (10). The gene expression data from two sets of CRC tissues, GSE17536 and GSE17537, were used as validation datasets to test the robustness of the gene signature in predicting survival (11,12). The clinical traits of patients in the three CRC cohorts are summarized in Table I.

 Identification of individual genes associated with acquired irinotecan or oxaliplatin resistance

Differentially expressed genes between the parent and resistant subclass cell lines from the GSE42387 dataset were analyzed by performing the univariate F-test with a randomized variance model and false discovery rate correction for each gene. A permutation test (n=10,000) was performed for each significant gene. Genes for which P<0.005 were selected for further analysis. All the analyses were performed using BRB-ArrayTools software (version 4.5) developed by Dr. Richard Simon and the BRB-ArrayTools Development Team (National Institutes of Health, National Cancer Center, Bethesda, MD, USA) (13).

 Construction of irinotecan or oxaliplatin resistance gene signatures associated with the survival of CRC

Significant genes identified as resistance-associated in the GSE42387 dataset were then applied to the GSE14333 training dataset involving clinical CRC samples. The univariate Cox proportional hazards model tool of BRB-ArrayTools was used to test for genes that significantly influenced patient survival (permutation time, 1,000). Genes with a significance threshold of 0.05 were selected. Hierarchical clustering of the CRC samples in the GSE14333 dataset was performed using the acquired resistance signatures.

Subsequently, gene signature models were developed based on a linear combination of the expression levels of survival-associated genes weighted by the regression coefficient derived from the univariate Cox regression analysis for the selected irinotecan- or oxaliplatin-resistance gene signatures. The patient groups were divided into high- or low-risk subgroups based on their risk scores being above or below the median value across all samples, respectively. The prediction accuracy of the gene classifiers was estimated by leave-one-out cross-validation.

 External validation of the models using expression data in two independent datasets of clinical CRC samples

The established gene classifiers were validated using datasets from two independent cohorts of patients with CRC (GSE17536 and GSE17537). Gene expression profiles were used to predict survival classification (high- or low-risk) using the same models developed in the training sets.

 Effect of combined drug resistance gene signatures on predicting clinical outcomes in CRC datasets

The predictive results of irinotecan- and oxaliplatin-resistance gene signatures, respectively, were combined and the patients of the training and validation datasets were further dichotomized into four sub-classes according to survival risk as follows: High-/high-, low-/high-, high-/low-, and low-/low- subgroups.

 Statistical analysis

Categorical variables were compared using the chi-squared test. Except for analyses that were performed using BRB-ArrayTools, all other statistical analyses were performed using MedCalc® software (version 8.1; MedCalc® software, Ostend, Belgium). Survival between groups was compared using the Kaplan-Meier method, and the differences were assessed using the log-rank test. P<0.05 was considered to indicate a statistically significant difference.

 Results <sec> <title>Determination of irinotecan- and oxaliplatin-resistance gene signatures for CRC using microarray datasets

Using the microarray results of the irinotecan- and oxaliplatin-resistant sublines and the parent cell lines from the GSE42387 dataset, 292 and 103 genes were identified to be significantly dysregulated in irinotecan- and oxaliplatin-resistant CRC cell lines, respectively, using a univariate F-test with a randomized variance model and false discovery rate correction in BRB-ArrayTools.

Subsequently, these resistance-associated genes were applied to the training cohort of clinical biopsies of CRC tumors (GSE14333). Using the Cox regression analysis, a 60-gene signature for irinotecan resistance and a 13-gene signature for oxaliplatin resistance were generated, which were associated with the poor survival of patients with CRC in the GSE14333 dataset (Tables II and III).

Using these two filtered gene signatures, a survival risk score system was developed by calculating a linear combination of irinotecan- or oxaliplatin-gene signature expression values weighted by their Cox regression coefficients. Based on the risk scores evaluated by the irinotecan- and oxaliplatin-resistance gene signatures, patients with CRC from the GSE14333 dataset were divided into two subclasses (high- and low-risk) according to the median values of the risk scores.

 The prognostic significance of irinotecan- and oxaliplatin-resistance gene signatures for CRC

As presented in Fig. 1A and B, the high-risk groups defined by the two resistance gene signatures in the GSE14333 dataset had significantly shorter overall survival times than the low-risk groups [irinotecan-resistance gene signature: hazard ratio (HR)=0.4607, 95% confidence interval (CI)=0.3369–0.6300, log-rank P<0.0001; oxaliplatin-resistance gene signature: HR=0.6119, 95% CI=0.4547–0.8233, P=0.0008]. Furthermore, the irinotecan- or oxaliplatin-resistance gene sets were able to hierarchically cluster the GSE14333 CRC tumors into two subgroups with varying distributions of outcome risk in an unsupervised manner (Fig. 2).

 Validation of the prognostic value of irinotecan- and oxaliplatin-resistance gene signatures in independent CRC cohorts

To further investigate the clinical relevance and evaluate the predictive value of the developed gene signature models, a preliminary evaluation in two independent cohorts of patient samples was performed. As presented in Fig. 1C-F, the robustness of the irinotecan- and oxaliplatin resistance-associated gene signatures may be also demonstrated in two large independent cohorts of patients with CRC: GSE17536 (irinotecan-resistance gene signature: HR=0.5318, 95%=0.3359–0.8419, P=0.0079; oxaliplatin-resistance gene signature: HR=0.5383, 95% CI=0.3400–0.8521, P=0.0114) and GSE17537 (irinotecan-resistance gene signature: HR=0.2827, 95% CI=0.1173–0.6813, P=0.0088; oxaliplatin-resistance gene signature: HR=0.2378, 95% CI=0.09773–0.5784, P=0.0023).

 Combination of irinotecan- and oxaliplatin-resistance gene signatures for the prognosis of CRC

The combination of the two signatures described above may further add prognostic power for the prediction of the survival of patients with CRC in training and validation datasets. As presented in Fig. 3, high-risk subgroups predicted by the two models had the poorest clinical outcomes in all three CRC cohorts.

 Discussion

In the present study, two sets of genes that may be used clinically to predict the prognosis in three independent CRC cohorts were identified from a series of experimental data on chemotherapy resistance. One gene set included 60 genes associated with irinotecan-resistance, while another gene set included 13 genes associated with oxaliplatin resistance. These gene lists were compared with the previously identified gene signature for CRC, revealing that cyclin K (CCNK) in the oxaliplatin-resistance signature in the present study was also identified in a 13-gene prognostic signature (ColoGuideEx) for stage II CRC prognosis (14), while CCNK in the irinotecan-resistance signature was included in a 42-gene signature predictive for sensitivity to radiochemotherapy in patients with locally advanced rectal cancer (15). Using Gene Set Enrichment Analysis, the present study identified that the 60-gene irinotecan-resistance signature significantly overlapped with genes that are dysregulated in numerous epithelial cancer cell lines overexpressing an oncogenic form of the KRAS GTPase proto-oncogene. It was additionally identified that these gene signatures were independent of other factors that affect the outcome of patients with CRC. Furthermore, the combination of these two novel signatures was shown to further improve their robustness in predicting the survival of patients with CRC.

Except for clinicopathological characteristics, responses to standard chemotherapy regimens, particularly standard first-line therapies such as irinotecan and oxaliplatin, are an important determinant affecting the clinical outcome of patients with CRC. Antitumor activity in irinotecan- and oxaliplatin-sensitive or refractory CRC is determined by the existing diverse gene expression patterns (9). Therefore, the results of the present study support the hypothesis that the patterns of expression of numerous drug-resistance genes may be successful in distinguishing between improved and poor outcomes for patients with CRC. However, different from other resistance-associated signatures for CRCs, these two gene classifiers are currently used to predict overall survival for CRC patients, but not to distinguish between sensitive and resistant tumors (9). The utility of the signatures identified in the present study for predicting chemotherapy activity remain to be clarified.

It may be hypothesized that individual genes in the two resistance signatures in the present study have functional significance in the process of developing irinotecan or oxaliplatin resistance. Among the two gene lists, transforming growth factor-β-induced transcript 1 has been validated as a potential biomarker to predict the effects of FOLFOX4 chemotherapy in patients with CRC (16). Tubulin α-1B chain (TUBA1B) in the oxaliplatin-resistance signature and inhibitor of DNA binding 4 (ID4), Ets-1, mucin 16, regulator of G-protein signaling 2, dual specificity protein phosphatase 4 and milk fat globule-EGF factor 8 in the irinotecan-resistance signature, have been identified as prognostic factors for CRC survival in previous studies (17–22). Functionally, TUBA1B was revealed to be associated with resistance to paclitaxel in patients with hepatocellular carcinoma (23). ID4 may participate in the chemoresistance associated with gain-of-function mutations in p53 (24). In particular, the two genes ectodysplasin A2 Receptor (EDA2R) and Von Willebrand factor A domain-containing 5B1 (VWA5B1) overlap between the irinotecan resistance- and oxaliplatin resistance-gene signatures. EDA2R, also termed XEDAR and tumor necrosis factor receptor superfamily member 27, is a member of the tumor necrosis factor receptor super-family, and is a novel p53 target with an important role in colorectal carcinogenesis (25–27). The function of the VWA5B1 gene remains unknown and its association with chemotherapy resistance has not yet been investigated. The results of the present study indicate that the roles of EDA2R, VWA5B1 and other gene signatures in the development of irinotecan or oxaliplatin resistance warrant further experimental investigation.

In conclusion, the present study developed two sets of gene signatures from chemoresistance-associated experimental microarray data that were able to effectively and reproducibly classify CRC tumors according to poor or improved prognosis. These findings highlight the importance of chemotherapy resistance in determining the prognosis of patients with CRC, and led the present study to propose that chemotherapy resistance-associated genes may be a novel source for the establishment of gene classifiers to categorize CRC into subgroups with varied clinical outcomes. Furthermore, the functional significance of these novel gene sets in developing drug resistance in CRC also requires further investigation.

 Acknowledgements

This study was supported by the Medicine and Health Project of Zhejiang Province (grant no. 2015RCB028), the Huzhou Science and Technology Fund (grant no. 2014GZ11), and the Science and Technology Planning Project of Guangdong Province (grant no. 2012B031800323).

 References GustavssonBCarlssonGMachoverDPetrelliNRothASchmollHJTveitKMGibsonFA review of the evolution of systemic chemotherapy in the management of colorectal cancerClin Colorectal Cancer14110201510.1016/j.clcc.2014.11.00225579803 WangFLiuRLeeSWSlossCMCougetJCusackJCHeparin-binding EGF-like growth factor is an early response gene to chemotherapy and contributes to chemotherapy resistanceOncogene2620062016200710.1038/sj.onc.120999917001310 BianchiPLaghiLDelconteGMalesciAPrognostic value of colorectal cancer biomarkersCancers (Basel)320802105201110.3390/cancers302208024212797 LaoVVGradyWMEpigenetics and colorectal cancerNat Rev Gastroenterol Hepatol8686700201110.1038/nrgastro.2011.17322009203 SimonsCCHughesLASmitsKMKhalid-de BakkerCAde BruïneAPCarvalhoBMeijerGASchoutenLJvan den BrandtPAWeijenbergMPvan EngelandMA novel classification of colorectal tumors based on microsatellite instability, the CpG island methylator phenotype and chromosomal instability: Implications for prognosisAnn Oncol2420482056201310.1093/annonc/mdt07623532114 WangLShenXWangZXiaoXWeiPWangQRenFWangYLiuZShengWA molecular signature for the prediction of recurrence in colorectal cancerMol Cancer1422201510.1186/s12943-015-0296-225645394 HurKToiyamaYSchetterAJOkugawaYHarrisCCBolandCRGoelAIdentification of a metastasis-specific MicroRNA signature in human colorectal cancerJ Natl Cancer Inst107pii:dju492201510.1093/jnci/dju49225663689 HuYChenHYYuCYXuJWangJLQianJZhangXFangJYA long non-coding RNA signature to improve prognosis prediction of colorectal cancerOncotarget522302242201410.18632/oncotarget.189524809982 JensenNFStenvangJBeckMKHanákováBBellingKCDoKNViuffBNygårdSBGuptaRRasmussenMHEstablishment and characterization of models of chemotherapy resistance in colorectal cancer: Towards a predictive signature of chemoresistanceMol Oncol911691185201510.1016/j.molonc.2015.02.00825759163 OhSCParkYYParkESLimJYKimSMKimSBKimJKimSCChuISSmithJJPrognostic gene expression signature associated with two molecularly distinct subtypes of colorectal cancerGut6112911298201210.1136/gutjnl-2011-30081221997556 SmithJJDeaneNGWuFMerchantNBZhangBJiangALuPJohnsonJCSchmidtCBaileyCEExperimentally derived metastasis gene expression profile predicts recurrence and death in patients with colon cancerGastroenterology138958968201010.1053/j.gastro.2009.11.00519914252 FreemanTJSmithJJChenXWashingtonMKRolandJTMeansALEschrichSAYeatmanTJDeaneNGBeauchampRDSmad4-mediated signaling inhibits intestinal neoplasia by inhibiting expression of β-cateninGastroenterology142562571.e2201210.1053/j.gastro.2011.11.02622115830 SimonRLamALiMCNganMMenenzesSZhaoYAnalysis of gene expression data using BRB-Array ToolsCancer Inform31117200719455231 AgesenTHSveenAMerokMALindGENesbakkenASkotheimRILotheRAColoGuideEx: A robust gene classifier specific for stage II colorectal cancer prognosisGut6115601567201210.1136/gutjnl-2011-30117922213796 RimkusCFriederichsJBoulesteixALTheisenJMagesJBeckerKNekardaHRosenbergRJanssenKPSiewertJRMicroarray-based prediction of tumor response to neoadjuvant radiochemotherapy of patients with locally advanced rectal cancerClin Gastroenterol Hepatol65361200810.1016/j.cgh.2007.10.02218166477 LiSLuXChiPPanJIdentification of Nkx2-3 and TGFB1I1 expression levels as potential biomarkers to predict the effects of FOLFOX4 chemotherapyCancer Biol Ther13443449201210.4161/cbt.1929822313639 UmetaniNTakeuchiHFujimotoAShinozakiMBilchikAJHoonDSEpigenetic inactivation of ID4 in colorectal carcinomas correlates with poor differentiation and unfavorable prognosisClin Cancer Res1074757483200410.1158/1078-0432.CCR-04-068915569977 TokuharaKOgataYNakagawaMShirouzuKEts-1 expression in vascular endothelial cells as an angiogenic and prognostic factor in colorectal carcinomaInt Surg882533200312731728 StreppelMMVincentAMukherjeeRCampbellNRChenSHKonstantopoulosKGogginsMGVan SeuningenIMaitraAMontgomeryEAMucin 16 (cancer antigen 125) expression in human tissues and cell lines and correlation with clinical outcome in adenocarcinomas of the pancreas, esophagus, stomach, and colonHum Pathol4317551763201210.1016/j.humpath.2012.01.00522542127 JiangZWangZXuYWangBHuangWCaiSAnalysis of RGS2 expression and prognostic significance in stage II and III colorectal cancerBiosci Rep30383390201010.1042/BSR2009012920001967 De VriendtVDe RoockWDi NarzoAFTianSBiesmansBJacobsBBudinskaESagaertXRossiSD'ArioGDUSP 4 expression identifies a subset of colorectal cancer tumors that differ in MAPK activation, regardless of the genotypeBiomarkers18516524201310.3109/1354750X.2013.81903823875912 JinushiTShibayamaYKinoshitaIOizumiSJinushiMAotaTTakahashiTHoritaSDosaka-AkitaHIsekiKLow expression levels of microRNA-124-5p correlated with poor prognosis in colorectal cancer via targeting of SMC4Cancer Med315441552201410.1002/cam4.30925081869 LuCZhangJHeSWanCShanAWangYYuLLiuGChenKShiJIncreased α-tubulin1b expression indicates poor prognosis and resistance to chemotherapy in hepatocellular carcinomaDig Dis Sci5827132720201310.1007/s10620-013-2692-z23625295 Dell'OrsoSGanciFStranoSBlandinoGFontemaggiGID4: A new player in the cancer arenaOncotarget14858201021293053 TanikawaCRiCKumarVNakamuraYMatsudaKCrosstalk of EDA-A2/XEDAR in the p53 signaling pathwayMol Cancer Res8855863201010.1158/1541-7786.MCR-09-048420501644 TanikawaCFurukawaYYoshidaNArakawaHNakamuraYMatsudaKXEDAR as a putative colorectal tumor suppressor that mediates p53-regulated anoikis pathwayOncogene2830813092200910.1038/onc.2009.15419543321 WongKKIzaguirreDIKwanSYKingERDeaversMTSoodAKMokSCGershensonDMPoor survival with wild-type TP53 ovarian cancer?Gynecol Oncol130565569201310.1016/j.ygyno.2013.06.01623800698

Kaplan-Meier curves of the low-risk and high-risk subgroups defined by the irinotecan-resistance and oxaliplatin-resistance gene signatures. (A) Irinotecan-resistance signature. (B) Oxaliplatin-resistance signature. The GSE17536 dataset's (C) irinotecan- and (D) oxaliplatin-resistance signatures. The GSE17537 dataset's (E) irinotecan- and (F) oxaliplatin-resistance signatures. These two gene signatures were able to stratify patients with colorectal cancer into low- and high-risk subgroups with significant differences in terms of the prognosis in the GSE14333 training dataset.

Unsupervised hierarchical clustering of the colorectal cancer samples from the GSE14333 dataset using drug resistance-associated gene signatures. The (A) irinotecan-resistance and (B) oxaliplatin-resistance gene signatures were able to hierarchically cluster the samples into two subgroups with varied distributions of outcome risk.

Kaplan-Meier curves based on the combined drug resistance gene signatures for patients with colorectal cancer in the three cohorts. In all three datasets, (A) GSE14333, (B) GSE17536 and (C) GSE17537, patients predicted as high-risk by the two gene signatures had the poorest clinical outcomes.

Clinical traits of patients in the three colorectal cancer datasets.

Characteristics GSE14333 GSE17536 GSE17537
Age, years
  <60   61   59 24
  ≥60 165 118 31
Gender
  Male 120 96 26
  Female 106 80 29
Location
  Left   93 N.A. N.A.
  Right 101 N.A. N.A.
  Unknown   32 N.A. N.A.
Grade
  1 N.A.   16   1
  2 N.A. 134 32
  3 N.A.   27   3
  Unknown N.A.   0 19
Stage
  1 41 24   4
  2 94 57 15
  3 91 57 19
  4   0 39 17

Irinotecan-resistance genes associated with the survival of patients with colorectal cancer in the GSE14333 dataset.

No. identified Gene symbol Accession no. Parametric P-value FDR Permutation P-value Hazard ratio SD of log intensities
  1 LIMS2 NM_001136037 0.0001736 0.0165 0.0002 0.642 0.601
  2 TUBA1B NM_006082 0.0005244 0.0249 0.0003 5.428 0.168
  3 KLHDC2 NM_014315 0.0008015 0.0254 0.0008 0.484 0.362
  4 PDGFC NM_016205 0.0018540 0.0412 0.0020 0.740 0.882
  5 TGFB1I1 NM_001042454 0.0022273 0.0412 0.0030 0.753 0.836
  6 ARHGAP24 NM_001025616 0.0029117 0.0412 0.0040 0.754 0.803
  7 PPAP2A NM_003711 0.0030347 0.0412 0.0029 0.666 0.620
  8 DNASE2B NM_021233 0.0071416 0.0848 0.0064 1.159 1.411
  9 KIRREL NM_001286349 0.0092085 0.0972 0.0088 0.713 0.618
10 EDA2R NM_001199687 0.0122321 0.1160 0.0132 1.214 0.995
11 VWA5B1 NM_001039500 0.0145063 0.1250 0.0132 1.203 1.002
12 DAPK1 NM_001288729 0.0320929 0.2540 0.0328 0.792 0.752
13 FBXL21 NM_012159 0.0422906 0.3090 0.0416 0.881 1.270

FDR, false discovery rate; SD, standard deviation.

Oxaliplatin resistance genes associated with the survival of patients with colorectal cancer in the GSE14333 dataset.

No. identified Gene symbol Accession no. Parametric P-value FDR Permutation P-value Hazard ratio SD of log intensities
  1 MMP16 NM_005941 0.0000037 0.000951 <1e-07 0.659 0.814
  2 GPHA2 NM_130769 0.0000248 0.00284 <1e-07 1.425 0.894
  3 IRX1 NM_024337 0.0000331 0.00284 0.0001 0.729 1.120
  4 FKBP6 NM_001135211 0.000192 0.01230 0.0001 0.710 0.829
  5 RAB17 NM_022449 0.0002755 0.01420 0.0005 1.752 0.570
  6 MYH7B NM_020884 0.0004024 0.01720 0.0003 0.855 1.728
  7 TLX2 NM_016170 0.000472 0.01730 0.0003 0.832 1.44
  8 HTR4 NM_000870 0.000621 0.01990 0.0009 0.737 0.700
  9 PIK3R5 NM_001142633 0.0007926 0.02260 0.0007 2.057 0.345
10 SCGB1D1 NM_006552 0.000989 0.02540 0.0014 0.805 1.177
11 P2RX3 NM_002559 0.0013767 0.03070 0.0017 0.825 1.326
12 ID4 NM_001546 0.0014566 0.03070 0.0012 0.788 1.090
13 A2M NM_000014 0.0017583 0.03070 0.0010 0.681 0.671
14 IPMK NM_152230 0.0017827 0.03070 0.0020 1.565 0.559
15 TEX13A NM_031274 0.0017903 0.03070 0.0022 0.817 1.120
16 CRYBB3 NM_004076 0.0020744 0.03330 0.0021 1.961 0.361
17 GPR112 NM_153834 0.002965 0.04480 0.0033 0.797 1.030
18 C12orf49 NM_024738 0.0031541 0.04500 0.0038 2.279 0.321
19 ECEL1 NM_004826 0.004649 0.06290 0.0056 0.736 0.76
20 CHADL NM_138481 0.0055728 0.06590 0.0063 0.778 0.782
21 CCNK NM_001099402 0.0056307 0.06590 0.0056 0.576 0.389
22 GAP43 NM_001130064 0.005916 0.06590 0.0075 0.767 0.809
23 TSPAN7 NM_004615 0.0060566 0.06590 0.0058 0.823 1.058
24 NOB1 NM_014062 0.0061516 0.06590 0.0059 1.691 0.403
25 SYT12 NM_001177880 0.0081639 0.08390 0.0089 1.203 1.114
26 NTM NM_001048209 0.0087813 0.08680 0.0091 0.854 1.245
27 SYCE2 NM_001105578 0.0095563 0.09100 0.0093 0.830 1.102
28 PRKACG NM_002732 0.0105488 0.09680 0.0119 0.834 1.127
29 RNF146 NM_001242844 0.0109557 0.09710 0.0124 0.593 0.384
30 KCNMA1 NM_001014797 0.0114415 0.09720 0.0121 0.855 1.245
31 ETS1 NM_001143820 0.0120369 0.09720 0.0112 0.707 0.59
32 EDA2R NM_001199687 0.0122321 0.09720 0.0132 1.214 0.995
33 ADARB2-AS1 NM_001098830 0.012478 0.09720 0.0121 0.815 0.858
34 RCBTB1 NM_018191 0.0136522 0.10200 0.0132 1.459 0.502
35 SNAI2 NM_003068 0.0144012 0.10200 0.0139 0.781 0.826
36 VWA5B1 NM_001039500 0.0145063 0.10200 0.0132 1.203 1.002
37 PXDN NM_012293 0.0147868 0.10200 0.0140 0.746 0.715
38 DUOX2 NM_014080 0.0155175 0.10200 0.0150 0.904 1.973
39 ADAMTS20 NM_025003 0.0155347 0.10200 0.0160 1.268 0.844
40 FGF11 NM_004112 0.0183579 0.11800 0.0171 0.791 0.759
41 ATXN1L NM_001137675 0.0191998 0.12000 0.0179 0.552 0.312
42 MUC16 NM_024690 0.0216876 0.12700 0.0231 1.165 1.141
43 APOBEC3B NM_001270411 0.0220595 0.12700 0.0228 1.226 0.917
44 RGS2 NM_002923 0.0222933 0.12700 0.0217 0.838 1.032
45 DUSP1 NM_004417 0.0225976 0.12700 0.0215 0.810 0.863
46 CHRNA1 NM_000079 0.0228582 0.12700 0.0240 0.806 0.774
47 MFGE8 NM_001114614 0.0232019 0.12700 0.0233 0.819 0.878
48 APOL5 NM_030642 0.026395 0.14100 0.0255 0.746 0.536
49 HAS2 NM_005328 0.0326012 0.17100 0.0324 0.803 0.737
50 AUH NM_001698 0.0354341 0.18200 0.0335 1.703 0.317
51 HAUS4 NM_001166269 0.0369736 0.18600 0.0360 1.445 0.400
52 ZFP57 NM_001109809 0.0377616 0.18700 0.0359 1.203 0.882
53 BOLL NM_001284358 0.040459 0.19600 0.0423 1.355 0.529
54 XPR1 NM_001135669 0.0415493 0.19800 0.0422 1.449 0.413
55 MLXIP NM_014938 0.0423385 0.19800 0.0430 0.770 0.575
56 CEACAM19 NM_001127893 0.0462696 0.21200 0.0461 1.118 1.423
57 KLF12 NM_007249 0.0481573 0.21400 0.0482 0.784 0.665
58 BLVRB NM_000713 0.0482109 0.21400 0.0474 1.286 0.588
59 GLTPD1 NM_001029885 0.0496136 0.214 0.0478 1.325 0.553
60 EID2 NM_153232 0.0498994 0.214 0.0528 0.605 0.309

FDR, false discovery rate; SD, standard deviation.