The Signature Of Micro-metastasis And Relapse Risk After Surgical Resection For Early Stage ER+ Breast Cancer

About 20%-30% of early stage breast cancer patients suffer relapse after surgery,and these patients need adjuvant therapy to reduce the risk of relapse.Researchers have used molecular signatures to help identify breast cancer patients who had potential micro-metastases before surgery.However,previously reported molecular signatures that identify the risk of postoperative micro-metastasis and recurrence in breast cancer patients lack the robustness of sample data measured by different measuring platforms.Therefore,in order to solve the above problems,we developed a signature which is robust across multiple profiling platforms based on the within-sample relative expression orderings(REOs)in the samples.Using the gene expression profile data of breast cancer samples measured by Affymetrix platform and Illumina platform respectively,we identified gene pairs whose within-sample REOs were significantly stable in the lymph node-positive(LN+)samples and the lymph node-negative(LN-)samples,with reversal REOs in the two types of samples.In the discovery cohort,we identified gene pairs whose REOs were significantly correlated with relapse-free survival(RFS)using the univariate Cox regression model and validated in three independent data cohorts profile measured by two types of platforms(Affymetrix and RNA-sequencing platform).We further investigated the genomic characteristics between the relapse high-risk group and relapse low-risk group using The Cancer Genome Atlas(TCGA)samples with multi-omics data.Using the dataset GSE7390 and the dataset GSE6532,we constructed a prognosis signature consisting of five gene pairs,named 5-GPS.In an independent validation cohort GSE4922 measured by the Affymetrix platform,the predicted lowrisk 63 patients had a significantly better RFS than the predicted 53 high-risk patients(hazard ratio =2.08,95%(CI):1.04-4.19,p=3.56E-02,C-index=0.59).The signature was also validated in another independent cohorts GSE2034 measured by the Affymetrix platform get the similar result.We applied 5-GPS to the RNA-sequencing data of stage I-IV breast cancer samples documented in The Cancer Genome Atlas(TCGA),and found that the proportion of the high-risk samples in the total number of patient samples during their stage was increased as the stage level increased.Multi-omics analysis revealed that the genomes of the high-risk patients were distinctly characterized with a very high degree of genome instability compared with the lowrisk patients.In conclusion,the proposed REO-based signature can be applied to assess the relapse risk for postoperative early stage breast cancer patients at the individual level,which is robust across multiple profiling platforms.It has potential clinical conversion value and can be used to assist in optimizing the patients’ therapy plan.