Identification Of Distinct Molecular Subtypes Of Uterine Carcinosarcoma

IntroductionUterine carcinosarcoma(UCS),also termed as malignant mesodermal mixed tumor or malignant mixed mullerian tumor(MMMT),is the most common one of uterine sarcomas.UCS occurs frequently in postmenopausal women and may also occur in premenopausal women.It is prone to misdiagnosis for the lack of obvious specificity in the clinic and the extremely difficult diagnosis before surgery.Its carcinomatous component is similar to high grade endometrioid,serous or clear cell endometrial carcinoma,UCS can thus be also regarded as Type II endometrial carcinoma.Although UCS is relatively rare,with an annual incidence rate at about 5.1-6.9 per 1,000,000 women,its mortality accounts for 15% of malignant tumors of the uterus due to the high metastasis and relapse rate and highly complex pathological context.Surgery-depend method is the main treatment of UCS,especially lymphadenectomy,which can significantly improve overall patient survival.Nevertheless,the five-year survival rate is still low(18%-39%).In the past decades,many malignant cancers(including bladder cancer,breast cancer,colon cancer,glioma,kidney cancer,leiomyosarcoma,ovarian cancer and prostate cancer)have been classified into different molecular subtypes based on their different molecular signatures.Such classifications had helped better understand these tumors,and had contributed to the development of more targeted and effective therapies.In this study,we describe the identification of two distinct molecular subtypes of uterine carcinosarcoma,each of which exhibits different gene expression patterns and clinicopathologic features.We also show that subtype II represents the high-grade UCS due to its high aggressiveness,and subtype II patients are less sensitive to the treatment than subtype I patients.The present study deepens our understanding of UCS and provides strategies to develop targeted therapies for UCS based on the different molecular subtypes.ObjectiveFirstly,the collection of UCS clinical samples lays the foundation of our experimental verification after data analysis.Subsequently,we analyzed expression profile data of GEO database and TCGA datasets by consensus clustering in order to identify two distinct UCS molecular subtypes,which was separately defined as subtype I and II.Furthermore,Significance Analysis of Microarrays(SAM-seq)were used to investigate two subtypes of differentially expressed genes which were analyzed by Gene Set Enrichment Analysis(GSEA),Gene ontology analysis(GO)and DAVID Functional Annotation.Cluster 3.0 and TreeView were used to do hierarchical clustering to view significantly over-expressed genes from each subtype.This research can bring a better understand the intrinsic molecular stratification in UCS and specific occurrence mechanism and provides strategies to develop targeted therapies for UCS based on the different molecular subtypes.MethodThirteen cases of UCS,which included rich clinical information(such as clinical symptoms,past history,family history,etc.)and pathological data(tumor grade after immunohistochemistry,genes Subtypes,etc.),were collected from general hospitals in Henan Province.Bioinformatic analyses: firstly,the RNA-seq databases of UCS were downloaded from both GEO(www.ncbi.nlm.nih.gov/geo/)and TCGA datasets.Results showed those analyzed expression-profile data from different databases was classified into two distinct molecular subtypes of uterine carcinosarcoma.Silhouette analysis(R package cluster)was then used to evaluate the accuracy of subtype assignments from Consensus Clustering Plus.Results validated that the two subtypes are the best type of UCS.Furthermore,the Sub-Map analysis verified that the independent subtypes analyzed from the two datasets GSE32507 and TCGA are highly correlated.Correlation analysis of molecular stratification and clinical information found that such clinical indicators as both the tumor invasiveness and weight were significantly associated with classification.Subtypes of differentially expressed genes were obtained by SAM-seq.Using GSEA,GO and DAVID,it was found that the two molecular subtypes of UCS presented different gene expression patterns.Lastly,Cluster 3.0 and TreeView were used to do hierarchical clustering to view TOP500 significantly over-expressed genes from each subtype,reflecting significant differences.Statistical analyses :The significance was assessed by the chi-square and Fisher exact tests and p value less than 0.05 was considered significant.ResultOur research separately analyzed the UCS databases downloaded from both GEO(www.ncbi.nlm.nih.gov/geo/)and TCGA datasets,and identified two different molecular subtypes(C1 and C2,L1 and L2 respectively).Correlation analysis of Sub-Map was used to verified the accuracy of subtype assignments.Through bioinformatics analysis,we found that two molecular subtypes of UCS have different gene expression patterns.We performed a gene set enrichment analysis(GSEA)and found that subtype II was enriched for genes associated with myoblast differentiation/muscle development,whereas the genes overexpressed in subtype I were associated with intercellular adhesion and apoptosis.Immediately,following a significant analysis(SAM-seq software)of highly expressed genes in each subtype,differentially expressed signaling pathways were obtained through Gene ontology(GO)and DAVID Functional Annotations.Such pathways as cell-cell adhesion,cellular movements,antigen processing and presentation,signaling transduction and enzyme activation were shown to be enriched in subtype I,whereas muscle development,muscle contraction,cytoskeleton filament and biosynthesis pathways were found to be enriched in subtype II.Hierarchical clustering analysis revealed a significant difference in significantly over-expressed genes from subtype I and II.In addition to different subtypes with significantly different gene expression patterns,each subtype also exhibits different clinicopathologic characteristics which can be downloaded from the TCGA database.We compared the clinicopathological characteristics of subtypes specific to UCS patients,and found that subtype I exhibited a low tumor invasion rate and a light weight tumor feature and subtype II presented a high invasion rate and a high weight characteristics.Subtype I may represent the low-grade UCS,and subtype II may represent the high-grade one.Conclusion1.Consensus clustering identifies two distinct molecular subtypes of UCS;2.Distinct molecular subtypes of UCS have different gene expression patterns.The target genes and signal pathways,highly enriched in different molecular subtypes,are indeed different.3.Distinct molecular subtypes of UCS have different Clinicopathologic characteristics.Patients with different subtypes have significant differences in tumor invasion rate,tumor weight,and stage.