Background: DNA-methylation-based machine learning algorithms have demonstrated powerful diagnostic capabilities, and these tools are currently emerging in many fields of tumor diagnosis and patient prognosis prediction. This work aimed to identify novel DNA methylation diagnostic biomarkers for differentiating cervical cancer (CC) from normal tissues, as well as a prognostic prediction model to predict survival of CC patients. Methods: The methylation profiles with the available clinical characteristics were downloaded from the Gene Expression Omnibus (GEO) database and The Cancer Genome Atlas (TCGA) program. We first screened out the differential methylation sites in CC and normal tissues and performed multiple statistical analyses to discover DNA methylation diagnostic markers that are used to distinguish CC and normal control. Then, we developed a methylation-based survival model to improve risk stratification. Results: A diagnostic prediction panel consists of five CpG markers that could predict cervical cancer versus normal tissue with highly correct rate of 100%, and cg16428251, cg22341310, and cg23316360 which in diagnostic prediction panel all could yield high sensitivity and specificity for detection of CC and normal in six cohorts (area under curve [AUC] > 0.8), in addition to excellent performance in discriminating between CC and normal sample. The diagnostic marker panel also effectively predicted the CIN3 versus normal tissue with high accuracy in two datasets (AUC = 0.80, 0.789, respectively). Furthermore, a prognostic prediction model aggregated two CpG markers that effectively stratified the prognosis of high-risk and low-risk groups (training cohort: hazard ratio [HR] 4, 95% CI: 1.7-9.6, P = 0.0021; testing cohort: hazard ratio [HR] 1.9, 95% CI: 1.2-3.1, P = 0.0072). Conclusion:The findings of our study showed that DNA methylation markers are of great value in the diagnosis and prognosis of CC.

BackgroundCachexia is defined as an involuntary decrease in body weight, which can increase the risk of death in cancer patients and reduce the quality of life. Cachexia-inducing factors (CIFs) have been reported in colorectal cancer and pancreatic adenocarcinoma, but their value in diffuse large B-cell lymphoma (DLBCL) requires further genetic research.MethodsWe used gene expression data from Gene Expression Omnibus to evaluate the expression landscape of 25 known CIFs in DLBCL patients and compared them with normal lymphoma tissues from two cohorts [GSE56315 (n = 88) and GSE12195 (n = 136)]. The mutational status of CIFs were also evaluated in The Cancer Genome Atlas database. Based on the expression profiles of 25 CIFs, a single exploratory dataset which was merged by the datasets of GSE10846 (n = 420) and GSE31312 (n = 498) were divided into two molecular subtypes by using the method of consensus clustering. Immune microenvironment between different subtypes were assessed via single-sample gene set enrichment analysis and the CIBERSORT algorithm. The treatment response of commonly used chemotherapeutic drugs was predicted and gene set variation analysis was utilized to reveal the divergence in activated pathways for distinct subtypes. A risk signature was derived by univariate Cox regression and LASSO regression in the merged dataset (n = 882), and two independent cohorts [GSE87371 (n = 221) and GSE32918 (n = 244)] were used for validation, respectively.ResultsClustering analysis with CIFs further divided the cases into two molecular subtypes (cluster A and cluster B) associated with distinct prognosis, immunological landscape, chemosensitivity, and biological process. A risk-prognostic signature based on CCL2, CSF2, IL15, IL17A, IL4, TGFA, and TNFSF10 for DLBCL was developed, and significant differences in overall survival analysis were found between the low- and high-risk groups in the training dataset and another two independent validation datasets. Multivariate regression showed that the risk signature was an independently prognostic factor in contrast to other clinical characteristics.ConclusionThis study demonstrated that CIFs further contribute to the observed heterogeneity of DLBCL, and molecular classification and a risk signature based on CIFs are both promising tools for prognostic stratification, which may provide important clues for precision medicine and tumor-targeted therapy.

This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited.

Background An effective diagnostic and prognostic marker based on the gene expression profile of classic Hodgkin lymphoma (cHL) has not yet been developed. The aim of the present study was to investigate potential markers for the diagnosis and prediction of cHL prognosis. Methods The gene expression profiles with all available clinical features were downloaded from the Gene Expression Omnibus (GEO) database. Then, multiple machine learning algorithms were applied to develop and validate a diagnostic signature by comparing cHL with normal control. In addition, we identified prognostic genes and built a prognostic model with them to predict the prognosis for 130 patients with cHL which were treated with first-line treatment (ABVD chemotherapy or an ABVD-like regimen). Results A diagnostic prediction signature was constructed and showed high specificity and sensitivity (training cohort: AUC=0.981,95% CI 0.933–0.998, P<0.001, validation cohort: AUC=0.955,95% CI 0.895–0.986, P<0.001). Additionally, nine prognostic genes (LAMP1, STAT1, MMP9, C1QB, ICAM1, CD274, CCL19, HCK and LILRB2) were screened and a prognostic prediction model was constructed with them, which had been confirmed effectively predicting prognosis (P<0.001). Furthermore, the results of the immune infiltration assessment indicated that the high scale of the fraction of CD8 + T cells, M1 macrophages, resting mast cells associated with an adverse outcome in cHL, and naive B cells related to prolonged survival. In addition, a nomogram that combined the prognostic prediction model and clinical characteristics is also suggested to have a good predictive value for the prognosis of patients. Conclusion The new markers found in this study may be helpful for the diagnosis and prediction of the prognosis of cHL.

Background: Current international prognostic index is widely questioned on the risk stratification of peripheral T-cell lymphoma and do not accurately predict the outcome for patients. We postulated that multiple mRNAs could combined into a single model to improve risk stratification and to guide Clinicians implementing personalized therapeutic regimen for these patients. Methods: The gene expression profiles with clinical characteristics were selected and downloaded from the Gene Expression Omnibus (GEO) database. weighted gene co-expression network analysis (WGCNA) was used to screening genes in selected module which most closely related to PTCLs. Then build a gene classifier using a Lasso Cox Regression model and validated the prognostic accuracy of this mRNA signature in an internal validation cohort. Finally, a prognostic nomogram was constructed and performance was assessed by calibration plot and the concordance index (C-index). Results: 799 WGCNA-selected mRNAs in black module were identified and a mRNA signature which based on DOCK2, GSTM1, H2AFY, KCNAB2, LAPTM5 and SYK for PTCLs was developed. Significantly statistical difference can be seen in overall survival of PTCLs between low risk group and high risk group(training set :hazard ratio [HR] 4.3, 95% CI 2.4–7.4, p<0·0001; internal testing set :hazard ratio [HR] 2.4, 95% CI 1.2–4.8, p<0·01).Multivariate regression demonstrated that the signature was an independently prognostic factor contrast to age and gender. Furthermore, receiver operating characteristic analysis indicated that this signature exhibited excellent diagnostic efficiency for overall survival. Moreover, the nomogram which combined the six-genes risk signature and multiple clinical factors suggesting that predicted survival probability agreed well with the actual survival probability. Conclusions: The signature is a reliable prognostic tool for patients with PTCLs and it has the potential for clinicians to implement personalized therapeutic regimen for patients with stage PTCLs.