Project description:In AML, most patients are initiated on standard chemotherapy and afterwards assigned to a post-remission strategy based on genetically-defined risk categories. However, outcomes remain heterogeneous, indicating the need for novel biomarker tests that can rapidly and accurately identify high-risk patients, allowing better stratification of both induction and post-remission therapy. As patient outcomes are linked to leukemia stem cell (LSC) properties that confer therapy resistance and drive relapse, LSC-based biomarkers may be highly informative. We tested 227 CD34/CD38 cell fractions from 78 AML patients for LSC activity in xenotransplantation assays. Comparison of microarray-based gene expression (GE) profiles between 138 LSC+ and 89 LSC? fractions identified 104 differentially-expressed LSC-specific genes. To obtain prognostic signatures, we performed statistical regression analysis of LSC GE against patient outcome using a training cohort of 495 AML patients treated with curative intent. A score calculated as the weighted sum of expression of 17 LSC signature genes (LSC17) was strongly associated with survival in 4 independent datasets (716 AML cases) spanning all risk categories in multi-variate analysis; an optimized 3-gene sub-score (LSC3) was prognostic in favorable risk subsets. These scores were robust across GE technology platforms, including the clinically serviceable NanoString system (LSC17: HR=2.73, P<0.0001; LSC3: HR=6.3, P<0.02). The LSC17 and LSC3 scores provide rapid and accurate identification of high-risk patients for whom conventional chemotherapy is non-curative. These scores will enable evaluation in clinical trials of whether such patients may benefit from novel and/or more intensified therapies during induction or in the post-remission setting.
Project description:In AML, most patients are initiated on standard chemotherapy and afterwards assigned to a post-remission strategy based on genetically-defined risk categories. However, outcomes remain heterogeneous, indicating the need for novel biomarker tests that can rapidly and accurately identify high-risk patients, allowing better stratification of both induction and post-remission therapy. As patient outcomes are linked to leukemia stem cell (LSC) properties that confer therapy resistance and drive relapse, LSC-based biomarkers may be highly informative. We tested 227 CD34/CD38 cell fractions from 78 AML patients for LSC activity in xenotransplantation assays. Comparison of microarray-based gene expression (GE) profiles between 138 LSC+ and 89 LSC? fractions identified 104 differentially-expressed LSC-specific genes. To obtain prognostic signatures, we performed statistical regression analysis of LSC GE against patient outcome using a training cohort of 495 AML patients treated with curative intent. A score calculated as the weighted sum of expression of 17 LSC signature genes (LSC17) was strongly associated with survival in 4 independent datasets (716 AML cases) spanning all risk categories in multi-variate analysis; an optimized 3-gene sub-score (LSC3) was prognostic in favorable risk subsets. These scores were robust across GE technology platforms, including the clinically serviceable NanoString system (LSC17: HR=2.73, P<0.0001; LSC3: HR=6.3, P<0.02). The LSC17 and LSC3 scores provide rapid and accurate identification of high-risk patients for whom conventional chemotherapy is non-curative. These scores will enable evaluation in clinical trials of whether such patients may benefit from novel and/or more intensified therapies during induction or in the post-remission setting.
Project description:Background: The relationships between cancer stem cells (CSCs), epithelial-to-mesenchymal transition (EMT), and the tumor microenvironment (TME) in bladder urothelial carcinoma (BLCA) remain unclear. Methods: We first constructed tumor stemness (TS) score using principal component analysis to quantify tumor stemness in BLCA. Then, we evaluated the clinical value of the TS score for predicting the response to tumor immunotherapy using immunotherapy cohorts. Finally, we built an EMT cell model by treating T24 cells with TGF-β and validated the relationship between the TS score and the EMT process in tumors by real-time quantitative PCR, cell invasion assays, and RNA-seq. Results: A TS scoring system was established with 61 TS-related genes to quantify the TS. The prognostic value of the TS score was then confirmed in multiple independent cohorts. A high TS score was associated with high EMT activity, CSC characteristics, high stromal cell content, high TP53 mutation rate, poor prognosis, and high tumor immunotherapy tolerance. Conclusion: The TS score provides an index for EMT and CSC research and helps clinicians develop treatment plans and predict outcomes for patients.
Project description:Patients with acute myeloid leukemia (AML) suffer dismal prognosis and the most adverse subpopulation within each tumor determines patient’s prognosis. To better understand challenging features in AML, we studied individual stem cells from a single AML sample, complementing genomic with in vivo functional studies. Primary tumor cells from an AML patient’s first and second relapse were transplanted into NSG mice to establish serially transplantable patient derived xenografts (PDX). In an innovative approach, twelve derivative PDX clones were generated thereof, each derived from a single AML stem cell as proven by molecular barcoding, and were color-marked to facilitate multiplex competitive in vivo assays. PDX clones consisted of four different genomic clusters; one cluster displayed resistance against Cytarabine treatment, while two other clusters harbored increased stem cell potential, indicating that stemness and treatment resistance had evolved independently in the sample. In vivo functional data correlated closely with the phylogenetic tree calculated from exome data. The Cytarabine-resistant cluster was characterized by a distinct gene expression profile, and a score thereof predicted outcome in large clinical patient data cohorts. Taken together, we provide proof of concept that intra-sample heterogeneity mimics inter-sample heterogeneity in AML. Stem cell disparities within a single sample allow insights into adverse characteristics of general importance for AML.
Project description:We aimed to predict obesity risk with genetic data, specifically, obesity-associated gene expression profiles. Genetic risk score was computed. The genetic risk score was significantly correlated with BMI when an optimization algorithm was used. Linear regression and built support vector machine models predicted obesity risk using gene expression profiles and the genetic risk score with a new mathematical method.
Project description:Nowadays, there are different ICU scoring systems to predict the likelihood of mortality, such as Acute Physiology And Chronic Health Condition (APACHE), Sequential Organ Failure Assessment (SOFA), and SAPS (Simplified Acute Physiology Score). Theses risk scores are based on the use of physiologic and other clinical data. However, the use of these score systems depend on the clinical trust in the reliability and predictions by physicians. In this work, we have evaluated the expression profile by microarray analysis from postsurgical patients with the aim of proposing a candidate set genes as a mortality risk score.