Unveiling complex diseases through patient stratification and omics analysisacutely decompensated cirrhosis as a case study

Résumé

Understanding complex diseases requires advanced methods to understand their inherent heterogeneity. Two effective approaches for unravelling this complexity are (I) patient stratification and (II) omics analysis. Combining these approaches can help identify clinically relevant patient subgroups and elucidate the molecular basis of such disease heterogeneity, paving the way for precision medicine. In this thesis, we illustrate the practical application of these approaches and demonstrate their effectiveness in acutely decompensated cirrhosis (AD), a complex disease with significant inter-individual variability. Our research aims to improve patient prognosis through these methods. Firstly, we developed ClustALL, a novel clustering approach for robust patient stratification using clinical data. ClustALL handles mixed data, missing values, and correlated variables, and is available as an R Bioconductor package. The ClustALL method proved effective in identifying AD patient subgroups. The derived AD subgroups showed relevant clinical profiles with demonstrated prognostic value across the course of the disease. Importantly, these results were validated in an independent AD cohort. We identified a high-risk subgroup of patients showing a marked acute systemic inflammatory profile, which can be associated with a dysfunctional immune response already described in AD. Secondly, given the association of AD with a dysfunctional immune response due to systemic inflammation, we studied the immune system at the molecular level to gather insights into the pathophysiology of AD and disease progression. Accordingly, we characterised the peripheral blood mononuclear cell populations of 16 AD patients and 4 healthy controls, profiling gene expression and protein surface markers at single-cell resolution. Our analysis revealed a specific CD14+ monocytic subpopulation in AD patients potentially linked to a higher risk of developing Acute-on-chronic liver failure (ACLF). These monocytes exhibited an exhaustion profile, which could lead to dysfunctional monocytes and contribute to disease progression. We defined and validated a transcriptional signature of these exhausted monocytes in two large independent AD cohorts. Overall, this thesis highlights the potential of bioinformatic tools to address patient heterogeneity. We developed a methodology for patient stratification using clinical data and demonstrated its application in AD. We also enabled the future general use of this novel approach by creating a user-friendly R package that can be used for other complex diseases. Furthermore, we have identified a monocyte subpopulation associated with AD progression by leveraging omics analysis, specifically single-cell analysis. These findings emphasise the importance of patient stratification and omics analyses in advancing precision medicine to improve patient management and outcomes.