Molecular bases of comorbidities and the impact of administered drug interactions

Abstract

The higher-than-expected probability of developing secondary diseases when already suffering from a previous – known as comorbidity – has become a significant health problem of the 21st century. Due to comorbidities, and magnified by aging, chronic diseases tend to accumulate, worsening patients’ quality of life, life expectancy, and hindering the choice of proper treatments. We can define two main sections within the framework of this thesis: one analyzing the patterns of co-administration of drugs known to interact, and the second studying the molecular bases of comorbidities. In the first one (Chapter I), we have analyzed Electronic Health Records from three populations with different healthcare systems: Blumenau (Brazil, public), Catalonia (Spain, public with co-payment), and Indianapolis (US, private). Stratifying by age, we have observed an increased risk of co-administering interacting drugs with aging that cannot be explained solely by higher co-administration rates. Stratifying by gender, we have found that women are at higher risk for co-administration of drug-drug interactions in the three populations, excepting men over their 50s in Indianapolis. In the second one, we have downloaded publicly available transcriptomic data analyzing disease and control samples. In Chapters II and III, we have conducted transcriptomic meta-analyses to study the similarities between differential expression profiles in Alzheimer’s disease, autism, and cancer. Through gene set enrichment analysis, we have observed that mitochondrial metabolism-related processes are altered in the same direction in Alzheimer’s disease and glioblastoma, and in the opposite direction in lung cancer, correlating with their described comorbidity relations (Chapter II). In the case of autism and cancer, oxidative phosphorylation and the immune system have been identified to be jointly altered in both diseases, with differential alterations depending on the cancer type (Chapter III). Finally, in Chapter IV, we have calculated similarities between patients’ differential expression profiles and used them to measure similarities between diseases, significantly recapitulating 25% of epidemiologically described comorbidities. Going one step further, we have grouped transcriptomically similar patients within each condition and measured similarities between subgroups of patients suffering from different diseases, identifying subgroup-specific similarities. The obtained results highlight the need to analyze comorbidities at the patient level