Leader: Licia Iacoviello (NEUROMED); Other collaborator(s): Fabiola Olivieri (INRCA); Giacomo Frati (NEUROMED);
Multimorbidity is a condition often recurrent in the old age, although the biological basis of the comorbidity between different chronic conditions is often poorly characterized. Within this task we aim to better clarify potentially underlying shared mechanisms by analyzing existing datasets of multimorbid and non-multimorbid subjects, so to identify predictive biomarkers of multimorbidity and specific clusters of comorbidities, possibly characterizing them from a clinical, instrumental and biological point of view. This may have notable implications on common clinical practice for the follow-up and treatment of multimorbid patients.
Partner NEUROMED has finalized, submitted and revised the manuscript entitled “Blood-based biological ageing and red cell distribution width are associated with prevalent Parkinson's Disease: findings from a large Italian population cohort”, which was accepted for publication on Frontiers in Endocrinology (Gialluisi et al., in press). Moreover, the same Partner created, merged and harmonized several databases to test if and whether PhenoAge acceleration may predict the incident risk of several other chronic conditions in the Moli-sani cohort. Tested events include cardiovascular disease, type 2 diabetes, (breast, lung, prostate and colorectal) cancer, as well as all-cause and cause-specific mortality and first hospitalization risks. Partner NEUROMED plan to analyze these outcomes until the completion of the project.
Main policy, industrial and scientific implications: Phenoage acceleration could be a reliable predictor of incident risk of sevearl clinical phenotypes associated with aging
Partner NEUROMED: In a random test set of the Moli-sani cohort (N=4,772, the same used in previous analysis (doi: 10.1007/s10654-021-00797-7), we computed SHAP (SHapley Additive exPlanations) values, which allow to establish the importance of each feature, clarifying how it influences the label and explain the output of any machine learning model. We implemented also visualization of SHAP values through dedicated R packages like shapviz. We also computed and checked the performance of the BA and the resulting DeltaAge measure (index of blood age acceleration based on 15 most influential features) in the totality of participants of the Moli-sani cohort passing QC, for which a PhenoAge measure was also available (N=23,858).
Partner INRCA: The analysis of some inflammation-based scores created by evaluating CRP and albumin (CAR, GPS, mGPS and hs-mGPS) with short term mortality in geriatric hospitalized patients (REPORTAGE INRCA) was performed. A paper on this analysis has been submitted and is now under revision.
Partner NEUROMED tested whether PhenoAge acceleration may predict the incident risk of all cause and cause-specific mortality and hospitalization risk. In particular, we focused on cancer mortality and hospitalization risk in the Moli-sani cohort. PhenoAge’s influence on the incident risk of fatal/non-fatal events of breast, prostate, lung and colorectal cancer was tested in incrementally adjusted multivariable Cox PH regression models.
Partner NEUROMED tested the joint influence of PhenoAge and BloodAge on cancer mortality and first hospitalization risk, as well as on incident fatal/non-fatal events of breast, prostate, lung, renal and pancreatic cancer, through Cox PH regressions. This revealed interesting protective influences of BloodAge acceleration on breast and prostate cancer, which were further deepened through targeted sensitivity and interaction analyses. These included weighted Cox regressions, analyses after regressing out glucose levels from the Δage measures, stratified analyses by cancer subtypes and interaction analyses with several risk and protective factors like menopause status at baseline,