Multi-system prediction of frailty

Background

Advanced ageing is associated with reductions in mental and physical reserves, or increased frailty, which strongly hinders healthy ageing by increasing one’s risk of falls, hospitalisations, care home placement and mortality. This project aims to identify the most predictive indices of frailty, including biomarkers (genetic, epigenetic, proteomic, metabolomic, transcriptomic, and brain magnetic resonance imaging), lifestyle, physical, psychological, environmental, and social indicators. It will draw on existing deeply phenotyped longitudinal cohorts from the UK and Sweden. Social and environmental variables will include standard measures of socio-economic position but also extend to physical measures of the built-up environment, to social interaction and support, and lifestyle factors such as exercise, diet, and smoking; psychological variables will include personality, mood, and cognitive wellbeing, and physical measures will include anthropometrics and body health. Genetic predictors will be based on polygenic scores which are created for each individual based on genome-wide genotyping and the results of genome-wide association studies which can reliably identify genetic variants associated with traits like behaviour and disease. The genetic association effects for each variant are combined to give an overall genetic risk score based on the person's DNA profile. Epigenetic information relates to the expression of genes, whether a gene is turned on or off, and here we use epigenetic scores of ageing and protein levels that have been built using methylation data (whether specific chemicals are present that tell the gene to switch on or off) and protein levels themselves. Brain volumetric and connectivity indices of ageing will also be modelled in two cohorts. By integrating these different types of data we aim to produce the most predictive model of frailty and improve understanding of the additive and interactive effects of individual predictors. The findings could inform prevention strategies and/or intervention targets at a population level that will reduce levels of frailty in older age but may also offer superior individual level prediction.

Aims

To build a predictive model of frailty using multi-sytem data in a training sample and test its performance in independent samples of population based cohorts.

To include both additive and interactive effects to understand whether some variables can mitigate frailty risk caused by other interacting variables.

To understand whether some variables mediate the causal association between a target predictor and frailty. 

Training Outcomes

The student will learn complex statistical modelling skills, some that are routinely applied in genetics and epigenetics, and others that are more bespoke to this project, such as machine learning approaches to prediction.

They will be able to handle big data, learning computer programming skills in several languages (e.g., R, python), and gain experience in open science practices.

They will develop a broad knowledge in areas of psychology, imaging, genetics, epigenetics, proteomics and ageing, particularly of the frailty construct with two of the supervisors able to offer a clinical perspective.

They will develop their oral and written presentation skills for both lay and scientific audiences.

References

Miles Welstead, Natalie D Jenkins, Tom C Russ, Michelle Luciano, Graciela Muniz-Terrera, A Systematic Review of Frailty Trajectories: Their Shape and Influencing Factors, The Gerontologist, Volume 61, Issue 8, December 2021, Pages e463–e475, https://doi.org/10.1093/geront/gnaa061

Flint, J.P., Welstead, M., Cox, S.R., Russ, T.C., Marshall A., Luciano M. Validation of a polygenic risk score for frailty in the Lothian Birth Cohort 1936 and English longitudinal study of ageing. Sci Rep 14, 12586 (2024). https://doi.org/10.1038/s41598-024-63229-y

Clare Tazzeo, Debora Rizzuto, Amaia Calderón-Larrañaga, Serhiy Dekhtyar, Alberto Zucchelli, Xin Xia, Laura Fratiglioni, Davide Liborio Vetrano, Living Longer But Frailer? Temporal Trends in Life Expectancy and Frailty in Older Swedish Adults, The Journals of Gerontology: Series A, Volume 79, Issue 1, January 2024, glad212, https://doi.org/10.1093/gerona/glad212

Smith, H.M., Moodie, J.E., Monterrubio-Gómez, K…,…., Russ.T,C,, Cox, S.R., Marioni, R.E. Epigenetic scores of blood-based proteins as biomarkers of general cognitive function and brain health. Clin Epigenet 16, 46 (2024). https://doi.org/10.1186/s13148-024-01661-7