Measuring Loss of Homeostasis in Aging
Individual biomarkers are often studied as indicators of abnormality, but a complex systems perspective suggests that further insight may be gained by considering biomarker values in the context of others. The concept of homeostasis implies that normal levels of a biomarker may be abnormal in relation to the levels of other biomarkers, and vice versa. On the premise that healthy physiological dynamics are constrained through regulation and thus converge towards certain profiles, results from our lab suggest that Mahalanobis distance (Dm), or the distance from the center of a distribution, can be used as a measure of physiological dysregulation. Specifically, Dm increases with age, and predicts mortality and many other health outcomes of age. Increase of signal with the inclusion of more biomarkers, and lack of sensitivity to biomarker choice confirm that dysregulation is indeed an emergent phenomenon. This approach can be applied at the organismal level or to specific physiological/biochemical systems. Here, in order to better understand the signal measured by Dm, we draw on the mathematical relationship between principal components (PCs) of the biomarkers and Mahalanobis distance. Our results characterize the relative distribution of biological information among the PCs, and suggest that careful removal of certain PCs in the calculation of Dm can significantly improve the biological signal.
KeywordsDysregulation Mahalanobis distance Principal components
- Cohen, A.A., Milot, E., Li, Q., Bergeron, P., Poirier, R., Dusseault-Bélanger, F., Fülöp, T., Leroux, M., Legault, V., Metter, E.J., Fried, L.P., Ferrucci, L.: Detection of a novel, integrative aging process suggests complex physiological integration. PLoS ONE 10, e0116489 (2015b)CrossRefGoogle Scholar
- Gastinel, L.N.: Principal Component Analysis in the Era of « Omics » Data. Principal Component Analysis, 23 (2012)Google Scholar
- Jolliffe, I.T.: Choosing a Subset of Principal Components or Variables. In: Principal Component Analysis, pp. 92–114. Springer, New York (1986)Google Scholar
- Reese, S.E., Archer, K.J., Therneau, T.M., Atkinson, E.J., Vachon, C.M., de Andrade, M., Kocher, J.-P.A., Eckel-Passow, J.E.: A new statistic for identifying batch effects in high-throughput genomic data that uses guided principal component analysis. Bioinformatics 29, 2877–2883 (2013)CrossRefGoogle Scholar