¹Institute for Geriatric Research of Ulm University Medical Center at Agaplesion Bethesda Ulm, Ulm, Germany
²Ulm University, Institute of Epidemiology and Medical Biometry, Ulm, Germany

Text

Introduction: Underlying pathophysiological mechanisms behind frailty are still not fully understood. Available literature analyzing blood-based biomarkers is mainly based on univariate analysis [1]. Evaluations considering the interplay between biomarkers, taking sex-specific differences into account, are lacking. We performed a comprehensive sex- and system-specific analysis of 35 blood-based biomarkers with respect to frailty based on the concept of the accumulation of deficits.

Methods: Baseline data from the population-based ActiFE study (≥65 years) was used. Frailty was defined through a frailty index (FI). The FI represents the proportion of age-related deficits present at a given time point, and ranges from zero to one. Individuals with a FI >0.2 are identified as frail. Biomarkers were analyzed sex-, and organ-/system-specific (hematologic: hemoglobin, hematocrite, thrombocytes, leucocytes, iron, transferrin, ferritin; renal: creatinine, cystatin C, urea; hepatic: alanine transaminase (ALT), aspartate transaminase, gamma-glutamyl transpeptidase (GGT), albumin, protein; lipid: high and low density lipoprotein (HDL-, LDL-) cholesterol; endocrine: dehydroepiandrosterone (DHEA), vitamin D, testosterone, parathyroid hormone, sex hormone-binding globulin, thyroid-stimulating hormone, free thyroxine, free triiodothyronine (fT3); metabolic: lactate dehydrogenase (LDH), uric acid; cardiac: N-Terminal pro B-type natriuretic peptide (NT-proBNP), high-sensitivity cardiac troponin T (hs-cTnT), high-sensitivity cardiac troponin I growth factors: growth differentiation factor 15 (GDF15), insulin-like growth factor 1 and inflammatory markers: Interleukin 6 (IL6), hs C-reactive protein (hs-CRP); with respect to frailty. Models were built with standardized biomarkers using backwards selection in Generalized Linear Models (GLM) with a quasi-binomial distribution for continuous and Logistic Regression (LR) for dichotomized FI, adjusting for age, education, smoking, alcohol and medications. The likelihood ratio test was employed for model comparison during the backwards selection. Residual Mean Squared Error (RMSE) for GLM models and area under the curve (AUC), sensitivity (Se), specificity (Sp), positive and negative predictive value (PPV, NPV) for LR models were estimated. The ActiFe study was approved by the University Ulm ethical committee (application no. 318/08).

Results: Among 1180 participants in the ActiFe cohort with 683 being men (FI median (Q1, Q3): 0.117 (0.063, 0.188) and 18.3% frail) and 497 women (FI median (Q1, Q3): 0.133 (0.070, 0.203) and 25.2% frail), GLMs showed a good fit of the data with GGT, HDL-, LDL-cholesterol and GDF15 overall, and sex-specific transferrin, ALT, testosterone, vitamin D, LDH, NT-proBNP in men (RMSE 0.064, Se 0.33, Sp 0.96, PPV 0.63, NPV 0.86), and leucocytes, cystatin C, DHEA, fT3, hs-cTnT in women (RSME 0.074, Se 0.57, Sp 0.94, PPV 0.76, NPV 0.87). LR models included less biomarkers with similar properties (AUC 0.83, Se 0.72, Sp 0.80, PPV 0.45, NPV 0.93 in men; AUC 0.85, Se 0.86, Sp 0.72, PPV 0.51, NPV 0.94 in women).

Conclusion: Obtained models provide important insights into sex-specific pathways related to frailty. Surprisingly, inflammatory biomarkers did not play a role when taking all other biomarkers into account. In addition, the models show good accuracy and discriminatory properties, suggesting their preliminary value in predicting frailty in ambulatory settings, as reported in hospitals settings [2].

The authors declare that they have no competing interests.

The authors declare that an ethics committee vote is not required.

Literatur

[1] Fritzenschaft L, Boehm F, Rothenbacher D, Denkinger M, Dallmeier D. Association of blood biomarkers with frailty-A mapping review. Ageing Res Rev. 2025 May 1;109:102761. DOI: 10.1016/j.arr.2025.102761
[2] Mailliez A, Leroy M, Génin M, Drumez E, Puisieux F, Beuscart JB, Bautmans I, Balayé P, Boulanger E. Development and validation of a biological frailty score based on CRP, haemoglobin, albumin and vitamin D within an electronic health record database in France: a cross-sectional study. BMJ Public Health. 2025 Mar 23;3(1):e001941. DOI: 10.1136/bmjph-2024-001941