Visit-to-visit HbA1c Variability, Dementia, and Hippocampal Atrophy among Adults Without Diabetes

Chenxi Li,Junyan Guo,Yining Zhao,Kaili Sun,Zeinab Abdelrahman,Xingqi Cao,Jingyun Zhang,Zhoutao Zheng,Changzheng Yuan,Huiqian Huang,Yaojing Chen,Zuyun Liu,Zuobing Chen

DOI: https://doi.org/10.1016/j.exger.2023.112225

IF: 4.253

2023-01-01

Experimental Gerontology

Abstract:Objectives: Adults without diabetes are not completely healthy; they are probably heterogeneous with several potential health problems. The management of hemoglobin A1c (HbA1c) is crucial among patients with diabetes; but whether similar management strategy is needed for adults without diabetes is unclear. Thus, this study aimed to investigate the associations of visit-to-visit HbA1c variability with incident dementia and hippocampal volume among middle-aged and older adults without diabetes, providing potential insights into this question. Methods: We conducted a prospective analysis for incident dementia in 10,792 participants (mean age 58.9 years, 47.8 % men) from the UK Biobank. A subgroup of 3793 participants (mean age 57.8 years, 48.6 % men) was included in the analysis for hippocampal volume. We defined HbA1c variability as the difference in HbA1c divided by the mean HbA1c over the 2 sequential visits ([latter - former]/mean). Dementia was identified using hospital inpatient records with ICD-9 codes. T1-structural brain magnetic resonance imaging was conducted to derive hippocampal volume (normalized for head size). The nonlinear and linear associations were examined using restricted cubic spline (RCS) models, Cox regression models, and multiple linear regression models. Results: During a mean follow-up (since the second round) of 8.4 years, 90 (0.8 %) participants developed dementia. The RCS models suggested no significant nonlinear associations of HbA1c variability with incident dementia and hippocampal volume, respectively (All P > 0.05). Above an optimal cutoff of HbA1c variability at 0.08, high HbA1c variability (increment in HbA1c) was associated with an increased risk of dementia (Hazard Ratio, 1.88; 95 % Confidence Interval, 1.13 to 3.14, P = 0.015), and lower hippocampal volume (coefficient, -96.84 mm3, P = 0.037), respectively, in models with adjustment of covariates including age, sex, etc. Similar results were found for a different cut-off of 0. A series of sensitivity analyses verified the robustness of the findings.