Assessment of Advanced Glycation End Products and Receptors and the Risk of Dementia

Abstract

Importance: Advanced glycation end products (AGEs) and their receptor (RAGE) are implicated in the pathophysiological processes of dementia and potentially underlie the association of diabetes with neurodegeneration. However, longitudinal studies examining this association are lacking.

Objective: To determine whether markers of the AGE-RAGE system are associated with prevalent and incident dementia and with cognition.

Design, setting, and participants: In this population-based cohort study including participants from the prospective Rotterdam Study, extracellular newly identified RAGE binding protein (EN-RAGE) and soluble RAGE (S-RAGE) were measured in plasma collected between 1997 and 1999 in a random selection of participants, and additionally in participants with prevalent dementia. Participants without dementia were followed up for dementia until 2016. Skin AGEs, measured as skin autofluorescence, and cognition were measured between 2013 and 2016 in participants without dementia. Data analysis was performed from June 2019 to December 2019.

Exposures: EN-RAGE, S-RAGE, and skin autofluorescence.

Main outcomes and measures: Prevalent and incident dementia and cognition, adjusted for potential confounders, including age, sex, diabetes, educational level, APOE ε4 carrier status, smoking, and estimated glomerular filtration rate.

Results: Of 3889 included participants (mean [SD] age, 72.5 [8.9] years; 2187 [56.2%] women), 1021 participants had data on plasma markers (mean [SD] age 73.6 [7.8] years; 564 [55.2%] women), 73 participants had dementia at baseline, and during 10 711 person-years of follow-up, 161 participants developed incident dementia. Compared with low levels, high EN-RAGE level was associated with a higher prevalence of dementia (odds ratio [OR], 3.68 [95% CI, 1.50-8.03]; P = .003), while high S-RAGE level was associated with a lower prevalence of dementia (OR, 0.37 [95% CI, 0.17-0.78]; P = .01). These associations attenuated in a longitudinal setting, with hazard ratios of 0.65 (95% CI, 0.42-1.01) for high EN-RAGE (P = .05) and 1.22 (95% CI, 0.82-1.81) for high S-RAGE (P = .33). Among 2890 participants without dementia (mean [SD] age, 72.5 [9.4] years; 1640 [57%] women), higher skin autofluorescence was associated with lower global cognitive function (adjusted difference in z score per 1-SD higher skin autofluorescence, -0.07 [95% CI, -0.11 to -0.04]), especially among carriers of the APOE ε4 allele (adjusted difference in z score per 1-SD higher skin autofluorescence, -0.15 [95% CI, -0.22 to -0.07]).

Conclusions and relevance: These findings suggest that the AGE-RAGE system is associated with cognitive decline and dementia cross-sectionally but not longitudinally. This indicates either a short-term association or reverse causality. Findings of cross-sectional associations between higher skin autofluorescence and lower cognitive function and an association with APOE status also warrant replication and prospective studies.

Figure 1.. Association of EN-RAGE and S-RAGE With Prevalent and Incident Dementia by Time Intervals

Associations of plasma levels with dementia incidence per cumulatively increasing duration of follow-up were obtained by censoring all participants still at risk at 4, 8, and 12 years after baseline, and after a total follow-up of 18.7 years. All estimates were adjusted for age, sex, diabetes, education, APOE ε4 carrier status, smoking, and estimated glomerular filtration rate. RAGE indicates Glycation End Products Receptor; EN-RAGE, extracellular newly identified RAGE binding protein; S-RAGE, soluble RAGE; HR, hazard ratio; and OR, odds ratio.

Figure 2.. Associations of Skin Autofluorescence With Cognitive Function

LDST indicates letter-digit substitution task; WFT, verbal fluency test; WLT, word learning test (delayed recall); and PPB, sum score Purdue pegboard tests, including tests on left hand, right hand, and both hands. The original score of the Stroop test (interference task) was inversely transformed. After transformation, a higher score corresponds to better performance. Coefficients are the difference in G-factor and z-scores of test parameters in association with 1-SD difference in skin autofluorescence, obtained from the linear regression adjusting for age, sex, diabetes, education, APOE ε4 carrier status, smoking status, and estimated glomerular filtration rate. For subgroup analysis by APOE ε4 carrier status, APOE ε4 carrier status was not used in the model for adjustment.