By John Gever, Deputy Managing Editor, MedPage Today

Published: Oct 23, 2013

Full Story:  http://www.medpagetoday.com/Neurology/Dementia/42446

Variations in glycated hemoglobin levels within the normal range were modestly correlated with performance on certain cognitive tests and with differences in hippocampal structure, German researchers found.

In a study of 141 healthy middle age and elderly adults recruited from the community with mean glycated hemoglobin (HbA1c) levels of 5.8%, each standard deviation increase in HbA1c was associated with decreases of 0.18 to 0.27 standard deviations in measures of delayed recall, learning ability, and memory consolidation (all P<0.05), according toAgnes Floel, MD, of Charite-University Medicine in Berlin, and colleagues.

These decreases translated to R²correlation coefficients of 0.06 to 0.16 after adjusting for other factors, the researchers reported online in Neurology.

Additionally, they were related in turn to reductions in hippocampal volume and to increased mean diffusivity within the hippocampus as measured by brain MRI scans, the researchers found.

“The present findings might lead to a better understanding of the mechanisms underlying the effect of chronically elevated glucose brain function and structure, and the interaction between these factors,” Floel and colleagues wrote.

They added that “lifestyle strategies” to achieve strict glucose control could prevent age-related cognitive decline, even in individuals with HbA1c levels currently considered normal — a hypothesis that should be tested in future trials, they noted.

Previous studies had linked blood glucose with cognitive performance in patients with dysregulated glucose metabolism and in those with levels in the normal range. Others had found correlations between blood glucose and hippocampal alterations, and still others had shown that declines in cognition and memory are mirrored in hippocampal structural changes. The study by Floel and colleagues is the first to connect all three factors in the same population.

They recruited participants 50 to 80 years old by advertising publicly in Berlin with no particular inclusion criteria other than a body mass index value of 25 to 30. Individuals with diagnosed type 2 diabetes, severe untreated medical or psychiatric illness, major neurologic or psychiatric conditions, existing cognitive impairment, or high daily consumption of caffeine, alcohol, or cigarettes were excluded.

Participants underwent blood sampling, brain MRI scans, and the German version of the Rey Auditory Verbal Learning Test, which includes tasks related to delayed recall, learning ability, and memory consolidation.

Mean age for the 141 selected for the study was 63 (SD 6.9), with mean educational attainment of 16.2 years (SD 3.1). About 20% carried the APOE epsilon-4 allele that predisposes to early onset of Alzheimer’s disease.

HbA1c values in the group averaged 5.8% (39.8 mmol/mol, SD 2.93). Floel and colleagues found that each of the three cognition parameters evaluated was significantly associated with HbA1c levels in the range of 4.3% to 6.5%.

For each standard deviation increase in HbA1c, the following effects on the cognition parameters were seen in multivariate analysis, expressed as fractions of standard deviations (reflecting adjustments for age, sex, BMI, physical activity, smoking, systolic blood pressure, hippocampal volume and mean diffusivity, blood glucose, blood insulin, and APOE genotype):

• Delayed recall: -0.27 (P=0.001)
• Learning ability: -0.22 (P=0.005)
• Memory consolidation: -0.18 (P=0.037)

Sex, age, and hippocampal volume were each found to be independent predictors of performance on one or more of these measures, the researchers indicated.

The correlations between cognitive function and hippocampal parameters accounted for some but not all of the associations between cognition and HbA1c, Floel and colleagues noted.

Using simple mediation models, they calculated that “between 2% and 3% of the overlap in variances” among the three cognition parameters was explained by hippocampal volume and mean diffusivity.

Floel and colleagues cited several limitations of the study, including that it was not a true population-based analysis and that potential confounding variables may not have been addressed.