Radiology: Brain volume may indicate pre-clinical Alzheimers
The automated measurement of temporoparietal brain region volumes is a highly accurate predictor of memory loss in healthy elderly persons, indicating that these underlying characteristics could help clinicians identify likely cases of pre-clinical Alzheimer’s and enable them eventually to prevent the disease’s progression, according to an article published in the June issue of Radiology.

Recent studies have discovered significant differences in temporal and parietal lobe volumes between cognitively healthy adults, individuals with mild cognitive impairment and those with probable Alzheimer’s disease. These volumes also appear to be accurate predictors of which patients will progress to Alzheimer’s, noted Gloria C. Chiang, MD, from the department of radiology at the University of California, San Francisco (UCSF), and co-authors.

Importantly, volumetric models developed thus far to differentiate healthy and cognitively impaired elderly persons have relied on evidence of advanced memory impairment, when irreversible brain degeneration may already have taken place. “As a result,” Chiang and colleagues explained, “there is growing interest in identifying individuals who are at risk for AD at an early stage, when memory preservation still may be possible.”

In an attempt to identify still earlier indicators of likely Alzheimer’s, the authors measured temporoparietal brain volumes in 149 cognitively healthy elderly individuals using MRI. Participants, who came from 56 institutions as part of the Alzheimer’s Disease Neuroimaging Initiative, were re-examined two years later to assess changes to memory and their association with baseline brain volumes.

Twenty-five of 149 subjects were identified as having experienced clinically important memory loss over the course of the two-year study. No confounding associations between age or education and memory loss were observed in the cohort.

The researchers investigated volume in 15 regions of the brain, testing a number of models to evaluate the predictability of various regions, additively and in interaction with one another.

As more regions of the temporoparietal lobes were included in the models, volume became a more accurate predictor of clinically important memory loss. The most accurate model included eight brain regions and showed an accuracy of 81 percent (area under the receiver operating characteristic curve, ROC) for differentiating healthy patients from those who experienced memory loss over the course of the study.

The volumes of various regions proved substantially more powerful when calculated as interactive characteristics, rather than simply adding the effects of each region’s volume on memory loss, which yielded an ROC accuracy of just 58 percent.

“An unexpected finding was that pairwise interactions significantly improved predictive accuracy,” Chiang and colleagues observed. “The pairwise interaction terms account for simultaneous volume variations across separate brain regions. The results imply that the effect of multiple regions of brain atrophy on memory is greater than the additive effect of each region alone.”

The authors added that additional studies would be necessary to fully understand the effects of these interactions between different regions of the brain.

The most accurate model developed by the authors included volumes of the hippocampus, parahippocampal gyrus, amygdala, superior middle gyrus, inferior temporal gyrus, superior parietal lobe and posterior cingulate gyrus.

Chiang and colleagues highlighted the congruity between these regions of the brain and those identified by previous researchers as being significantly affected by Alzheimer’s disease. The authors argued that this finding provides additional support for the concept of preclinical Alzheimer’s.

The researchers emphasized the importance of additional studies with longer follow-up periods to validate their results and concluded that their findings indicate that latent differences exist in individuals that will likely progress to Alzheimer’s.

“We hypothesize that the decliners in our cohort may have had greater underlying Alzhimer’s disease-related abnormalities compared with the nondecliners, resulting in volumetric differences and subsequent memory decline. Furthermore, the decliners had marginally higher baseline memory scores, suggesting that volumetric differences may be more predictive of and precede clinical memory decline.”