By Alon Seifan, MD, MS
Assistant Professor of Neurology,
Weill Cornell Medical College,
Memory Disorders Program

Dr. Seifan reports no financial relationships relevant to this field of study.

Synopsis: In a prospective, longitudinal, cohort study of an asymptomatic, multi-ethnic Dallas community, brain MRI biomarkers measuring volume were associated with cognitive functions, as measured by the Montreal Cognitive Assessment.

Source: Gupta M, et al. Association of 3.0-T brain magnetic resonance imaging biomarkers with cognitive function in the Dallas Heart Study. JAMA Neurol 2015;72:170-175. doi:10.1001/jamaneurol.2014.3418.

In persons presenting with neurological complaints, the clinical significance of MRI biomarkers is often obvious. Neurologists are expert at determining whether pathology explains a given symptom. In community-dwelling adults who don’t seek neurological attention, however, the clinical significance of brain differences, including differences in white-matter disease distribution and cortical and subcortical atrophy, remains to be fully understood. Age-related brain changes are difficult to quantify using visual inspection alone. A significant proportion of asymptomatic adults harbor findings that might be considered pathologic in symptomatic individuals. In addition, individuals often harbor multiple simultaneous brain pathologies. These are critical research challenges because a better understanding of the clinical significance of readily available brain MRI biomarkers could help with earlier identification of individuals at risk for neurological disease. The study by Gupta et al takes a step forward by documenting cross-sectional associations between brain MRI biomarkers and cognitive outcomes in a younger, community-dwelling sample.

The Dallas Heart Study II was an extension of the initial Dallas Heart Study, a longitudinal, multi-ethnic, population-based cohort study of Dallas County residents. Between 2007 and 2009, a total of 2082 participants underwent brain MRI imaging and brief cognitive testing. For this study, non-English speakers and persons with known brain disease were excluded, leaving a final sample of 1645 subjects, which was 60% female, 48% African American, and 15% Hispanic, with a mean age of 50 years and an educational level of 13 years. Cognition was assessed using the Montreal Cognitive Assessment (MoCA). Brain volume was measured using 3.0-T MRI and quantified using the functional MRI of the brain software. The following volumes were quantified: white matter hyperintensity volume (WMH), total brain volume (TBV), gray matter volume (GMV), white matter volume (WMV), cerebrospinal fluid volume (CSFV), and hippocampal volume (HCV).

Results demonstrated a statistically significant (although low in magnitude) relationship between each MRI brain biomarker and total MoCA scores after adjusting for demographics (age, sex, years of education, self-reported race/ethnicity), with the exception of the association between white matter hyperintensity volume and MoCA score, which lost statistical significance after adjusting for demographics. Of the six MRI biomarkers, a combination of three (GMV, CSFV, and HCV) together formed the best predictive model related to total score on the MoCA. Years of education modified the association between cognition and each MRI biomarker except for WMH. Of all subdomains on the MoCA, the visuospatial/executive domain had the strongest correlation with brain volumes.

The results suggest that clinically significant changes in brain volume can be detected in community-dwelling adults, even at younger ages. The small size of the association (1% of variance in cognition) can likely be explained by the fact that the MoCA (due to its brevity) is not the most nuanced of cognitive measures. The findings confirm a protective effect of education on cognition in the presence of pathology. The findings suggest that prior observations of clinically significant consequences of cortical atrophy in older individuals might also extend to younger individuals. Importantly, the study was done in a multi-ethnic population that was mostly non-Caucasian, allowing generalization of previous brain biomarker study results to a more diverse population.

Strengths of the study include the fact that it was population-based, it included several measures of brain anatomy, and it included individuals of diverse backgrounds. Population-based sampling minimizes selection bias and may allow for more generalizable conclusions. Using population-based approaches also captures samples with a high prevalence of vascular risk factors. Prior associations between MRI biomarkers and cognitive outcomes have yet to be replicated in more diverse populations. Studying several different MRI biomarkers at once, each of which represents a different type of potential underlying pathology, allows for speculation about the mechanisms by which brain anatomy might relate to cognition. For example, hippocampal volume may represent underling AD pathology, and white matter changes or ventricular size could represent cerebrovascular disease.


A few limitations are worth mentioning. The quantification method used in this study did not allow for region-of-interest based analysis. This precludes any conclusions to be drawn regarding specific brain structures. Also, although the authors performed some analyses to justify treating age and brain atrophy as co-linear, it is currently understood that the relationship between brain atrophy and age is not linear. Also, the study did not account for presence of neurodevelopmental disorders such as learning disability; that is critically important because up to 10% of adults may have atypical neuroanatomy related to neurodevelopment. In fact, a host of early life factors besides education can influence the final attained brain size and cognition, including total brain size and hippocampal volume. Although the study used multiple MRI biomarkers, it could not account for the inter-relatedness of each of these structures. Connectivity-based approaches are important because brain volumes are intricately interconnected.

The next wave of cognitive imaging biomarker studies should include an even more comprehensive set of imaging biomarkers, including functional, connectivity-based, and ligand-based methods. Future studies need to account for early life exposures, including childhood adversity, socioeconomic status, and nutritional deprivation. Also, future studies are required to extend these findings to populations in which vascular risk factors are not so prevalent. Addressing potential vascular components using functional imaging is important to explain some of the variance in cognition (cognition depends on blood flow not just brain structure). Future studies are still required to fully account for the full range of neurodiversity that exists both within and across cultures. To truly test brain-behavior associations, future studies will need to use more nuanced cognitive measures. In the meantime, the study by Gupta et al suggests that even in younger individuals without overt neurological presentations, brain volumes matter.