Diversity in Aging-Related Neuroimaging Research

People of self-identified Black and Hispanic/Latinx ancestry are highly underrepresented in aging-related neuroimaging research.¹ In 1 of the largest multisite neuroimaging studies of age-related changes to date, reported enrollment included only approximately 5% Black and 2.5% Hispanic/Latinx participants.² The inclusion of Black and Hispanic/Latinx and other historically marginalized groups in neuroimaging research is critical for a number of reasons, beyond the fact that Black people in the US are twice as likely to develop dementia compared with white people.^3,4 The elevated risk of dementia and corresponding structural and functional brain changes are complex beyond “inclusion of race as a covariate” in statistical models for analyses, if included in studies at all. Rather, this elevated risk for dementia and brain-related changes on neuroimaging reflects a dynamic interplay of biopsychosocial factors throughout the lifespan. Inclusion of diverse cohorts who reflect the full range of biologic and health considerations, racial and ethnic identities, and psychosocial and economic factors is not only the ethically appropriate approach but also holds greater potential to advance scientific understanding of the unique and interactive contributions of these factors.

Health-Related Disparities

Several health-related risk factors for developing dementia and brain-related changes, such as metabolic (eg, type 2 diabetes mellitus), cardiovascular (CVD), and cerebrovascular (CBV) disease, are higher in Black and Hispanic/Latinx people and other historically marginalized populations.^5-7 Consequently, higher rates of increased white matter hyperintensity (WMH) burden and decreased cerebral perfusion have been observed alongside higher rates of CVD.^8-10 Functional and structural age-related brain changes, however, extend beyond consideration of group health disparities. Previous studies have shown a higher volume of overall deep WMHs and decreased cerebral perfusion among Black individuals that are independent of other CVD risk factors (eg, hypertension/higher diastolic blood pressure).^7,10 A study showed that the association between vascular disease history and WMHs was stronger among Black participants compared with white and Hispanic/Latinx participants, suggesting the greater risk of developing WMHs for Black individuals goes above and beyond CVD risk.⁸ Other differential effects of risk factors for Black individuals have also been observed (eg, sex, inflammation, family history of dementia),^11-13 further highlighting the need to include not only racial and ethnic identity, but also other social determinants of health and associated health disparities in dementia-related and neuroimaging research.

Psychosocial Factors & Social Determinants of Health

Beyond the elevated prevalence of health-related risk factors, greater consideration of psychosocial factors and social determinants of health, including racial and ethnic identity, as they relate to dementia and brain-related changes on neuroimaging is critical. Nonbiologic risk factors for dementia and brain-related changes, such as higher rates of adverse childhood experiences (ACEs), are disproportionally prevalent in Black, Hispanic/Latinx, and other underserved and underrepresented communities.¹⁴ These psychosocial factors, including higher rates of poverty and lower socioeconomic status (SES), are intertwined with racial and ethnic identities and health disparities. These factors also may directly contribute to volumetric differences observed across groups throughout development. For example, smaller hippocampal and prefrontal volumes in people with lower SES can be observed as early as childhood and adolescence.¹⁵

Untangling the influences of racial and ethnic identity, psychosocial factors, and health-related risks on brain-related changes on neuroimaging becomes even more complex considering, that, in addition to influencing these changes directly, some of these factors (eg, poverty and ACEs) are strong predictors of health behaviors and status in adulthood that compound the risk of adverse outcomes, (eg, the influence of WMH via the elevated risk of CVD).^6,7,16-20 Considering this, it is unsurprising that decomposing the effects of psychosocial factors, health disparities, and racial or ethnic identity on WMH and brain volume changes is highly challenging. Further evidence for the complex interactions among these factors is the observation that different factors can have different effects across groups defined by racial and ethnic identity. For example, education—commonly considered a proxy for cognitive reserve—attenuates adverse effects of structural brain changes (ie, WMH, cortical thinning, and hippocampal volume) on memory and language functions in white people but not for Black or Hispanic/Latinx people.²¹

Why Diversity & Inclusion in Research Matter

Racial and Ethnic Identities Influence Biologic Variation

Taken together, the findings highlighted above point to the fact that racial and ethnic identity, health-related factors, and psychosocial considerations interact to influence dementia risk and corresponding brain-related changes on neuroimaging. Consideration of holistic biopsychosocial factors is essential in understanding the risk of dementia, as well as identifying accurate biomarkers and therapeutics that are robust in the face of these complexities. In addition to the ethical and moral imperative to include diverse groups in all aspects of research, doing so is also beneficial because it can advance the field’s understanding of disease risk, onset, and progression. For example, as part of a pilot proteomic study of Alzheimer’s disease (AD), 185 AD-associated proteins were found to be differentially expressed in white participants compared with Black participants as a result of the diverse cohort that was included.²²

Inclusion of diverse populations is critical in aging, neuro-imaging, and clinical trial research, because interactions of racial identity and the molecular AD biomarker APOE Ε4 have been observed.¹¹ Inclusion of diverse groups ensures these biomarkers and associated neuroimaging and functional outcomes have diagnostic utility and generalizability to all members of the general population. Furthermore, inclusion of diverse populations within aging, neuroimaging, and clinical trial research ensures that observed biomarker changes (eg, serum or CSF molecular concentrations, volumetric changes) reflect comparable functional changes across all groups. For example, positive associations between cerebral blood flow with memory and executive functioning were observed as being stronger in white participants with primarily European ancestry compared with participants with South Asian or Afrocaribbean ancestry.²³ This illustrates how an intervention targeted to improve or maintain a particular function (ie, increasing cerebral perfusion) could lack efficacy for particular groups and the way in which a lack of diversity in research would fail to detect this difference.

Neuropathology studies further emphasize the need for diversity and inclusion, providing important data that must be integrated with neuroimaging and other biomarker studies to fully inform understanding of the biologic basis of dementia—a key determinant of prevention and therapy. The magnitude of this point is seen in results from a positron emission tomography (PET) study of amyloid deposition in a diverse cohort of adults age 67 to 88, who did not have dementia. A large effect of racial identity on β-amyloid levels—broadly comparable to the effect of APOE genotype in the study—was found only after controlling for vascular risk factors, WMHs, APOE genotype, cognitive status, age, and sex.²⁴

Additionally, mixed pathologies are more commonly seen in Black individuals in clinic-based studies, with a higher frequency of Lewy body disease in those who sought care.²⁵ Hispanic/Latinx individuals also have more mixed pathologies with both more severe vascular disease and notable differences in tau and hippocampal sclerosis.^26,27 How these pathologic manifestations translate to notable structural and functional neuroimaging changes and clinical signs and symptoms needs to be further studied in diverse populations.

Sex as a Biologic Variant

Neuroimaging findings for AD in male and female participants show that women with mild cognitive impairment tend to have greater atrophy rates and greater cognitive and clinical decline.^28-30 Similar sex-related differences on PET have been shown in 2 independent large-scale databases with greater rates of tau accumulation in female individuals.^31,32 These findings, although informative, had limited diversity within the cohort prohibiting investigation of the biopsychosocial complexities discussed earlier in this article.

Neuropathologic data also highlight differences across sex. Female participants in studies have been more likely to have neurofibrillary tangles, AD pathology, and mixed AD with cerebrovascular pathology,^33,34 whereas male participants are more likely to have cortical Lewy bodies and Lewy body disease without concomitant AD pathology.^34,35 No doubt, determining the biologic basis of dementia will require translational research considering racial and ethnic identity and sex-related differences of in vivo biomarkers measured from the prodromal period through postmortem neuropathologic assessment.

Consideration and inclusion of sex as a biological variant (SABV) in all analyses and research investigations, must become routine practice with analyses reported across the entire research spectrum from academic meeting presentations to peer-reviewed publications, clinical trial analyses, and outcome reports.

Conclusion

The importance of inclusion in research is essential because of the potential downstream effects on clinical implementation and patient utilization. A systematic review and meta-analysis of how dementia-related resource use differs across groups defined by racial and ethnic identity reported that Black and Hispanic/Latinx individuals were notably underrepresented in US clinical trials of investigational treatments for dementia and significantly less likely to enroll in research trials.³⁶ It could not be determined whether the underrepresentation is caused by low interest in enrollment, low trust of research in general, poor recruitment efforts by researchers to engage diverse populations, or some other factor.

In addition to concerns related to the generalizability of therapeutic efficacy, limited participation has potential for downstream effects on implementation and resource utilization. The same meta-analysis found that Black people were 30% less likely to be prescribed cholinesterase inhibitors, even in the context of comparable social service utilization across groups.³⁶ Ultimately, inclusion of diverse populations will allow for integrative consideration of biopsychosocial complexities and the potential for increased utilization of research results by those who have been historically underrepresented through increased engagement in all phases of research.