The Multi-Partner Consortium to Expand Dementia Research in Latin America (ReDLat): Driving Multicentric Research and Implementation Science

Abstract

Dementia is becoming increasingly prevalent in Latin America, contrasting with stable or declining rates in North America and Europe. This scenario places unprecedented clinical, social, and economic burden upon patients, families, and health systems. The challenges prove particularly pressing for conditions with highly specific diagnostic and management demands, such as frontotemporal dementia. Here we introduce a research and networking initiative designed to tackle these ensuing hurdles, the Multi-partner consortium to expand dementia research in Latin America (ReDLat). First, we present ReDLat's regional research framework, aimed at identifying the unique genetic, social, and economic factors driving the presentation of frontotemporal dementia and Alzheimer's disease in Latin America relative to the US. We describe ongoing ReDLat studies in various fields and ongoing research extensions. Then, we introduce actions coordinated by ReDLat and the Latin America and Caribbean Consortium on Dementia (LAC-CD) to develop culturally appropriate diagnostic tools, regional visibility and capacity building, diplomatic coordination in local priority areas, and a knowledge-to-action framework toward a regional action plan. Together, these research and networking initiatives will help to establish strong cross-national bonds, support the implementation of regional dementia plans, enhance health systems' infrastructure, and increase translational research collaborations across the continent.

Keywords: Alzheimer's disease; Latin America; SDOH; SES; dementia; fronto-temporal dementia; genetics; implementation science.

Figure 1

The ReDLat initiative. Systematic comparisons between LAC and US samples of AD and FTD via a novel, multimodal approach. The multimodal patterns will be assessed with different measures of (A) genetic risk (Aim 1), (B) imaging markers boosted by computational approaches, and (C) harmonized and novel measures of cognitive profiles and SDH (Aim 2). These data sources will be (D) integrated and compared among countries through machine learning (Aim 3) to unveil the main commonalities and differences between US and LAC samples. Tier 1 (T1): Larger study (Aim 1 & 3). Tier 2 (T2, smaller study with deep neurocognitive investigation (Aim 2 & 3). D, Data; Q&F, Quality & feature extraction; N, normalization; T, test; VS, visualization; Neuropsychiatric E, Neuropsychiatric evaluations. Reproduced with authorization from (1).

Figure 2

Machine Learning approach for the discovery of discriminant multidomain features between US and LAC patients with AD and FTD. Genetic, cognitive, SES, SDH, and imaging data are preprocessed with specific normalization methods to extract reliable features. After a feature-based quality assessment (QA), samples are separated in healthy controls (HCs) and patients. HC samples are used to apply normalization and harmonization methods over patient samples, enabling the correction of site-dependent bias in the data. Then, machine learning (ML) and data mining (DM) methods are used for multi-domain classification systems (CS), to find robust features, and to develop visualization dashboards. Each CS performs a progressive feature elimination process to find the most important features and to assess the stability of the model performance using a k-fold cross-validation over the training partition. Finally, performance and generalization in the classification are assessed via test independent partition from the training set.

Figure 3

ReDLat Kickoff meeting at San Francisco, CA. On January 27 2020, regional leaders, local investigators, ReDLat members, as well as authorities from the Global Brain Health Institute (GBHI), Alzheimer's Association, the Tau Consortium, the National Institute of Health (NIH) and other organizations met at UCSF in a US-Latin American Networking on Dementia Symposium. Co-hosted by GBHI and the UCSF Memory and Aging Center, the symposium served to launch ReDLat. Reproduced with authorization from (58).

Figure 4

ReDLat pre-existing data. (A) Estimates of cases with MRI (T1, rs-fMRI, or DTI) and/or DNA per country. (B) Number of participants with DNA and MRI data per diagnosis and per country. (C) Mutations already identified across countries. (D) Summary of the cognitive and functional assessments available in each country.

Figure 5

Testimonies from Peru highlighting different dimensions of the coronavirus outbreak and their impact on older people, and patients with cognitive decline and their families. The pictures above illustrate the people's vulnerabilities and the unpreparedness of the health system. Top left inset: Enrique (64 years old, Trujillo) suffers from diabetes mellitus but has been unable to get medication for 2 months. He is a shoe repairer with a small mobile stall and, after months of quarantine, he has to go out to work. Top right inset: Juana (64 years old, Trujillo) is a merchant diagnosed with coronavirus 3 months ago, which led to her needing supplemental oxygen and intravenous medications. Given the collapse of the hospitals, she was treated at home by her daughter. She thought she might lose her life, unable to perform simple activities (such as walking and eating) without great effort. Bottom left inset: Enedina (65 years old, Lima) lives with her youngest son who lost his job due to the pandemic restrictions. They live in a precarious room, without electricity, water or drainage. Bottom right Inset: On the other side of Lima, 83-year-old Mrs. Rosita lives with her family in a wealthy district. Her daughter has noted typical dementia symptoms, which have exacerbated since the quarantine. She doesn't understand the isolation, needs constant monitoring and urgently requires a neurological evaluation, but there are no services available due to the pandemic. Photos and testimonies from Peru documented by Alexander Kornhuber and Maritza Pintado Caipa. Individuals and relatives portrayed in the photos have provided written consent for reproduction. Reproduced with authorization from (126).

Figure 6

Dementia Public policies in Latin America. (I) Public policies accessibility. (A) Probability of response frequency regarding accessibility by sector. (B) Probability of response frequency regarding accessibility by age. (C) Probability of response frequency regarding accessibility by region. (D) Interaction of probability of response frequency of accessibility by country. (II) Public policies transmission. (E) Probability of response frequency regarding transmission by private sector. (F) Probability of response frequency regarding transmission by the public sector. (III) PPKI (public policy knowledge index). (G) Probability of response frequency regarding high PPKI by academic degree. (H) Probability of response frequency regarding high PPKI index by age. (I) Probability of response frequency regarding high PPKI by the public sector. (J) Probability of response frequency regarding PPKI by public region. (K) Probability of response frequency regarding PPKI by country. IV Aging. (L) Proportion of responses about aging stigma. (M) Proportion of responses about interest in aging and dementia manual. (N) Proportion of responses about interest in a data-sharing platform. Significance (p values): effects significance (**p ≤ 0.05, ***p ≤ 0.01), model significance (°p ≤ 0.1, °p ≤ 0.05, °p ≤ 0.01). Academic degree: 1: No reported education, 2: Technicians, 3: Tertiaries, 4; Certificates, 5: Undergrads, 6: Hospital Interns, 7: Post-graduate Specialization, 8: Master's Degree, 9: Ph.D. Reproduced with authorization from (127).

Figure 7

Priority levels assigned to core areas and challenges via a knowledge inquiry and related actions timelines. LAC-CD regional experts (N = 248) were presented with a survey and were asked to rank the 5 areas and associated challenges in order of priority. We calculated the percentage of respondents who rated these within the top two priorities and used these to rank both areas and challenges. The right inset shows the timeline for the proposed actions. Experts were also asked to deliver their views about a feasible timeline to address these challenges and actions (0–5 or 5–10 years) (% = Mean % of responses). Reproduced with authorization from (2).

Figure 8

Knowledge-to-action framework. The diagram captures challenges posed by dementia and the related mapping of key actions. Such actions are linked to specific working groups that have been included in the framework. This approach comprises a biomarker framework (LAC-BF), genetics and epidemiology workgroup (LAC-GEW), dementia platform (LAC-DP), clinical trial program (LAC-CTP), non-pharmacological interventions (LAC-NPI), and an LAC network for translational research (LAC-NTR). Reproduced with authorization from (2).