A new genetic map developed in Brazil has identified 129 possible therapeutic targets to combat Alzheimer's disease. The study utilizes artificial intelligence and RNA analysis to expand the search for treatment methods and uncover new possibilities for diagnosing this disease, which currently lacks a cure.
Expanding Target Search
The work was conducted by the Brazilian startup BioDecision Analytics. The project aims to increase the number of investigated targets beyond proteins traditionally associated with Alzheimer's disease, such as beta-amyloid and tau. According to the publication Pesquisa para Inovação from FAPESP, scientists analyzed genetic data from thousands of samples to detect changes correlating with the disease. This work may help researchers and pharmaceutical companies find new avenues for developing medications and blood tests.
For decades, most research has focused on beta-amyloid. When this protein accumulates abnormally in the brain, it forms plaques that disrupt communication between neurons and cause inflammatory processes. For a long time, beta-amyloid was the only known target for Alzheimer's therapy, necessitating the exploration of other areas.
Study Methodology
To create the genetic map, the team used 2870 samples of brain tissue and blood from both healthy individuals and patients diagnosed with Alzheimer's disease. The data were obtained from the Sequence Read Archive (SRA), a public database supported by the US government. Brain samples included five different regions selected for their specific functions: the hippocampus (related to memory), the cingulate cortex (related to information processing), Wernicke's area (related to language), the visual cortex (responsible for vision), and Broca's area (involved in communication).
Researchers considered the separation of these areas crucial because the brain is not a homogeneous tissue; each region possesses its own genetic characteristics, and mixing this information could obscure significant changes.
Role of Artificial Intelligence
Data processing was carried out using the BDASeq® platform, developed by BioDecision with support from FAPESP's Small Business Innovation Research Program (PIPE). The tool analyzes RNA sequencing to identify genes whose activity changes in people with Alzheimer's disease. Since RNA is involved in protein production in cells, these changes may indicate processes involved in the development of the disease.
João Rafael Dias Pinto, co-founder of BioDecision, explains that combining various statistical methods helps reduce errors. He noted that most studies use only one statistical method, leading to a large amount of noise.
The platform integrates eight statistical techniques and includes an artificial intelligence layer that correlates genetic data with clinical information. Thus, researchers can prioritize changes with the greatest potential for creating drugs or diagnostic tools.
Diagnostic Prospects
The analysis revealed over 4200 genes with altered expression in Alzheimer's patients. Among them, 75 genes are linked to neuronal plasticity—the ability of neurons to form new connections, a process affected by the disease. Eight genes from this group had not previously been indicated as potential therapeutic targets in prior studies. Another group consisted of 74 genes altered in both the brain and blood, making them candidates for biomarkers.
Of these genes, 21 showed an accuracy above 90% in differentiating between individuals with Alzheimer's disease and healthy individuals. Araldi stated that today, Alzheimer's diagnosis is clinical, lengthy, and imprecise. He added that this discovery provides the possibility of conducting diagnosis through blood analysis, which is much more accessible.
The results were presented at the AD/PD Congress held in Copenhagen, Denmark. Although the findings are promising, researchers emphasize that several more stages are required before drugs or tests based on these findings become available. BioDecision also plans to expand access to the platform via the cloud so that small laboratories can perform advanced analysis without depending on large computing structures.