Predição de resistência à aspirina em gestantes com o uso de inteligência artificial e metilação global do DNA

Abstract

Cardiovascular diseases (CVD) represent the leading cause of mortality worldwide and in Brazil. The history of complications such as preeclampsia (PE) during pregnancy is the most relevant risk factor for assessing the cardiovascular (CV) risk of women in the premenopausal period. Aspirin, which inhibits platelet aggregation and thrombosis, is used in the secondary prevention of CVD and PE. However, the efficacy of this prophylaxis is limited in cases of aspirin resistance, as high-risk pregnant women may still develop PE despite appropriate and rational use of aspirin. Comorbidities such as chronic hypertension, obesity, and potential genetic factors lead to mechanisms that interact with oxidative stress and may promote platelet activation, reducing the individual response and effectiveness of aspirin. These factors may also be associated with nucleotide polymorphisms in platelet receptor genes. The use of artificial intelligence (AI) can help identify women who may develop PE due to aspirin resistance by establishing connections between clinical, laboratory, and genetic data in the assessment of aspirin responsiveness. Evaluating the global methylome of DNA in peripheral blood during pregnancy will allow for a comprehensive consideration of the genetic landscape in the occurrence of aspirin resistance and non-responsiveness to PE prophylaxis, as well as its future consequences such as CVD. This approach may contribute to new strategies by highlighting the vulnerability of these individuals and suggesting alternative treatments or doses for PE prevention in those who do not respond to standard aspirin doses. Given the high cost of methylome evaluation, the use of an AI model can optimize costs by quantifying the contribution of clinical, laboratory, and genetic data in predicting aspirin resistance during pregnancy, ensuring efficient resource utilization. The overall objective of this study is to predict aspirin resistance during pregnancy using an artificial intelligence model. This is an observational cohort study with two distinct phases: (A) Retrospective to develop and validate an artificial intelligence model for predicting aspirin resistance in pregnant women (IAGRA); (B) Prospective to validate the IAGRA model in predicting laboratory aspirin resistance and to correlate clinical aspirin resistance with laboratory aspirin resistance. The study population will consist of pregnant women who underwent a screening test, with risk calculation for PE using the algorithm from the Fetal Medicine Foundation, between October 2010 and December 2025.