Researchers are particularly intrigued by differences in the immune system. Women typically mount stronger immune responses than men and have more adaptive immune cells (B cells) that produce antibodies. These antibodies have different roles: some protect the body against foreign pathogens, like viruses and bacteria, while others, called defensive autoantibodies (AAbs), recognize the body’s own natural proteins and help clear dead or damaged cells. However, AAbs can malfunction and attack healthy cells and tissues, which leads to autoimmune diseases like multiple sclerosis and rheumatoid arthritis. Like Alzheimer’s disease (AD), autoimmune diseases disproportionally affect women. A few small-scale studies suggest that changes in levels of specific AAbs are associated with AD and that they occur even before the onset of amyloid plaque and tau tangle pathologies. Given these various threads, it is possible that autoantibodies contribute to AD, particularly in women. A better understanding of the role of AAbs in AD could lead to novel biomarkers, improved predictions about disease risk and tailored therapeutic strategies.
Dr. Chen leads the NYU Women’s Health Study, which has enrolled approximately 14,000 women between the ages of 35 and 65 who gave blood samples during the enrollment period (sometimes multiple). These women were followed for over 30 years through health-related surveys. These data are linked to clinical diagnoses and official death records through several databases. To date, this valuable cohort has been primarily used for cancer-related studies. However, the Chen team recently linked their data to the Centers for Medicare and Medicaid Services (CMS) database for people over 65, which opens the door for aging-related studies. Using the CMS data, Dr. Chen identified approximately 1,800 women in the NYU Women’s Health Study cohort who were clinically diagnosed with late-onset AD. With this new information, the Chen lab proposes to identify AAbs in blood samples collected from these women in midlife and determine if including this data improves the accuracy of predicting who will go on to develop AD. They are using a novel, comprehensive assay (protein microarray) to measure circulating AAbs. The assay was validated in a subset of blood samples, and AAbs targeting AD-related proteins—such as the amyloid precursor protein and TREM2—were reliably measurable and remained stable over time. Here, Dr. Chen hypothesized that naturally occurring defensive AAbs measured in midlife are associated with lower AD risk, whereas AAbs related to autoimmunity, neuroinflammation, or neurodegeneration are associated with increased risk of AD.
The Chen team is testing this hypothesis across two aims. In the first aim, they are identifying which midlife AAbs are associated with Alzheimer’s risk. They are including 120 AD cases and 120 matched controls from the larger cohort and ensuring that non-white participants comprise 50% of all samples. The top AAbs candidates—those most strongly associated with AD risk—will be used in the second aim. In Aim 2, several machine learning approaches are being used to build a computational model that reliably predicts AD risk. This model is integrating data on the AAbs identified in Aim 1, along with other key variables, including lifestyle risk factors such as education, smoking, obesity, physical activity and hypertension. The machine learning model is being trained on data from approximately 1,500 matched AD cases and controls. The team will determine whether including the top AAbs identified in Aim 1 improves AD risk prediction beyond traditional midlife risk factors.
At the end of the first year of funding, the team is making excellent progress toward both aims. Using samples from the NYU Women’s Health Study cohort, they have measured over 22,000 AAbs in 60 matched pairs of women and found that certain AAbs linked to neuroinflammation, neuronal health and brain structure are associated with later Alzheimer’s risk. These results provide the first prospective evidence that naturally occurring immune responses may play an early role in disease development. The team is measuring AAbs in another 60 matched pairs and combining these biological markers with midlife lifestyle and health data to test machine learning models in 120 matched pairs. These proof-of-concept analyses are showing it is feasible to integrate biological and non-biological factors to identify individuals at higher risk for AD. In the next year, they will expand these analyses to refine their computational models to determine if incorporating AAbs significantly improves risk prediction beyond traditional midlife risk factors. Together, these studies are advancing a new strategy for early Alzheimer’s detection based on immune and lifestyle signatures.
