Meta‐GWAS of amyloid burden endophenotype combining PET and CSF results

• Published in: Alzheimer's & dementia Vol. 17; pp. e052333 - n/a
• Main Authors: Puerta, Raquel, de Rojas, Itziar, Hernandez, Isabel, Montrreal, Laura, Lage, Carmen, Quintela, Inés, Aguilera, Nuria, García‐González, Pablo, Sotolongo‐Grau, Oscar, S, Alonso‐Lana, Rodríguez, Eloy Rodríguez, Orellana, Adelina, Marquié, Marta, Sáez, María Eugenia, Carracedo, Angel, Tárraga, Lluís, Moreno‐Grau, Sonia, Boada, Mercè, Sanchez‐Juan, Pascual, Ruiz, Agustín
• Format: Journal Article
• Language: English
• Published: United States 01-12-2021
• ISSN: 1552-5260; 1552-5279
• DOI: 10.1002/alz.052333

Background
It is well known that Alzheimer’s disease (AD) has a strong genetic component. Even though it is a highly heritable disease, a big fraction of AD heritability remains to be elucidated. Genetic analysis of endophenotypes tightly linked to the disease might help to identify or confirm AD genetic risk factors.
Methods
GWAS results Amyloid β (Aβ42) in cerebrospinal fluid (CSF) or positron emission tomography (PET) measures were combined in a single meta‐analysis. Six independent GWAS were run using a generalized linear model with four principal components and status as a covariate (PLINK software). The datasets had data for two different amyloid endophenotypes: CSF‐Aβ42 levels from ADNI, Marqués de Valdecilla Hospital (Santander) and Fundació ACE (N= 1051); or PET‐SUVR (amyloid global standardized uptake ratio) from ADNI and Fundació ACE (N= 607). GWAS results were meta‐analysed using the sample size weighted method (METAL software). To reduce false positive findings, only genetic signals with opposite effect direction in PET and CSF endophenotypes were considered. Finally, we tried to replicate previously known genes for AD (de Rojas et al., MedRxiv 2019).
Results
After selecting SNPs with opposite effect on CSF and PET associations (N= 325051), we only observed genome wide significance in APOE locus (rs429358, P‐value= 2.202×10‐48) in the meta‐GWAS. We also observed several suggestive signals in chromosomes 3, 5 and 6 (rs72977014, rs13164192 and rs4960198); further studies are needed to add evidence of AD association. Finally, relative to known AD genes we were able to replicate CR1 (rs4844610, P‐value= 0.045) signal.
Conclusion
By combining data from both β‐amyloid endophenotypes (CSF and PET) we were able to replicate some well‐known AD genes indicating that this approach could be a useful strategy to identify novel variants increasing the knowledge of AD genetic architecture.