Abstract Alzheimer’s disease (AD) is one of the most complicated progressive neurodegenerative brain disorders, affecting millions of people around the world. Ageing remains one of the strongest risk factors associated with the disease and the increasing trend of the ageing population globally has significantly increased the pressure on healthcare systems worldwide. The pathogenesis of AD is being extensively investigated, yet several unknown key components remain. Therefore, we aimed to extract new knowledge from existing data. Ten gene expression datasets from different brain regions including the hippocampus, cerebellum, entorhinal, frontal and temporal cortices of 820 AD cases and 626 healthy controls were analyzed using the robust rank aggregation (RRA) method. Our results returned 1713 robust differentially expressed genes (DEGs) between five brain regions of AD cases and healthy controls. Subsequent analysis revealed pathways that were altered in each brain region, of which the GABAergic synapse pathway and the retrograde endocannabinoid signaling pathway were shared between all AD affected brain regions except the cerebellum, which is relatively less sensitive to the effects of AD. Furthermore, we obtained common robust DEGs between these two pathways and predicted three miRNAs as potential candidates targeting these genes; hsa-mir-17-5p, hsa-mir-106a-5p and hsa-mir-373-3p. Three transcription factors (TFs) were also identified as the potential upstream regulators of the robust DEGs; ELK-1, GATA1 and GATA2. Our results provide the foundation for further research investigating the role of these pathways in AD pathogenesis, and potential application of these miRNAs and TFs as therapeutic and diagnostic targets. Keywords: Alzheimer’s disease, GABAergic synapse pathway, retrograde endocannabinoid signaling, differentially expressed genes 1. Introduction Alzheimer’s disease (AD) is the leading cause of dementia, affecting between 70 to 80% of older adults with dementia [[48]1]. Currently, over 50 million people are living with the disease worldwide, and this number is estimated to rise to 150 million in 2050, exacerbating an already constrained healthcare system unless preventive strategies are implemented [[49]2,[50]3]. AD is characterized by initial memory loss and learning impairment, followed by cognitive dysfunction. The disability progresses significantly throughout the disease course, culminating in death within 5–12 years of the onset of symptoms [[51]3,[52]4]. The current treatments only provide symptomatic relief without mitigating disease progression. Thus, there are a growing number of studies focusing on potential therapeutic agents to combat AD more directly [[53]5,[54]6,[55]7]. Most of these studies are focused on two main pathological hallmarks of AD: senile plaques (SPs) composed of amyloid beta (Aβ) peptides; and neurofibrillary tangles (NFTs) composed of hyperphosphorylated tau proteins [[56]3,[57]4,[58]5,[59]8]. Although results of clinical trials have been underwhelming for the past 25 years, recently an anti-amyloid β antibody, Aducanumab, received an accelerated FDA approval, requiring further clinical trials to confirm the estimated efficacy [[60]1]. Repeated failure in clinical trials has challenged our understanding of this multifactorial disease, leading to recent studies concentrating on advancing our knowledge of the underlying mechanisms of AD pathogenesis to find druggable targets. High-throughput ‘omics’-based research including genomics, transcriptomics and proteomics has made a significant contribution to our current understanding of AD [[61]9,[62]10,[63]11]. However, currently the biological data are generated at a higher pace than they are being interpreted. Thus, there is an urgent need to summarize and extract new knowledge from the existing data. We have used meta-analysis to summarize and extract the most reliable data from existing results of multiple studies, taking advantage of the increased statistical power of larger combined sample sizes [[64]12,[65]13]. Several meta-analyses have been performed on microarray gene expression datasets of different brain regions to identify altered pathways involved in AD, as brain regions are differentially affected by AD during the course of disease progression [[66]12,[67]14]. The hippocampus is one of the earliest brain regions to be affected and thus most studies have been aimed at this region. However, data from the other regions, especially those that are less affected by AD such as the cerebellum, are not well-studied and are sometimes even excluded from meta-analyses [[68]15,[69]16]. Exploring these changes in less affected brain regions may open new avenues to enhance the molecular understanding of AD pathogenesis and may reveal key disease mechanism in affected brain regions [[70]17]. Therefore, in this study we have combined multiple gene expression datasets from five brain regions including the hippocampus, cerebellum, frontal, entorhinal and temporal cortices of AD patients and healthy controls, and used robust rank aggregation (RRA) meta-analysis to find robust differentially expressed genes (DEGs) between AD cases and healthy controls. We further investigated enriched pathways, related miRNAs and transcription factors of these DEGs. 2. Materials and Methods 2.1. Search Strategy A comprehensive search was performed through the National Center for Biotechnology Information (NCBI) Gene Expression Omnibus (GEO) datasets ([71]https://www.ncbi.nlm.nih.gov/geo/, accessed on 25 March 2021) to identify eligible data from inception to March 2021. The following key words were used: “Alzheimer”, “hippocampi”, “hippocampus”, “entorhinal”, “temporal”, “frontal” and “cerebellum”, then three filters including ‘Homo sapiens’, ‘Series’ and ‘Expression profiling by array’ were applied. In addition, references of all included studies