Background: Iron trafficking and accumulation has been associated with Alzheimer's disease (AD) pathogenesis. However, the role of iron dyshomeostasis in early disease stages is uncertain. Currently, gene expression changes indicative of iron dyshomeostasis are not well characterised, making it difficult to explore these in existing datasets. Results: We identified sets of genes predicted to contain Iron Responsive Elements (IREs), and used these to explore iron dyshomeostasis responses in transcript datasets involving (1) cultured cells under iron overload and deficiency treatments, (2) post-mortem brain tissues from AD and other neuropathologies, (3) 5XFAD transgenic mice modelling AD pathologies, and (4) a zebrafish knock-in model of early-onset, familial AD (fAD). IRE gene sets were sufficiently sensitive to distinguish not only between iron overload and deficiency in cultured cells, but also between AD and other pathological brain conditions. Notably, we see changes in 3' IRE transcript abundance as amongst the earliest observable in zebrafish fAD-like brains and preceding other AD-typical pathologies such as inflammatory changes. Unexpectedly, while some 3' IRE transcripts show significantly increased stability under iron deficiency in line with current assumptions, many such transcripts instead show decreased stability, indicating that this is not a generalizable paradigm. Conclusions: Our results reveal iron dyshomeostasis as a likely early driver of fAD and as able to distinguish AD from other brain pathologies. Our work demonstrates how differences in the stability of IRE-containing transcripts can be used to explore and compare iron dyshomeostasis responses in different species, tissues, and conditions.