Application of molecular targeted therapies for esophageal adenocarcinoma (EAC) has been limited by a lack of druggable oncogenic drivers. We propose that synthetic-lethal interactions may provide new opportunities for targeted therapies in EAC. We have taken an integrated multi-omics approach incorporating Perturb-Seq and in vivo tumourigenesis assays to perform high-throughput characterisation of >70 high-confidence EAC driver genes, followed by genome-wide CRISPR-Cas9 knockout screens in isogenic models of EAC tumourigenesis to identify driver specific gene dependencies. Our overall goals are to (i) enhance our understanding of EAC tumourigenesis, (ii) identify potential opportunities for therapeutic interventions targeting EAC drivers via synthetic lethal-like approaches, and (iii) reduce the complexity of genetic heterogeneity by categorising EAC drivers with similar phenotypic outcomes. Through our approach we have identified complex crosstalk between the tumour suppressor SMAD4 and regulation of mTOR signalling, with specific downstream effects on translational reprogramming in EAC. Mutation or loss of SMAD4 occurs in up to 20% of EAC, but not pre-malignant tissue (Barrett’s metaplasia), and is sufficient to promote transformation of pre-malignant cells in our in vivo tumourigenesis model. In this model, xenotransplanted SMAD4-deficient Barrett’s metaplasia cells formed invasive, metastatic tumours after a period of latency. SMAD4 deficient cells had downregulated expression of 4E-BP1, which inhibits EIF4E, the cap-dependent translation initiation factor. This was accompanied by increased mTOR activity, including phosphorylation and inactivation of 4EBP1. Moreover, we found that SMAD4-deficient cells preferentially upregulate cap-dependent translation at the expense of IRES mediated translation. Furthermore, perturbation of additional negative regulators of mTOR signalling in combination with SMAD4 knockout exacerbated these effects and accelerated tumourigenesis in vivo. We have extended these findings to a model of Barrett’s esophagus patient-derived organoids (PDOs) and observed increased proliferative potential of our genetically modified PDOs. Finally, analysing gene ontologies for differentially expressed genes from Perturb-seq revealed that driver-dependent transcriptional changes can be categorized into a smaller number of functional pathways allowing us to potentially consider groups of drivers as functional units. This work advances our understanding of EAC tumourigenesis, provides new mechanistic insights into SMAD4-driven transformation as well as novel potential therapeutic avenues for SMAD4-deficient EAC.