Gliomas are cancers of the brain with a five-year survival rate of 22%, with current standard of care unable to prevent disease progression. The leading edge of the tumor is a key architecture involved in both recurrence and progression, but spatially resolved characterisation of this region is lacking. Increasing the dimensionality at which we interrogate glioma – into their native 3D space and across omics platforms – will improve our understanding of the underlying mechanisms causing biological dysfunction. We have developed a spatial multi-omic integration pipeline, SMINT, to integrate transcriptomics and metabolomics at the leading edge across serial sections of IDH-mutant glioma. We found that nuclei-only segmentation, while containing only 40% of segmented cell transcripts, enables accurate cell type annotation across different tissues and platforms, but cannot account for multinucleated cells. By combining spatial transcriptomics and metabolomics, our integrative analysis demonstrated tissue regions that are transcriptionally distinct with an associated unique metabolic landscape, identifying increased OPC-like tumor cells that may drive invasion. By interpolating between sections via Kriging, we can extend our pipeline to develop a 3D model of disease architecture (SMINT-3D). Validated with mouse brain and human glioma using MERFISH, Xenium and CosMx, Kriging displays promising accuracy (60-70%) in predicting the spatial changes to cellular neighborhoods across serial sections spaced over 200 microns apart, offering a cost-effective approach to reconstructing 3D tissue. There is significant need to improve our understanding of the leading edge in glioma, which will be enhanced by innovative spatial multi-omic strategies. Investigating glioma by combining our SMINT pipeline with Kriging interpolation will improve our ability to identify spatial drivers of tumorigenesis in 3D, and develop effective treatments for patients.