Introduction

Metabolic dysfunction-associated steatotic liver disease (MASLD) is the most common chronic liver disease worldwide, and its progression from simple steatosis through steatohepatitis (MASH), fibrosis, cirrhosis and hepatocellular carcinoma (HCC) is driven by cellular transitions that bulk transcriptomics cannot resolve. Single-cell, single-nucleus and spatial technologies have transformed this landscape, but the resulting human datasets are fragmented and technically heterogeneous, limiting the cross-study integration that a unified molecular map would require. Existing reviews synthesise biology rather than assessing whether the underlying datasets are actually accessible, annotated and suitable for an integrated spatially resolved human MASLD progression meta-atlas.

Aims & Methods

We aimed to systematically catalogue and grade the actionability of human liver single-cell and spatial transcriptomic datasets across the entire MASLD axis, to define a feasible backbone for an integrated meta-atlas. Following PRISMA 2020, we searched PubMed (frozen March 2026), supplemented by backward and forward citation searching and a targeted bioRxiv/medRxiv preprint search, and screened 2,357 unique records. Eligible studies generated de novo high-resolution transcriptomic data. For each, we extracted disease context, tissue source, modality, platform, yield, repository accessions and metadata, and graded integration-readiness across data accessibility, granularity, metadata completeness and technical harmonisability.

Results

Seventy-eight datasets met eligibility (35 core-metabolic data producers, 10 foundational healthy references, 33 secondary HCC references), published 2017–2026, with 83% appearing from 2022 onward. scRNA-seq predominated (47/78), followed by spatial methods (33), snRNA-seq (21) and multiome/ATAC (8); 19 datasets combined single-cell and spatial profiling, the configuration most useful for integration. Disease-axis coverage was markedly uneven: MASH (52/78) and HCC-containing tissue (51/78) dominated and 38 datasets included healthy reference tissue, whereas early MASL/simple steatosis was captured in only 11 datasets, leaving the earliest, most therapeutically tractable transition severely under-sampled. On integration-readiness, 42/78 datasets (54%) were openly reusable, 31 (40%) conditionally integrable (controlled access, author request, or pending publication) and 5 (6%) not obtainable; controlled-access deposition (26 datasets, concentrated among genotype-/eQTL-linked cohorts) and incomplete metadata were the principal barriers.

Conclusion

Human single-cell and spatial profiling of the MASLD axis is rich, accelerating, but uneven and fragmented, and only about half of it is openly integration-ready. The dominant obstacles to a unified molecular map are infrastructural rather than biological: sparse early-disease sampling, controlled-access deposition and inconsistent metadata. This audit provides a reproducible, curated roadmap and identifies spatially resolved profiling of early MASL and bridging fibrosis, with standardised metadata and open deposition, as the field's most actionable priorities for assembling a spatially resolved human MASLD meta-atlas.