The diagnostic problem
Whether an inflamed pulp is treated conservatively or extirpated hinges on a diagnosis, reversible versus irreversible pulpitis, that rests heavily on subjective pain assessment and sensibility testing. Zaky and colleagues state the known weakness plainly: these clinical proxies may not reflect the histopathological and molecular state of the tissue, and the discordance can lead to overtreatment of teeth that might have responded to pulp-preserving care. Their pilot applies spatial transcriptomics to ask what the tissue actually looks like beneath the clinical labels.
What the pilot found
Using Visium CytAssist spatial transcriptomics on four human dental pulp samples, with cell deconvolution, differential gene expression, and pathway scoring, the team profiled healthy pulp and clinically diagnosed reversible and irreversible pulpitis, attending to coronal regions adjacent to carious lesions. The headline observation is a single discordant case: one sample meeting clinical criteria for symptomatic irreversible pulpitis exhibited spatial features more similar to the reversible sample, including a comparable immune-to-fibroblast ratio and similar activation patterns of TLR4 and neuroinflammation-related pathways. Across the irreversible samples, genes involved in immune signalling, cell migration, and tissue remodeling were differentially expressed. Increased PTN and CXCL14, together with lower expression of ENG (CD105), SELE, COL4A1, CXCL1, and CXCL13, emerged as candidate markers associated with a reversible-like molecular profile in this dataset.
How far the evidence goes
This is T3 at most on our ladder at /method/, and a narrow instance of it: four samples, no validation cohort, candidate markers rather than a tested classifier. One discordant case is a hypothesis about misdiagnosis, not a measured misdiagnosis rate. Nothing here yet changes a treatment decision. The value is directional. If a molecular readout of pulp state ever becomes practical, it sits exactly at the decision point that governs the pulp-dentin-repair programme we track at /programs/pulp-dentin-repair/: every vital pulp therapy and regenerative endodontic procedure depends on correctly identifying tissue that can still recover. A diagnosis that is wrong in either direction either sacrifices recoverable pulp or attempts regeneration in tissue that cannot support it. The immune-to-fibroblast ratio and the PTN/CXCL14 signature are the kind of concrete, checkable candidates a follow-up cohort study could confirm or kill.
Where we differ from the coverage
Spatial transcriptomics papers attract write-ups about tests that will “save teeth from root canals”. Four samples cannot support that claim, and the authors do not make it: the paper itself is framed as exploratory, and its markers are explicitly candidates. The accurate summary is that the first spatial maps of inflamed human pulp exist, they already show one case where the molecular state and the clinical label diverge, and the field now has named markers to validate. The next dated entry in the /ledger/ for this thread should be a validation cohort, not a product.
Provenance: every claim above traces to the published abstract of Zaky et al., Journal of Translational Medicine (2026), per our method at /method/.