State of the field, fifth pass

Eight days after the fourth pass, the ladder is unchanged: one program holds adult safety data and nothing more, the repair wing gained its first animal result in the peptide GSK-3β lane, and an independent review has now drawn the field’s central boundary from outside the developer orbit.

Written 2026-11-02. This is the fifth pass; it supersedes the fourth pass of 2026-10-25, which stays published and unchanged in the archive. The evidence window is the eight days since the fourth pass, plus items that entered the record after that pass was written. The window’s primary results sit on the pulp and dentin repair wing: a rat pulp injury study of the Wnt agonist peptide L803-mts, the first in vivo entry in the peptide GSK-3β lane this site tracks, and a head-to-head review of the two dental stem cell types that finds no winner. On the whole-tooth side, the window brought developmental genetics, a 24-patient cilia study linking C2CD3 to tooth number, shape and eruption timing, a whole-exome cohort in tooth agenesis with neurodevelopmental disorders that named candidates and proved none, and a study separating two enamel-defect genes on the eruption-failure axis. One review matters beyond its page count: an Egyptian dental-school group surveyed the anti-USAG-1 literature independently and concluded, in its own words, that regeneration after acquired tooth loss should be viewed as a translational hypothesis rather than an experimentally validated therapeutic outcome. That is this record’s standing line, now shared by a group with no stake in the program. None of it is human efficacy data for a third dentition, and no tier moved.

Program bases were re-checked against the research pool, the trial registries, and the sponsor’s disclosure pages. The standing re-verification gap is unchanged and is restated rather than smoothed over: the jRCT registry host still blocks this pipeline host, so the lead program’s registry state is carried from the 2026-10-05 verification, 28 days old and inside the 120-day window.

Tier ladder

• Tier 5: Replicated human efficacy. Held by: nothing.
• Tier 4: Human efficacy or safety data, single program. Held by: anti-USAG-1 / TRG-035.
• Tier 3: Small or single-center human data, narrow endpoint. Held by: pulp and dentin repair.
• Tier 2: Animal result, no human protocol proposed. Held by: bioengineered tooth germ, dental epithelial organoids, tooth-root organoids.
• Tier 1: In vitro only, or mechanism without a tissue result. Held by: nothing at this pass.

Per-program assessment

Anti-USAG-1 / TRG-035 (Toregem BioPharma, Kyoto University): tier 4, held

The human program has still answered only one question: whether a single ascending dose is tolerated in adults. Registry re-verification was attempted again this pass and failed again: the jRCT host did not respond to this pipeline host for a second consecutive pass. The stated registry state therefore still rests on the third pass’s verification of 2026-10-05: recruitment status Complete, a target of 30 healthy adult males aged 30 to 65, a single intravenous dose in five ascending steps, a primary endpoint of safety, and secondary endpoints of pharmacokinetics and anti-TRG-035 antibody frequency (kitano-2024-phase1). No outcome data is posted. A check of clinicaltrials.gov this pass again found no TRG-035 record there, consistent with the US program still being pre-IND (toregem-2025-fda-preind). This is the second pass carrying the registry basis at one remove; if the host block persists, the basis will age out of the 120-day window and the tier itself will need a re-verification source or an explicit stale marking.

On the sponsor’s side, the newsroom has produced nothing since the fourth pass recorded the 20 October 2026 booth visit; the latest disclosure that touches the science remains the 17 August 2026 completion of the PMDA clinical trial notification investigation for the Phase IIa trial (toregem-2026-pmda-ctn). No enrollment, no dosing, and no results for Phase IIa have been disclosed, and the sponsor’s own next disclosed step remains institutional ethics review at participating sites.

The window’s one publication on the program is a review, and it earns its place in this assessment for who wrote it rather than what it found. Tawfik and colleagues at MSA University in Giza, with no connection to the Kyoto or Toregem orbit, surveyed the anti-USAG-1 evidence in Gene (30 October 2026) and restated the record this site holds: in mice and ferrets, anti-USAG-1 treatment promotes additional or regenerated tooth-like structures in congenital-agenesis models; the registered human work is a Phase I safety and pharmacokinetics study in Japan; and evidence for regeneration after acquired tooth loss is limited, so that application should be viewed as a translational hypothesis rather than an experimentally validated outcome (covered in the analysis stream). A review moves no tier. What it changes is who draws the line: an independent, non-originator group now states the same boundary in the peer-reviewed record.

Tier 4 stands unchanged: human safety data exists for a single program, carried on a registry entry last verified 28 days ago and stated as such, and nothing more. No peer-reviewed human result set is public and no efficacy endpoint has been tested in anyone.

What would change this assessment: an efficacy endpoint in the intended population, children with congenital anodontia, reported with its statistical plan rather than by press release; or a dated human image showing a tooth in a site that was radiographically empty before treatment. A re-verified registry entry would also restore the basis to fully current status.

Bioengineered tooth germ: tier 2, held

The window’s addition to this wing is developmental patterning, not capability. A PNAS study of 24 patients carrying rare or novel variants in C2CD3, a primary-cilia gene, catalogs dental defects in tooth number, shape and eruption timing, and uses hedgehog-signaling data to offer a testable, explicitly hedged hypothesis for the evolutionary loss of teeth in birds (covered in the analysis stream). It sharpens the developmental map of how tooth number is set; it contains no reconstitution result, no transplant, and nothing that closes the program’s central gaps. A companion genetics item this window, an Italian whole-exome cohort of ten patients with tooth agenesis and neurodevelopmental disorders, prioritized 30 variants in 28 genes and concluded that none is established as the cause of the dental phenotype (covered in the analysis stream); a study of FAM20A and FAM83H enamel-defect families separates two genes on the eruption-failure axis (covered in the analysis stream). All three are background for the patient population and the developmental circuitry, not program evidence.

The mammal basis is untouched: a transplanted bioengineered tooth germ erupted, occluded, and responded to mechanical and pain stimuli in an adult mouse (2009, ikeda-2009); chemically defined culture supports mouse tooth reconstitution with BMP-driven enamel induction (zhang-2024-chemically-defined-tooth-reconstitution); decellularized porcine tooth-bud scaffolds seeded with dental and endothelial cells form tooth-like tissues, including periodontal-ligament-like tissue with Sharpey’s fibers, in adult minipigs (zhang-yelick-2025-decellularized-tooth-bud); and the qualifying results stand, lingual-buccal positional memory in dissociated and reaggregated mouse dental mesenchyme (kim-2025-dental-mesenchyme-position), the secretome not replacing living mesenchyme for tooth induction in mice (birjandi-sharpe-2025-tooth-germ-secretome, with the February 2026 publisher correction, birjandi-sharpe-2026-secretome-correction), and the marker-defined human tooth-germ progenitor that forms dentin, pulp-like tissue and bone in graft and rodent models and forms no tooth (Yu et al., 2026, doi:10.1016/j.stemcr.2026.103121). No new mammal whole-tooth result appeared this window, and no human protocol has been proposed.

What would change this assessment: independent replication of tooth formation in a large animal by a group with no commercial stake, a credible human cell-sourcing plan, or reconstitution of a tooth germ from the marker-defined human progenitor with developmental control demonstrated.

Dental epithelial organoids: tier 2, held

No news this pass. The basis stands: human dental epithelial organoids with a stemness phenotype and in vivo assessment (hemeryck-2022), supported by a soluble Notch agonist maturing human ameloblast organoids to enamel-like mineral under the mouse kidney capsule (patni-2026-ameloblast-organoid), PITX2-positive embryonic oral epithelium induced from human pluripotent stem cells (nakashima-2025-oral-epithelium), and a biomimetic elastin-like matrix nucleating enamel-like mineral on acid-etched human enamel and dentin ex vivo (hasan-2025-biomimetic-enamel-restoration). The program sits at the organoid/model boundary, not at a replacement tooth.

What would change this assessment: organoid-derived enamel tissue of native thickness and orientation, integrated in a living jaw.

Tooth-root organoids: tier 2, held

No news this pass. The basis stands: self-assembled tooth root organoids from postnatal human dental stem cells organize dentin-, pulp-, cementum-, and ligament-like tissue with in vivo assessment (calabrese-2024), and the 2026 scaling result still constrains the route to a larger construct: smaller root-organoid constructs lose the layered mineral pattern, pointing to a minimum cell mass for proper patterning (calabrese-2026-root-organoid-scaling). Eruption, crown formation, innervation, and periodontal integration remain unresolved, and nothing has been placed in an animal jaw.

What would change this assessment: eruption, crown formation, innervation, and periodontal integration of an organoid-derived construct in an animal jaw.

Pulp and dentin repair: tier 3, held

The tier basis stands and the window’s additions qualify it, each short of the ladder’s human bar in a different way. The one new primary result is the first in vivo entry in the peptide GSK-3β lane this site tracks: Rajasekar and colleagues, the Hong Kong group behind this site’s October analysis of their earlier cell-dish work, report in the International Endodontic Journal (published online 26 October 2026) that L803-mts, a myristoylated GSK-3β-inhibitory peptide used as a Wnt agonist, promoted reparative dentinogenesis in a rat pulp injury model at mineral density comparable to mineral trioxide aggregate, 1.17 versus 1.32 gHA/cm3, with significantly lower mineral volume, 0.03 versus 0.06 mm3 (covered in the analysis stream). In vitro under an inflammatory challenge the peptide increased mineralization and shifted macrophages toward an anti-inflammatory phenotype. Three boundaries hold even at the level of the author-supplied abstract, the paywalled full text not being accessible: the parity claim concerns mineral density only, with half the mineral volume of the reference material; the model is a rat pulp injury, not carious pulpitis in a patient; and the Wnt attribution is the authors’ interpretation, with no rescue or loss-of-function experiment in the abstract-level record. It is a bounded proof of progress in the same mechanistic family as tideglusib, not a parity claim with the clinical standard; tier basis is unchanged (neves-2017).

Second, a Japanese narrative review comparing SHEDs and DPSCs for cell-based pulp therapy finds no winner: animal readouts split, human histology covers one re-traumatized tooth, and no SHED cell-based vital pulp therapy study exists (covered in the analysis stream). It is a useful map of a crowded lane, and it underscores the standing qualification: the human histology base for cell-based repair remains a handful of single cases. Third, an in-vitro study in the Journal of Dental Sciences finds that acidic pH, the condition an infected canal actually presents, cuts apical-papilla stem cell proliferation by about 57 percent through PI3K/AKT/mTORC1 suppression, without apoptosis: the arrest is reversible, and the group’s self-assembling buffering peptide hydrogel, which pulls medium back toward neutral, restores proliferation about 3.5-fold in culture (covered in the analysis stream): mechanism-level support for scaffold design, no tissue result, and a reminder that alkaline dressing chemistry can be as hostile as acid, with pH 10.4 outright cytotoxic in this assay. A meta-analysis of 21 animal studies of extracellular-vesicle therapy in pulp regeneration entered the record this window as well, pooling to large effect sizes on mineralization and angiogenesis with 82 to 91 percent heterogeneity and an unclear risk of bias in every included study (covered in the analysis stream); the underlying paper has been in the research pool since July, and the synthesis qualifies the crowded animal-lane evidence rather than strengthening it. A preprint reanalysis of a 25-formulation electrospun-antibiotic-fiber campaign, finally, finds that water-soluble carriers may dump their payload within hours against a one-to-four-week dressing window (covered in the analysis stream); a preprint and a carrier study, it constrains delivery engineering and nothing more.

None of this moves the tier. The tier still rests on the randomized trial of autologous deciduous-tooth pulp stem cell implantation in immature permanent incisors: 26 patients evaluated with 24-month imaging, regenerated pulp tissue with vessels and nerves (xuan-2018), supported by a 2026 small randomized trial of platelet-rich fibrin versus mineral trioxide aggregate for direct pulp capping in 20 adult molars (asthana-2026-prf-mta-pulp-capping), and qualified by the best available human histology after a regenerative endodontic procedure: dentin-like tissue without an organized odontoblast layer, sparse neurovascular ingrowth, apical osteoid (yang-2026-rep-histology-human). What is regrown in people remains repair-adjacent tissue, not a rebuilt pulp-dentin complex, let alone anything beyond the pulp space.

What would change this assessment: controlled human evidence that repair extends beyond the pulp space, true enamel regrowth rather than surface remineralization, or human histology showing an organized odontoblast layer and organized neurovascular ingrowth after a regenerative procedure.

Myth vs record

1. Claim: TRG-035 Phase I completed “with no serious adverse events reported,” and first approval around 2030 is a realistic timeline. Verdict: the safety characterization is unverifiable as stated and the timeline is a projection. The sponsor disclosed completion of Phase I dosing on 19 May 2026 (toregem-2026-financing), but no peer-reviewed result set is public, the registry entry carries no outcome data, and “no serious adverse events” is a claim about data no one outside the sponsor can inspect. A dated approval timeline, however plausible its author finds it, is not evidence and moves nothing.
2. Claim: the TRG-035 Phase I participants are adults missing at least one molar. Verdict: unsupported. The registry record this site holds describes 30 healthy adult males aged 30 to 65 receiving a single intravenous dose in five ascending steps, with safety and pharmacokinetic endpoints (kitano-2024-phase1). A missing-molar enrollment criterion does not appear in that record.
3. Claim: the tooth-regrowth drug will also let adults regrow teeth lost to decay or injury. Verdict: overstated. The intended population is children with congenital anodontia; the congenital-agenesis animal models are the evidence base; and the 30 October 2026 review by Tawfik et al. states that application to acquired tooth loss should be viewed as a translational hypothesis rather than an experimentally validated outcome. Sponsor-facing materials that present anti-USAG-1 as biological tooth replacement for people who have lost teeth sit on the hypothesis side of that line.

What would change the field-level assessment

1. A dated human image showing a tooth in a site that was radiographically empty before treatment.
2. An efficacy endpoint, in children, reported with its statistical plan rather than by press release.
3. Independent replication of tooth formation in a large animal by a group with no commercial stake.

Corrections applied this pass

• None logged since the fourth pass; the inbox is empty. The previously logged duplicate-analysis retirement remains in force. One deferred item carried without change: the withdrawal decision on the duplicate regenerative-endodontics-versus-apexification page remains open, awaiting a deliberate on-record human decision.

Cited here

• kitano-2024-phase1: first-in-human Phase I announcement and registry record, single ascending dose, adults; registry last verified 2026-10-05, re-verification attempted and failed this pass.
• toregem-2026-pmda-ctn: PMDA CTN investigation completed for Phase IIa, 17 August 2026.
• toregem-2025-fda-preind: sponsor disclosure of FDA response to pre-IND meetings, 27 November 2025.
• toregem-2026-financing: Phase I complete and Phase II preparation disclosure, 19 May 2026.
• murashima-suginami-2021: anti-USAG-1 rescues tooth agenesis in mice and ferrets.
• ikeda-2009: bioengineered tooth germ functional in an adult mouse.
• zhang-2024-chemically-defined-tooth-reconstitution: chemically defined mouse tooth reconstitution.
• zhang-yelick-2025-decellularized-tooth-bud: decellularized scaffolds in minipigs.
• kim-2025-dental-mesenchyme-position: positional memory in dental mesenchyme.
• birjandi-sharpe-2025-tooth-germ-secretome: secretome does not replace living mesenchyme.
• birjandi-sharpe-2026-secretome-correction: publisher correction to that study.
• hemeryck-2022: human dental epithelial organoids with in vivo assessment.
• patni-2026-ameloblast-organoid: Notch-agonist-matured ameloblast organoids, enamel-like mineral in mice.
• nakashima-2025-oral-epithelium: PITX2-positive oral epithelium from human pluripotent stem cells.
• hasan-2025-biomimetic-enamel-restoration: biomimetic enamel-like mineral ex vivo.
• calabrese-2024: self-assembled tooth root organoids from human dental stem cells.
• calabrese-2026-root-organoid-scaling: size scaling disrupts the layered mineral pattern.
• xuan-2018: randomized autologous pulp stem cell trial, 26 patients, 24-month imaging.
• asthana-2026-prf-mta-pulp-capping: PRF versus MTA direct pulp capping randomized trial.
• yang-2026-rep-histology-human: human histology after a regenerative endodontic procedure.
• neves-2017: tideglusib promotes reparative dentin in mice.
• Tawfik et al. 2026, Gene 150472, doi:10.1016/j.gene.2026.150472: independent anti-USAG-1 review, acquired-loss regeneration named a translational hypothesis; analysis stream only, not yet in the keyed bibliography.
• Rajasekar et al. 2026, International Endodontic Journal, doi:10.1111/iej.70298: L803-mts reparative dentinogenesis in a rat pulp injury model; analysis stream only.
• Baldeon-Gutierrez et al. 2026, Journal of Clinical Medicine 15(20):8162, doi:10.3390/jcm15208162: SHED versus DPSC pulp repair review, no winner; analysis stream only.
• Zhang et al. 2026, Journal of Dental Sciences 21(4):2092-2099, doi:10.1016/j.jds.2025.07.3469: acidic pH reversibly arrests apical-papilla stem cell proliferation without apoptosis, buffering peptide hydrogel rescue, in vitro; analysis stream only.
• Kurmanalina et al. 2026, Science Progress 109(1), doi:10.1177/00368504261433133: extracellular-vesicle therapy meta-analysis across 21 animal studies; analysis stream only.
• Kantaputra et al. 2026, PNAS, doi:10.1073/pnas.2603057123: C2CD3 cilia variants and tooth patterning across vertebrates; analysis stream only.
• Yu et al. 2026, Stem Cell Reports, doi:10.1016/j.stemcr.2026.103121: SLITRK6-positive human tooth-germ progenitor, dentin, pulp-like tissue and bone in graft and rodent models, no whole tooth; analysis stream only.