What was compared
James Fernando, Peiyan Shen, Yi Yuan, Coralie Reynolds and Eric Reynolds at the Centre for Oral Health Research, Melbourne Dental School, ran a head-to-head instrument comparison rather than an agent trial (Journal of Dentistry, 2026). They created artificial subsurface enamel lesions in extracted human third molars, split each specimen into experimental and control halves, and exposed the experimental halves to casein phosphopeptide-amorphous calcium fluoride phosphate (CPP-ACFP) for eight days. They then measured percent remineralization (%R) three ways on the same samples: conventional laboratory micro-CT (Cµ-CT), synchrotron radiation computed tomography (SR-CT), and transverse microradiography (TMR), the destructive reference standard that requires sectioning the specimen. Repeated-measures ANOVA with Tukey post hoc comparisons and Bland-Altman analysis were used to compare the methods.
The size of the gap
The two tomography modalities agreed with each other but not with the reference. Mean %R was 20.77 plus or minus 7.12 percent by Cµ-CT and 20.55 plus or minus 6.27 percent by SR-CT, values not significantly different from one another; TMR returned 49.29 plus or minus 3.23 percent, and both XMT figures were significantly lower at p<0.00001. Put plainly, both tomography methods measured less than half the remineralization that microradiography did on the same lesions. Significant method effects also appeared for the underlying mineral-density variables (delta Zr at p<0.001 and delta Zd at p<0.02). The authors attribute part of the shortfall to imaging artifacts in both XMT modalities that compromise assessment of the enamel surface zone, the outer layer where much of the recovery is concentrated.
Why this matters for the record
Nearly every ex vivo enamel-remineralization number in the literature, including the agent studies this site tracks, carries an implicit instrument label, and this study puts a magnitude on how much that label matters. Percent remineralization is not a property of the agent alone: it is a property of the agent, the lesion model, and the measuring instrument. The finding also bounds how yesterday’s white-spot result should be read: the P11-4 plus CPP-ACP pH-cycling study reported its outcome under its own measurement setup, and its 20 to 30 percent margins should not be lined up against %R values produced by a different instrument as if they were the same ruler. The authors’ own conclusion is calibrated: TMR remains the more sensitive and accurate reference for ex vivo remineralization work, particularly at the surface zone, while XMT still has a role where its non-destructive nature matters, namely longitudinal tracking of mineral changes in the same specimen.
Boundary of the result
This is one in vitro comparison: one lesion model, one agent, one eight-day treatment window, extracted molars. It does not test in situ or clinical remineralization, and it does not show that tomography is wrong about mineral content in general; the authors explicitly position XMT as the appropriate tool for non-destructive or longitudinal mineral measurement, just not as a drop-in replacement for TMR in %R endpoint analysis. It also does not rank CPP-ACFP against any other agent, since no second agent was tested.
Where we differ from the coverage
Coverage of remineralization products routinely quotes percent-remineralization figures without naming the instrument that produced them, which lets numbers measured by different methods compete as if they were commensurable. This study is a direct demonstration that they are not: on identical lesions, two respectable instruments disagreed with the reference by more than a factor of two. Any future efficacy claim on this record should carry its measurement method alongside the number.
Provenance: grounded in the complete structured abstract of the CC-BY article (Journal of Dentistry 2026, DOI 10.1016/j.jdent.2026.107111, PMID 42880482), verified verbatim against two independent primary indexes (Europe PMC and OpenAlex); the publisher full text was bot-blocked from this host, so every number above is traceable to that abstract and to no other source. Method and sourcing standard at /method/.