Platelet-rich fibrin and CBCT-derived radiographic attenuation in post-extraction sockets: a secondary analysis of a split-mouth randomized controlled trial

Background Tooth extraction is followed by alveolar remodelling that may complicate later implant placement. Platelet-rich fibrin (PRF) is widely used to support socket healing, but quantitative imaging comparisons within the same patient remain scarce, and cone-beam computed tomography (CBCT) grey values are difficult to interpret because they are not calibrated in the way medical computed tomography is. This report is an exploratory secondary analysis of a split-mouth randomized controlled trial whose dimensional outcomes have been published previously. It addresses a question the earlier report did not: whether radiographic attenuation follows a different trajectory in PRF-treated and control sockets, and how far that conclusion survives alternative analytical choices. Methods The parent trial randomized 36 post-extraction premolar sockets from 10 patients to PRF ( n = 19) or spontaneous healing controls ( n = 17), with every patient contributing sockets to both conditions. CBCT scans were acquired at 30 and 120 days under identical parameters, and attenuation was recorded at two intra-alveolar locations, points A and B, defined as in the previous report. No new participants, interventions, CBCT acquisitions, or radiographic measurements were added. Linear mixed-effects models with a socket-level random intercept were fitted separately for each location using restricted maximum likelihood and Kenward–Roger degrees of freedom. The Group × Time interaction was the central longitudinal estimand; four PRF versus control contrasts were adjusted using the Holm procedure. Because the trial randomized within patients, dependence between sockets from the same patient was examined through a patient-aggregate analysis, cluster-robust CR2 standard errors, a cluster bootstrap, and a leave-one-patient-out analysis. Results At 30 days the estimated difference was 34.42 HU-equivalent units at point A (95% CI − 31.69 to 100.53; Holm-adjusted p = 0.558) and 35.15 at point B (95% CI − 29.57 to 99.87; Holm-adjusted p = 0.558). At 120 days PRF-treated sockets showed higher values at both locations: 120.91 units at point A (95% CI 54.80 to 187.02; Holm-adjusted p = 0.0023) and 119.52 at point B (95% CI 54.80 to 184.24; Holm-adjusted p = 0.0023). The Group × Time interaction was 86.49 units at point A (95% CI 46.47 to 126.51) and 84.37 at point B (95% CI 43.39 to 125.35). Socket-weighted and patient-based analyses pointed in the same direction at 120 days and for the interaction. At 30 days the two estimators diverged and the patient-averaged bootstrap interval included zero. No adverse events were observed during follow-up. Conclusions CBCT-derived radiographic attenuation increased more between 30 and 120 days in PRF-treated sockets than in control sockets, and this pattern was directionally consistent across socket-weighted and patient-based analyses. Evidence at 30 days was estimator-dependent and remains inconclusive. Because the measurement was not calibrated against medical computed tomography or histology, these findings do not establish differences in bone quality, histological maturation, or implant-related outcomes. Inference rests on ten patients and should be read as exploratory. Trial registration ClinicalTrials.gov, NCT06900036, first submitted on 15 March 2025 and first posted on 28 March 2025. Registration in this registry was retrospective. https://clinicaltrials.gov/study/NCT06900036 .