Unexpected Bone Resorption in Mentum Induced by the Soft-Tissue Filler Hyaluronic Acid: A Preliminary Retrospective Cohort Study of Asian Patients

Xiaoshuang Guo and colleagues recently reported that hyaluronic acid could induce resorption in the mandibular bone and that resorption was positively correlated with the injection volume per time.1 The authors concluded their article with the worrisome warning that patients should be fully informed preoperatively about massive resorption from large-volume injection of soft-tissue fillers.1 Undoubtedly, if confirmed, this might be a real concern in cosmetic medicine, raising doubts about filler safety. Nevertheless, while reading the article, many criticisms were raised from our research group and expert colleagues in the maxillofacial surgery unit. In addition to dismissing any covariate estimation and correlation plotting, the authors claim they found a significant difference between non–hyaluronic acid–treated controls and hyaluronic acid–treated subjects. However, they erroneously compared percentages, not absolute measures of mandibular thickness. Using percentages, they compared two differently distributed populations: a nonparametric (controls) and a parametric (hyaluronic acid–treated) distribution (Anderson-Darling test: controls, p = 0.7248; treated, p = 2.767 × 10−5). While comparison test results were highly significant (Wilcoxon Mann Whitney U test, p = 4.4887 × 10−8), the Pearson test for ungrouped data resulted in p = 0.309 (two-sided). The comparison of chin thickness in a one-way analysis of variance with a Tukey post hoc test gave F = 0.807 and p = 0.704; in an unpaired t test, p = 0.67542; and in a Wilcoxon test, p = 0.79667, thereby rejecting the high significance reported by the authors using their same data. In this respect, our conclusion is that the evidence reported by the authors contains bias, which prevents readers from trusting their published evidence as sound and reliable. A possible reason is the incorrect collection of data, spoiled by outliers as indicated by the Cook distance test, affecting regression (p = 0.309). Figure, Supplemental Digital Content 1, below, shows the linear regression for hyaluronic acid–treated and control subjects (p = 0.309, estimated intercept = 88.507 ± 5.228 SD). [See Figure, Supplemental Digital Content 1, which shows (above, left) data distribution in a Pearson correlation test showing two different distributed populations of data; data are averages of percent proportion on the left side, taken from supplementary results provided with the article. (Above, right) The Cook distance test shows a modality to identify points that negatively affect regression evaluation. The measurement is a combination of each observation’s leverage and residual values; the higher the leverage and residuals, the higher the Cook distance. (Below) Regression plotting on data used in above, left panel, https://links.lww.com/PRS/E865.] The reports by Guo et al. showed further flaws, such as that controls may not represent an adequately matched comparison group, as subjects in this cohort were a merging of individuals seeking out contour surgery. This introduces further bias, as the severity of contour deficit was likely greater than that in a group of subjects receiving hyaluronic acid to the chin. Moreover, the authors did not further specify the method used; the same quality-of-life questionnaire results as a nonstandardized, “homemade” interview panel, despite the existence of qualified, approved methods in the field.1–4 The reported mechanism may be caused by pressure-related events, as reported with silicone chin implants.5 Furthermore, the balance between osteoclasts and osteoblasts, impaired by hyaluronic acid fillers, lacks evidence. The low-molecular-weight hyaluronic acid (< 8 kDa), which interestingly is used in osteoarthritis,6 has been reported to enhance both osteoclast formation and function in vitro,7 and yet hyaluronic acid less than 8 kDa is not a component of hyaluronic acid fillers. On the other hand, high-molecular-weight hyaluronic acid has been reported to suppress osteoclast formation in animal experimental models.7 In conclusion, both study design and data managing were not robust enough, in our opinion, to assess a causative link between hyaluronic acid and bone resorption. DISCLOSURE The authors have no financial or conflicts of interest to report. Salvatore Chirumbolo, M.Sci.Department of Neurosciences, Biomedicine and Movement Sciences Riccardo Nocini, M.D.Department of Surgery, Dentistry, Paediatrics and GynaecologyUnit of Otorhinolaryngology Dario Bertossi, M.D.Department of Surgery, Dentistry, Paediatrics and GynaecologyUnit of Maxillofacial SurgeryUniversity of VeronaVerona, Italy