Material and methods

This cross-sectional multicenter study was approved by the Research Ethics Committees of the University Hospital Antônio Pedro from the Fluminense Federal University (33791314.3.0000.5243), School of Dentistry of Ribeirão Preto from the University of São Paulo (50765715.3.0000.5419), Positivo University (80846317.8.0000.0093), and the University of Pittsburgh (no. 12080056). Written informed consent was obtained from all participants or their legal guardians.

Clinical and radiographic records from Brazilian subjects were assessed. Orthodontic patients from private practice and graduate orthodontic clinics in 3 cities (Rio de Janeiro [mean age, 24.7 ± 10.7 years; 124 male, 210 female], Manaus [mean age, 26.5 ± 11.0 years; 51 male, 79 female], and Ribeirão Preto [mean age, 14.9 ± 7.0 years; 65 male, 68 female]) located in different regions of Brazil were selected. In addition, a sample of patients presenting severe skeletal sagittal and/or vertical disharmonies, which required orthognathic surgery, were recruited from another center in the city of Curitiba [mean age, 29.3 ± 9.1 years; 33 male, 65 female] to test if the studied genetic variants are also involved in more severe phenotypes. Sample characteristics, location setting, and ethnic composition of each city were already described in a previous study. Patients who received previous orthodontic treatment, who had concomitant systemic medical conditions, craniofacial congenital or syndromic anomalies, a substantial number of teeth missing, or suffered facial trauma were excluded.

Digital cephalometric tracings of preorthodontic or preorthognathic lateral cephalograms, with the mandible in centric occlusion positioning, were performed by previously trained orthodontists using Dolphin 3D Imaging software (version 8.0; Dolphin Imaging and Management Solutions, Chatsworth, Calif). The Steiner’s ANB and Ricketts’ NBa-PtGn angles were measured to determine sagittal (skeletal malocclusion) and vertical (facial type) skeletal jaw relations. Subjects were classified as Class I (0°-4°), Class II (>4°), or Class III (<0°) and as mesofacial (87°-93°), dolichofacial (<87°), or brachyfacial (>93°) according to the ANB and NBa-PtGn angles, respectively. In addition, angular and linear measurements were obtained in both sagittal and vertical dimensions ( Fig 1 ; Table I ). Linear measurements were only assessed on radiographs with a ruler included in the image that enabled the corresponding size calibration in the Dolphin 3D Imaging software.

Reference points and lines used to perform analyses. Angular measurements (°): NBa-PtGn, SN-GoGn, y-axis, SNA, SNB, and ANB. Linear measurements (mm): Co-Go, S-Go, ANS-Me, N-Me, S-N, Co-Gn, Go-Pg, and Ptm′-A′.

The SNPs were chosen on the basis of their minor allele frequency >10%. Those selected in GHR (rs2910875, rs1509460, and rs2973015) are located in loci previously reported as candidate regions for mandible-related phenotypes. ^{, , , , ,} SNPs in IGF2R (rs2277071, rs6909681, and rs6920141) were selected by their functional implications or location according to the SNP database ( www.ncbi.nlm.nih.gov/SNP ). The characteristics of the studied SNPs are shown in Table II .

Genomic DNA was extracted and purified from buccal mucosa cells from saliva as previously reported. Genotyping was blindly performed by polymerase chain reactions (PCR) using end-point analysis and TaqMan assay on a real-time PCR system (Applied Biosystems Prism QuantStudio 6 Flex PCR System, Thermo Fisher Scientific, Foster City, Calif) according to an established protocol.

Results

The distribution of genotypes followed Hardy-Weinberg equilibrium (data not shown). There were no associations between the skeletal malocclusion or facial type and the genotype or allele frequencies for any SNP assessed in GHR and IGF2R in the total sample. Associations were detected when specific cephalometric measurements were assessed ( P <8.333 × 10 ⁻³ ) ( Table III ). Rare homozygotes for the GHR rs2973015 marker showed increased measures for the lower anterior facial height (ANS-Me) and mandibular sagittal lengths (Co-Gn and Go-Pg). In contrast, common homozygotes for the IGF2R rs6920141 presented reduced measures for these dimensions (ANS-Me and Go-Pg). In addition, the less common homozygotes for the IGF2R rs2277071 had reduced maxillary length (Ptm′-A′).

The meta-analytical approach replicated the associations of rs2973015 with ANS-Me, rs2277071 with Ptm′-A′, and rs6920141 with Go-Pg. Subjects with the AA genotype for GHR rs2973015 showed higher ANS-Me measurements than subjects carrying the GG genotype ( P = 0.008, I ² = 49%; Fig 2 ). In contrast, subjects with the AA genotype in IGF2R rs2277071 presented reduced measures for Ptm′-A′; this effect was stronger when both A alleles were present (recessive model for AA vs AG + GG ; P = 0.03, I ² = 29%; Fig 3 ). Furthermore, subjects carrying at least 1 C allele for IGF2R rs6920141 had higher Go-Pg measurements (dominant model for CC + CT vs TT ; P = 0.02, I ² = 48%; Fig 4 ). No other association remained significant when the single effect size measurement for the different cohorts was calculated in meta-analytic approaches ( Supplementary Figs 1-3 ).

Forest plots showing the effect size on ANS-Me (mm) measure according to the genotype presented for the GHR rs2973015. SD, standard deviation; CI, confidence interval.

Forest plots showing the effect size on Ptm′-A′ (mm) measure according to the genotype presented for the IGF2R rs2277071. SD, standard deviation; CI, confidence interval.

Forest plots showing the effect size on Go-Pg (mm) measure according to the genotype presented for the IGF2R rs6020141. SD, standard deviation; CI, confidence interval.

The Curitiba sample showed significant differences in angular measurements of both jaws (SNA and SNB) for rs1509460 in GHR ( P <8.333 × 10 ⁻³ ). Complete results are presented in the Supplementary Data ( Supplementary Tables I-V ).

Discussion

The roles of GH, IGFs, and their receptors have been well described and recognized for participating in the physiology of skeletal growth and development. The expression of some of them in the mandibular condyle, ^{, ,} where cartilage-mediated growth occurs, suggests they can influence the mandibular morphology. Previous studies have suggested an association of missense mutations in GHR with the mandibular ramus height. ^{, , ,} Our results showed that this gene could also affect the sagittal dimensions of the mandibular body.

Although P561T in exon 10 of GHR has already been associated with horizontal and longitudinal variations in the morphology of the mandible, ^{, ,} this is the first study reporting association between rs2973015 and this dimension. Our findings indicated that this SNP could also affect lower anterior facial height. Interestingly, a previous study identified, by a principal component analysis, the involvement of this marker in the genetic background of horizontal and vertical maxillomandibular discrepancies in Brazilians. Because subjects carrying the AA genotype in the total sample presented significantly greater measures for Co-Gn and Go-Pg in the present study, we suggest that this SNP could be considered as a prognostic indicator of mandibular prognathism. It is important to mention that the genetic contribution of this SNP could vary on different populations; the present findings differ from other previously reported findings that found no association between this SNP and mandibular prognathism in a sample from the United States. In addition, the associations unveiled in the present study between this genetic variant and the measurements Co-Gn and Go-Pg were not replicated when the total sample was divided according to the region. These negative findings, however, are likely a direct result of a lack of statistical power because the cohorts contributing to these analyses had reduced sample sizes.

An interesting finding is that GHR could also be associated with the severity of skeletal variations. Subjects carrying the GG genotype for rs1509460 showed significantly higher angular measures for both jaws. This result from the cohort that had patients who underwent orthognathic surgery (therefore, from patients with more severe craniofacial deformities) could suggest a regulatory role of this genetic variant on the size and not as a contributor to a specific craniofacial characteristic.

Our results suggesting a role of IGF2R in variations of the craniofacial morphology may be explained by the regulatory function of the extracellular IGF levels of the receptor that this gene codes. IGF2R is a transmembrane receptor that transports IGF2 to lysosomes. IGF2R can remove IGF2 from circulation and extracellular medium and thus regulate the growth-promoting function of this growth factor. Increased systemic IGF2 levels could cause embryonic overgrowth and perinatal lethality ; in contrast, overexpression of IGF2R in a tissue-specific manner has been shown to cause a local reduction in organ size. It is important to mention that IGF2R not only binds to IGF2 but also to IGF1, although with less affinity. This means that alterations in IGF2R could also modify IGF1 levels and consequently their effect on target tissues. A previous study showed that Igf1 null mutant mice exhibited decreased craniofacial size and prominent changes in the facial and cranial areas. In humans, a previous case-control study of Chinese subjects found no association between IGF1 and mandibular prognathism. However, mandibular prognathism was defined on the basis of the facial features, presence of crossbite, and ANB angle (under −2.0°), and there could be a contribution of other structures besides the mandible for this phenotype.

Considering that IGFs are GH mediators and are believed to also stimulate growth independently by promoting cartilage and bone development, the association of rs6920141 with the Go-Pg measure may be due to the following: (1) SNPs in IGF2R that may alter the role of this receptor modifying the IGF levels exerting its function on the condyle and (2) the condylar growth that contributes to vertical height gains but also to the sagittal length of the mandible. In addition, a sagittal linear measurement for the maxilla (Ptm′-A′) showed a difference depending on the genotype for rs2277071. We believe that IGF has a role in the development of the maxilla because in the embryonic period, its presence in the maxillary prominences has been detected. Furthermore, although the maxilla has an intramembranous ossification, this bone is enlarged and displaced, to some extent, by the influence of the structures of the skull base that have endochondral ossification. Therefore, it appears that IGF2R may influence both the maxillary and mandibular morphology.

Some of the associations found in the combined sample did not remain significant when the single effect size measurement for the different cohorts was calculated in a meta-analysis. It has been reported that, as a whole, the ancestry contribution of Brazilians is 62% European, 21% African, and 17% Amerindian; however, the genetic admixture varies greatly depending on the region evaluated. For example, although the European contribution in the Southern region is higher with 77%, the Northern region has the largest contribution of Amerindian ancestry with 32%. The subjects included in our analyses belonged to cities located in different regions of the country; Rio de Janeiro and Ribeirão Preto are in the Southeast, Manaus in the North, and Curitiba in the Southern region of Brazil. Therefore, undetected population substructure is possible, and implementing approaches to minimize this effect is warranted.

Conclusions

In summary, our results provide additional evidence that GHR contributes to the determination of mandibular morphology. In addition, we report that IGF2R is a possible gene associated with variations in craniofacial dimensions.