The Influence of Sex and Ancestry on Three-dimensional Palate Shape

Abstract

Modern human palate shape has been reported to vary by sex and ancestry, but limitations in the methods used to quantify shape and in population coverage have led to inconsistent findings. In the present study, we aim to characterize the effects of sex and ancestry on normal-range three-dimensional palate shape through landmark-based morphometrics.

3D digital dental casts were obtained and landmarked from 794 adults of European (n=429), African (n=295) and East Asian (n=70) ancestry. Principal component analysis (PCA) was conducted to identify patterns of shape variation present in our cohort, and canonical variates analysis (CVA) was performed to test for shape differences between sexes and ancestries.

PCA showed that three PCs, explaining 76.52% of variance, linked higher palatal vault with either a relative reduction in anteroposterior or mediolateral dimensions. CVA showed that males had wider and shorter palates with more posteriorly located maximum vault depth than females. Individuals of African ancestry, having higher vaults with more posteriorly located maximal depths, also had wider and shorter palates, while individuals of European ancestry had narrower and longer palates with more anteriorly located maximum vault depths. Individuals of East Asian ancestry showed the shallowest vaults.

It was found that both sex and ancestry influence palate shape, suggesting a possible genetic component underlying this variation. Additionally, our findings indicate that vault height tends to co-vary with anteroposterior or mediolateral dimensions. Further investigation of these morphological patterns may shed light on possible links to common congenital anomalies such as orofacial clefting.

Introduction:

The modern human palate varies widely in shape and specific shape profiles are generally considered reliable indicators of population affinity and sex¹. The morphology of the modern human palate has generally been addressed in to the context of forensic identification; less is known about the biological basis for morphological variation within and among humans and its relationship to craniofacial anomalies². A number of early forensic anthropology studies devised systems to classify palates into a few key shapes (e.g. elliptical or hyperbolic) in an attempt to develop a simple method for use in the field^{3, 4}. Attempts to quantify palate shape soon followed. Using traditional anthropometry, Byers et al., and Burris and Harris measured the palates of individuals from different ancestral groups and reported a number of differences in length and width, many of which were sex dependent^{5, 6}. More recently, Maier et al. assessed 3D palate curvature and depth (vault height) from skulls assigned European, African, Native American, and Hispanic ancestry¹. Using a curve fitting approach, they found that in the majority of cases the shape of the palate matched an elliptical, hyperbolic, and parabolic profile and that there was a tendency to cluster by palate shape along ancestral lines. While most of these forensic studies have generally reported the ability to successfully discriminate ancestry on the basis of palate shape, Clark et al., applying a geometric morphometric approach, found little evidence to support palate shape differences between African-derived and Indigenous American groups⁷. Questions remain therefore about the degree and consistency with which palate shape differs among some populations.

There are some potentially important limitations in prior studies of normal-range palate variation. For example, while European-derived, African-derived, and Native American populations are well represented in studies of comparative palate morphology, few studies have included samples of recent Asian ancestry. Furthermore, relatively few studies have considered the height (or depth) of the palatal vault in their assessments. Maier et al. did attempt to assess vault height by calculating a simple linear distance and did find some evidence that this feature could distinguish ancestral groups, with European-derived individuals exhibiting significantly shallower vaults compared to African-derived and Hispanic groups. A limitation here, however, was that this feature was only considered in isolation and not in relation to other aspects of palate shape. Moreover, they did not control or factors such as the presence of torus palatinus, which varies in incidence by population⁸ and can dramatically alter the topology of the palatal vault and have an effect on palate shape⁹. Another limitation to direct linear anthropometric measurements is not controlling for potentially varying overall palate size among sexes and populations, which may impact palate shape.

Beyond forensic questions of identification, the assessment of palate shape along demographic lines is potentially important for other reasons. Palate shape may provide clues to the etiology of facial malformations. Orofacial clefting is the most common craniofacial birth defect in humans, affecting as many as 1/700 births worldwide¹⁰. Clefts occur when, during the first trimester, the paired processes that give rise to the primary or secondary palate fail to meet and/or fuse leaving a gap in the tissue^{10, 11}. Such palatal clefts are most often isolated (i.e., non-syndromic) defects, and can occur along with a cleft lip (CLP) or involving only the secondary palate (CPO). Findings from mouse models and human epidemiology show a distinction etiologically and pathogenetically between palatal clefts that occur with cleft lip and those that occur affecting the secondary palate only in isolation^11–14. Similar to palate shape, clefts also show large differences in prevalence by ancestry and sex. CLP occurs more commonly in males while CPO also occurs at roughly twice the frequency in females^{10, 11}. Asian-derived populations exhibit the highest rates of clefting, African-derived populations exhibit the lowest, and European-derived populations are intermediate¹⁰.

The factors that contribute to the population and sex differences in OFC incidence are still largely unknown. A long-standing hypothesis states that the shape of certain facial features may be a predisposing factor for orofacial clefting¹⁵, the idea being that those at higher risk for clefting will tend to exhibit certain facial features (e.g. increased upper facial width) more often than the general population. Both increased midface retrusion^{16, 17} and altered palate shape² have been reported in the unaffected biological parents of individuals with orofacial clefts. This suggests that palate shape may be a potential cleft risk phenotype. Comparing the average palate shapes of different population groups spread across the risk continuum for clefting would provide an additional test of this hypothesis.

In the current study, we investigate the effects of sex and ancestry on normal-range palate shape. We accomplish this by applying landmark-based morphometrics to 3D dental impressions collected on males and females of recent East Asian, European, and African ancestry. We will test whether palate shape differs among these demographic groups and describe the nature of these differences by modeling the shape variation visually. In our discussion, we will also comment on whether the observed shape differences can be meaningfully linked to orofacial cleft risk.

Methodology:

Study Population:

The study sample was comprised of 794 individuals recruited as unrelated healthy controls for a large international genetic study of orofacial clefting. All participants were at least 18 years of age (mean age = 35.4; maximum age = 80.2) and screened for a personal and family history of oral and craniofacial malformations (such as orofacial clefting) and prior trauma or surgery involving the palatal region. Individuals were also excluded if they presented with torus palatinus, missing maxillary canines, or missing maxillary first molars, as these features interfered with data collection. Individuals from three ancestral groups were represented: 429 US-Whites of European ancestry (157 males; 272 females), 295 individuals of African ancestry (157 males; 138 females) and 70 individuals of East Asian ancestry (42 males; 28 females). US-Whites of European ancestry were recruited from Pittsburgh and Lancaster, Pennsylvania. Individuals of African ancestry were recruited primarily from Lagos, Nigeria (N = 281), in addition to a small number recruited from Colombia (N = 8) and Puerto Rico (N = 6). Individuals of East Asian ancestry were recruited from The Philippines in Manilla (N= 35) and the US (N = 35). These ancestral groups were self-reported by the study participants based on US census demographic categories. Individuals who reported mixed ancestry (i.e. more than one race) were excluded.

Data acquisition and Phenotype Capture:

Individuals provided their written informed consent prior to participation. The protocol was approved by Institutional Review Board (IRB) of the University of Pittsburgh and locally at each recruitment site. Demographic data were recorded as well as dental, medical, and social history through in-person interviews. Maxillary impressions were taken by standard hydrocolloid impression materials and poured into plaster casts. The casts were then laser-scanned (3Shape, Copenhagen, Denmark). The scanned models were cleaned and processed using the 3Shape Orthoanalyzer software and stored as 3D meshes.

Morphometric Data Collection:

Landmarking was performed directly on the 3D digital models using 3dMDVultus software (3dMD Inc., Atlanta, Georgia, USA). Seven landmarks were placed, comprising three midline and two bilateral landmarks. These landmarks were located at the tip of the incisive papilla (IP), the deepest points of the gingival crevice (or cementoenamel junction) at the right and left canines (CR & CL) and the deepest points of the gingival crevice at the right and left first molars (6R & 6L), the midline between canines (CM), and the midline between first molars (MM). The midline landmarks were placed in a viewing orientation that is perpendicular to the anterior incisal/occlusal plane (Figure 1). The observer was blinded to both sex and ancestry during landmarking. For intra-observer validation, a subset of 30 casts were landmarked twice, separated by at least 24 hours, yielding intraclass correlation coefficients for coordinates in each of the three axes (x,y,z) ranging from 0.871 to 0.998, indicating low error.

Figure 1.

Landmarks on a dental cast in axial and sagittal views. IP = Incisive papilla. CR = Right canine. CL = Left canine. 6R = Right maxillary first molar. 6L = Left Maxillary First Molar. CM = Midline at canines. MM = Midline at first molars.

Statistical Analysis & Visualization:

Landmark configurations were subjected to Generalized Least-squares Procrustes Superimposition. This involved translation, rotation, and scaling into a common orientation. Geometric morphometric analyses were performed on the Procrustes transformed coordinates with two approaches. First, we applied principal component analysis (PCA) in order to visualize the key patterns of variation in palatal morphology in the combined dataset. Principal components (PC) contributing at least 5% of variance were considered here.

Second, we applied canonical variates analysis (CVA) on the landmark configurations to compare palate shape directly among groups, including stratified analyses by sex and ancestry. Like PCA, CVA is a data reduction technique, but it is operationalized to maximize differences between pre-defined groups in a dataset¹⁸. As such, CVA allows for a visualization of the aspects of shape variation most responsible for separating groups. P-values were determined based on permutation testing (10,000 permutations) of the Procrustes and Mahalanobis distances generated from CVA. These complementally metrics represent the distance between group centroids in shape space; statistically significant distances indicate that the groups are able to be distinguished or separated on the basis of shape¹⁹. Morphometric tests and subsequent visualizations were generated with MorphoJ, R packages geomorph and Morpho, and 3D Slicer^20–23.

Results:

On average, males had larger palates than females (comparison of centroid sizes; p < .0001). By ancestry, African-derived individuals had the largest palates while European-derived individuals had the smallest (p < .0001). The effect of allometry (defined as size-related difference in shape) was minimal, with centroid size contributing to only 2.32% of the variation in shape.

Principal Component Effects:

PCA yielded a set of 21 PCs out of which the first five explained 92.53% of variance in shape cumulatively. The remaining PCs each explained less than 5% of the shape variance and were not considered further. The first three PCs, explaining 76.52% of total variance, showed patterns linking higher palatal vault with either a relative reduction in the anteroposterior or mediolateral dimensions (Figure 2). The first PC (34.39%) showed a pattern of variation in which a higher palatal vault was coupled with a shorter anteroposterior dimension. The second PC (24.05%) coupled a higher palatal vault with a shorter mediolateral dimension. The third PC (18.09%) showed a higher posterior palatal vault coupled with a shorter anteroposterior dimension. The fourth (9.65%) and fifth (6.36%) PCs were related to the depth of the palatal vault; each showed varying positions of maximal palatal vault depth along the sagittal plane in the anteroposterior direction with no effect on anteroposterior or mediolateral dimensions. (Figure 2).

Figure 2.

Principal component effects in the combined sample in axial, lateral and frontal views. Red = Positive Effect. Yellow = Negative Effect.

Canonical Variates Analysis:

CVA yielded statistically significant shape differences in pairwise group comparisons by both sex and ancestry. Those differences were present in both the combined and stratified samples by sex and ancestry (Supplemental Tables 1 and 2; Figures 3 and 4).

Figure 3.

Canonical variate analysis of sex. The warp shows the effect of the canonical variate of sex in the combined sample. Males represented the positive effect while females represented the negative effect. Warp scale factor = 4.

Figure 4.

Canonical variates analysis of ancestry. The warps show the effects of CV1 and CV2 of ancestry in the combined sample. African-derived individuals represented the positive end of CV1 while European-derived individuals represented the negative end of CV1. Asian-derived individuals represented a relatively negative effect of CV2 in comparison to the other ancestries. Warp scale factor = 4.

By sex, with ancestries combined, males had shorter anteroposterior and wider mediolateral dimensions than females. In addition, males had higher posterior palatal vaults while females had higher anterior palatal vaults. The pattern of sex differences observed in the combined sample were similar within each ancestry (Supplemental Table 1; Figure 3).

By ancestry, the first canonical variate (CV) represented a shape continuum with African-derived individuals on one end and European-derived individuals on the other (with East Asians intermediate). This CV was associated with variation in the position of maximal vault height in the sagittal plane (more anterior or more posterior) as well as overall anteroposterior and mediolateral palatal dimensions. Compared to other groups, African-derived individuals tended to have more posteriorly positioned palatal vaults and palates that were shorter and wider compared to European-derived individuals. The second CV separated individuals of East Asian ancestry from the other two groups. This CV was associated with overall height of the palatal vault and East Asian-derived individuals tended to have the shallower vaults compared to the other groups. A similar pattern of ancestral shape differences was also found when males and females were analyzed separately (Supplemental Table 2; Figure 4).

Discussion:

The present study attempts to characterize patterns of normal variation in palatal morphology in a large ethnically diverse cohort. We found that palate shape differed significantly between males and females and among ancestry groups.

Sexual dimorphism in modern human craniofacial morphology has been studied through multiple methodological approaches and is well documented in the literature²⁴. There is evidence from the literature that the facial profile in females tends to be more convex than in males, which would have direct implication for palate shape^25–27. Our findings largely confirmed these reports, with females exhibiting longer palates anteroposteriorly with more anteriorly located maximal palatal vault height. Prior studies investigating sexual dimorphism in linear anthropometric dimensions of the palate found that males had overall larger palatal sizes with wider palatal widths and higher palatal vaults^{1, 5, 6, 28–30}. Our findings, using geometric morphometrics, revealed additional more subtle sex differences. For example, males had wider mediolateral and shorter anteroposterior dimensions than females. Females had higher anterior palatal vaults while males had higher posterior palatal vaults. In addition, these sexual dimorphic patterns in palate shape study were observed separately within each of the three ancestral groups.

Variation in facial morphology by ancestry has also been extensively studied³¹. Regarding palatal morphology specifically, comparisons of African and European ancestries involving anthropometric measurements showed that European-derived individuals consistently showed smaller measurements than African-derived individuals^{5, 6}. It was also found that the region contributing most to the difference in palatal width between African and European ancestries lied more anteriorly at the canine/premolar region, whereby European-derived individuals showed more convergent (or narrower) anterior palates shaped more like a catenary curve in comparison to U-shaped arches of African-derived individuals⁶. Our findings showed that, African-derived individuals showed wider mediolateral dimensions and shorter anteroposterior dimensions while European-derived individuals showed narrower and more constricted palates in comparison to the other ancestries. Regarding the palatal vault depth, Maier et al. reported that individuals of African ancestry had the highest palatal vaults while those of European ancestry had the shallowest¹. Again, these findings were reported as direct linear anthropometric measurement from fixed points. Our data shows that Europeans had the smallest centroid sizes and after scaling for unit centroid size, Asians were found to have the overall shallowest palatal vaults. Almeida et al. compared Asians to Europeans using cephalometry and reported that Asians had more obtuse cranial base flexure angle³². This cephalometry finding can be translated to a shallower palatal vault. This is because the cranial base flexure angle increases with the downward growth trajectory of the midface. Our findings also showed than Asian-derived individuals had the shallowest palatal vaults. The ancestral differences found in the current study were also observed separately within each sex. A potential limitation of the present study is the fact that the ancestral groups were self-identified by participants, which can result in bias. As such, our ancestral categories should be considered broadly constructed and likely include numerous subgroups and ethnic variations. This fact also hinders our ability to explore potential subtle shape difference among subgroups within the same broadly-defined ancestral groups. Approaches that assign ancestry based on genetic data are preferred but were not possible for this cohort at this time. Despite the broad nature of our ancestral groups, it was still possible to identify systematic differences in palate shape.

With regard to orofacial clefting, epidemiological patterns of cleft subtypes have consistently shown sexual dimorphism, with clefting of the lip with or without clefting of the palate (CL/P) occurring more commonly in males, and CPO occurring more commonly in females^{10, 11, 33}. Our sex-specific findings show that males tend to have shorter and wider palates with higher posterior palatal vaults while females have narrower, more slender palates with higher anterior vaults. This pattern of morphological dimorphism might provide insight on etiology towards the observed epidemiological patterns of orofacial clefting by sex, coupling findings of a palate that is shorter, wider and higher vaulted posteriorly with CL/P, whereas a palate that is narrower, more slender and higher vaulted anteriorly with CPO. Furthermore, the reported prevalence of CPO by population varies widely, while still unknown in some parts of the world^{11, 33}. Burg et al. reported CPO prevalence to be highest among some populations of European ancestry³³. Our ancestry-specific findings point towards individuals of European ancestry having narrower, more slender palates with more anteriorly located maximal palatal heights in comparison to other ancestries. These morphological patterns might represent an interesting risk factor for the development of orofacial clefting, warranting further investigation in relation to the genetic architecture of the human palate.

We observed that certain aspects of palate shape tended to co-vary. Based on our PCA, we found that a higher palatal vault was linked to decreased sagittal or transverse dimensions. This supports the results of Galvez et al., who conducted a cross-sectional study on a large cohort of children (N = 1065) and reported an association between a higher palatal vault with anterior and/or posterior cross-bites, which are an indication of deficient anteroposterior or mediolateral maxillary dimensions³⁴. Moreover, Huang et al. conducted a three-dimensional study of palatal morphology and reported that those with skeletal Class II malocclusions also had reduce palatal vault height²⁹. These types of integrated patterns are expected since the growth of the hard palate must be considered within the context of the entire cranio-maxillofacial complex³⁵.

In conclusion, we found that both sex and ancestry influence modern human palate shape, suggesting a possible genetic component underlying this variation. Additionally, our findings indicate that palatal vault height and anteroposterior and/or mediolateral palatal dimensions tend to co-vary. Further investigation of these morphological patterns may shed the light on possible links to common congenital anomalies affecting the midfacial region, such as orofacial clefting.