Bilateral Asymmetry in the Tooth Relationships of Orthodontic Patients

It is well known that people's faces are not perfectly symmetric.1,2 For example, facial expressions typically are more obvious on a person's left side due to right hemispheric dominance.3 Clinically, left-right symmetry of the underlying skeletodental structures generally is a treatment goal,4,5 and studies suggest that symmetric faces are deemed more attractive.6,7 

Some asymmetries are acquired, for example, because of chewing side preference8 or trauma,9,10 but most left-right differences have no specific, identifiable etiology.11,12 Most asymmetries are subtle, requiring precise bilateral comparisons for their detection. These are evident when comparing the measurements of paired structures, but go unnoticed on casual clinical appraisal.13,14 

The purpose of the present study was to quantify left-right asymmetries in the dental relationships of samples of routine orthodontic patients studied according to Angle's classification of malocclusion. We assessed the kinds of asymmetry, their distributions and magnitudes in the dental arches, and correlations among them.

Pretreatment dental casts of Angle class I, II, and III malocclusions were assembled from a cohort of orthodontic patients who met three criteria:

• —All had intact permanent dentitions (excluding third molars), and none had prior treatment.

• —All were whites living in the US Midsouth to reduce variation.15,16 

• —No patient had a branchial arch syndrome, facial cleft, or any other condition known to enhance the risk of asymmetry.

Proportionate samples were collected by Angle class and sex. Total sample size was 211 individuals. Mean age at pretreatment was 14.0 years (SD = 2.1 years; range = 11 to 23 years).

Measurements were made using digital-readout sliding calipers on full-mouth dental casts. Interarch relationships were assessed with the casts in maximum intercuspation.17,18 Five sorts of variables were measured from each subject (Figures 1 and 2):

1. Deviation of the incisor midlines; a mandibular deviation to the right was given a positive sign.

2. Incisor overjet was measured separately on the left and right central incisors.

3. Canine deviation was the horizontal distance from the cusp tip of the maxillary canine to its normal position in the embrasure between the mandibular canine and first premolar.

4. Buccal segment relation (BSR) parallels Angle's molar classification,19 but on a continuous scale, where the horizontal distance of the buccal groove of the mandibular first molar is measured relative to the mesiobuccal cusp tip of the maxillary first molar. An idealized Class I relationship has a BSR of 0 mm; Class II relationships are given a negative value.

5. Arch chords are the straight-line distances from the incisive interdental papilla measured to (A) the distal-most aspect of the canine and (B) the distobuccal aspect of the first molar. Chords were measured from the midline to the canine (from central incisor through canine) and to the first molar (from central incisor through first molar) in each of the four quadrants.20,21 The variables primarily assess dentoalveolar asymmetries.

The technical error of measurement was assessed using the conventional Dahlberg statistic,22 namely

where X_1i and X_2i are the first and second measurements of specimen i. The unit of measurement does not cancel out, so d is expressed as the average millimetric difference attributable to measurement imprecision. Double determinations of 135 measurements yielded an average error of just 0.068 mm.

Two sorts of left-right asymmetry are examined,23,24 namely fluctuating asymmetry (FA) and directional asymmetry (DA). FA occurs for homologous dimensions when the sample distribution of the left-right differences is centered on zero.25 DA occurs when the mean of the distribution is shifted away from zero. DA is identified when the group average differs significantly from zero based on a one-sample t-test.26 Preserving the signs of the left-minus-right (L − R) differences,

where n is the number of cases measured.

The magnitude of FA is expressed as the absolute value of the side difference of a variable within each case. DA will confound the measure of FA,27 so the average DA for a sample is subtracted on a case-specific basis to center average L − R on zero:

Two-way analysis of variance was used to test for differences in the magnitude of asymmetry among Angle's three classes and between sexes.28 Statistics were evaluated as two-tail tests at α = .05.

Dental relationships can vary because of differences in the tooth positions within the supporting bone and because of size differences of the supporting arches. This mixture of sources is shown in Table 1, where 10 of the 15 variables have statistically significantly different mean sizes among Angle's classes.

Additionally, 11 of the tests between the sexes are significant. These statistics are largely confirmatory of (1) the facial proportions that characterize the three Angle classes and (2) the larger mean skeletodental dimensions in males. Summarily, these results show that (1) overjet and canine discrepancies are largest in class II and smallest in class III cases and (2) BSR is negative in class II cases, near-zero in class I, and positive in class III cases, simply because this is the fundamental trait used to classify the malocclusions. “Sex” is included in these analyses to control for the well-documented issue that men's arch dimensions tend to be larger than women's.29,30 

DA occurs when there is a systematic trend for subjects to have larger dimensions on one side, so the average L − R difference is offset away from zero.27  Table 2 shows that DA does not depend on Angle's class or on the subject's sex; none of the analysis of variance tests is significant. This warranted pooling the sample, and Table 3 shows that four of the eight variables exhibit directional asymmetry (Figure 3), namely the (1) canine relationship, (2) BSR, (3) maxillary 1-6 chord, and (4) mandibular 1-6 chord. In all four instances, the left dimension tends to exceed the right. Note that these small group averages obscure the considerable interindividual variation, where some asymmetries are quite obvious. Statistically, by repeated-measures analysis of variance, all four variables exhibit the same magnitude of DA (P = .08), with a grand mean of about half a millimeter (but with ranges exceeding a centimeter).

Table 4 lists the correlation matrix for the eight measures of DA, and roughly half of the associations (13/ 28) are significantly different from zero. With DA, the sign of the difference is retained, so positive correlations mean that both direction and magnitude of the asymmetries covary. The strongest correlations are between the arch chords within each jaw. The asymmetries in arch chords coincide with shifts of the dental midline. The 1-6 chord is defined by teeth that emerge at much younger ages than teeth in the midarch,31 suggesting that the asymmetries are already established once the incisors and first molars emerge. The correlations between the midline deviation and the 1-3 chords are of the same magnitude as for the 1-6 chords, so asymmetric positions of the late-emerging canine do not seem to affect the midline shift.

FA is the magnitude of difference between the sides (Table 5). Of the eight variables, three differ significantly by Angle class, although some differences are a bit complex (Figure 4). The interaction term is statistically significant for canine relationship, disclosing that the pattern of variation across Angle's classes is different in the two sexes. The canine relationship is statistically the same—about 1.2 mm—across all three Angle classes in females, but, in males, FA is significantly higher (about 2.6 mm) in the class II sample. Overjet FA differs among Angle's classes in that the average is significantly smaller in the class III sample, especially for females (though the interaction term is not strictly significant).

The third interclass difference is for BSR, but this is an artifact of sample selection. The class I sample was selected on the basis that both left and right BSR were close to zero, so these are contrived “special cases” where asymmetries in BSR were explicitly diminished. This classification bias is obvious in Table 5 where it accounts for the highly significant difference for BSR. Of note, this bias does not discernibly affect the magnitudes of asymmetry across the other variables.

Table 4 lists the matrix of correlations for FA (upper right of matrix). The question here is whether the magnitude of asymmetry in one variable is associated with the magnitude in another. Of the 28 correlations, 16 are significant. The two largest correlations are between arch chords 1-3 and 1-6 in the maxilla (r = .62) and the mandible (r = .67). These correlations are intuitive as the 1-3 chord is incorporated in the 1-6 chord, and Solow32 and others have commented on the geometrically dependent associations of overlapping dimensions. Two other, highly significant correlations involve deviations of the dental midline with BSR and with the canine relationship. In both situations, a deviation of the dental midline to one side corresponds to canine and molar deviations to the opposite side.

Kula et al33 found that malocclusions with greater anteroposterior (AP) discrepancies—specifically greater overjet—tend to have greater left-right asymmetries, and the present study shows this as well. Two measures of AP discrepancy are assessed to emphasize this point, overjet and BSR. Table 6 lists the correlations (Spearman's ρ) between the severity of the AP discrepancy and magnitude of the L − R asymmetries. As overjet increases, so does its L − R asymmetry (P < .0001) and, likewise, asymmetry of the canine relationship (P = .008). Kula's study33 focused on class II malocclusions; the present analysis generalizes the associations to the whole range of AP disharmonies. Associations are also significant for BSR, but the relationships are negative (simply because of the way BSR is coded). Across the spectrum of BSR (about −6 to +8 mm), the smaller the BSR (ie, more class II) the greater the (A) midline deviations, (B) overjet asymmetry, and (C) canine asymmetry. The association between the magnitude of BSR and the extent of canine asymmetry is representative of these relationships (Figure 5). We speculate that the principle underpinning these associations is the lack of intercuspation and occlusal guidance in cases where overt AP discrepancies leave the anterior teeth susceptible to greater left-right variations.

FA is thought to result from the accumulation of minor stochastic events during development.34 The two sides of the body are assumed to have the same genetic information, so phenotypic asymmetries result from the accumulation of minor differences between teeth in the two quadrants of an arch. During development any number of environmental issues may cause FA, such as side differences in times of primary tooth exfoliation (or extraction), position and orientation of the developing successor's tooth buds, differences in eruptive tempos and pathways, differences in tooth emergence and sequence, positions of antagonists, and so on.

Analyses (Table 5) show that the amounts of asymmetry for overjet and canine relationship differ among Angle's classes, but asymmetries of the chord distances do not. Overjet itself is small or even negative in class III malocclusions, and the present study shows that the left-right differences in overjet are significantly smaller in class III cases, especially in females, and most discordant in class II malocclusions. In severe cases of class III malocclusions, the incisors are in crossbite so the maxillary anterior region becomes the “contained arch,” which helps adjust the left and right central incisors symmetrically, so they are less asymmetric (though in crossbite).

Canine relationships are less symmetric in class II malocclusions because the maxillary canines are relatively forward and do not have the mandibular canine-premolar embrasure for guidance and stability, thus freeing them to exhibit greater asymmetry. Aside from the incisor and canine relationships, the other measures of FA are independent of Angle's class, which agrees with previous studies.4,14,35,36 

There are four variables (Figure 3) where the left quadrant characteristically exceeds the size of the right, a situation termed DA.37 In a classic craniometric study of human skulls by Woo,38 25 left-right paired dimensions were measured in some 900 skulls and tested for DA. Woo found that the skull is a collage of compensating side differences. In the midface (mandibles were not measured) there were significant L > R asymmetries—just as found here for arch chords. Neurobiologists attribute the facial directionalities to compensatory adjustments for right hemispheric dominance.39,40 

Few orthodontic studies have measured the same variables on both sides of the arch, so the comparative data are meager. Biggerstaff and Wells41 used a Cartesian coordinate system applied to occlusal views of dental casts. They offered no explanation for their finding that average arch length was longer on the left side (L > R), just as found here. Cassidy et al42 also measured the dental arches of orthodontic patients. They too found that “the left side of the arch is slightly but systematically larger than the right.”

Dentoalveolar asymmetries tend to be intercorrelated, probably because of dental compensations— asymmetries in one part of the arch contribute to other asymmetries because of the geometry of the dentition.43 

• Bilateral asymmetry is prevalent in the occlusion of routine orthodontic patients.

• The magnitudes of most asymmetries are equivalent across all three categories of Angle's classification.

• There are few sex differences in the magnitude of bilateral asymmetry.

• Asymmetries are greatest in severe class II malocclusions, probably because their anteriors have no functioning antagonists for guidance and stability.

• DAs, where the left arch dimensions are larger than the right, are confirmed in these data, and the cause may be hemispheric size differences in the central nervous system.