INTRODUCTION

Class II malocclusion is a common orthodontic concern that requires comprehensive treatment planning.1 Treatment of Class II malocclusion is frequently initiated in mid to late mixed dentition, wherein crowding and/or an increased overjet becomes of greater concern to patients and parents. Earlier correction of Class II abnormalities could be suggested in patients with significant occlusal discrepancies, increased risk of trauma to protruding upper incisors, and impaired masticatory functions.2

Class II correctors are appliances specifically designed to improve a combination of skeletal and dental components found among Class II division 1 malocclusions. Among the available Class II correctors, fixed options are gaining popularity. Systematic reviews3,4 have shed some light on what fixed Class II correction devices appear to produce during treatment of mild to moderate Class II malocclusion. Short-term changes include a combination of skeletal and dental modifications. Skeletal modifications include both maxillary restriction and mandibular repositioning, and dental effects consist of mandibular incisor proclination and maxillary molar distalization.

Cephalometric analysis is a valuable tool used for diagnosis and treatment planning of dental malocclusion and underlying skeletal discrepancies. In view of the fact that malocclusion is the product of an interaction between the alignment of erupting teeth in their basal bone and the skeletal position of the basal bone itself, cephalometric analysis can be used to evaluate dentoalveolar proportions and to elucidate the anatomic basis for jaw- and tooth-related abnormalities in the sagittal plane.1

The Xbow (pronounced “crossbow”) appliance is an orthodontic device that is used in late mixed or early permanent dentition before full fixed orthodontic treatment is initiated. It consists of a maxillary (Hyrax type) and a mandibular (lingual and labial arches) rigid frame linked together through a resilient fixed spring (FRD, Unitek, Monrovia, Calif). In comparison with the Herbst, bands are used instead of crowns in the first molars. Also, instead of a telescopic spring, which forces the mandible permanently forward (as in a Herbst), the Xbow springs allow the condyles to settle back if the patient forces the springs. Its main goal is to rapidly correct the occlusion in mild to moderate Class II malocclusions.5 Since the time of its introduction, only two published studies have reported on the Xbow appliance. One study6 focused on the evaluation of short-term skeletal and dental effects from lateral cephalograms; the other7 discussed treatment-originated lower incisor proclination according to vertical facial types. Both reported mild mandibular incisor proclination with significant variability after Xbow use. Facial type did not appear to significantly affect the amount of lower incisor inclination.

Because large variability has been observed in the magnitude of lower incisor proclination during Xbow treatment,6,7 it would be clinically beneficial to know whether initial values of 20 commonly used cephalometric variables could predict the final lower incisor proclination. This would potentially allow clinicians to treatment plan preventive measures or even discard use of the Xbow for specific patients. Therefore the objective of this study was to evaluate which cephalometric variables, or combination, can predict lower incisor inclination in mild to moderate Class II patients treated with the Xbow appliance.

MATERIALS AND METHODS

Methods

The sample was obtained from the private practice of two clinicians, both with significant experience in treating patients with the Xbow appliance. It consisted of 249 patients significantly treated with this appliance as part of phase I orthodontic treatment. All patients had both pretreatment and posttreatment lateral cephalograms taken between September 2002 and September 2009. As part of the coding required to send patients' radiographs between Canadian provincial boundaries, all identifiers, including age and gender, were removed. Based on a previous publication,6 the mean age of the patients was assumed to be around 11 years 11 months at T1 (Time 1, radiograph taken before appliance insertion) and 13 years 2 months at T2 (Time 2, radiograph taken 6.4 months after Xbow treatment). The mean time that the Xbow appliance was activated in the mouth was 4.5 months. The remaining difference corresponded to the time elapsed between T1 and actual insertion of the appliance. Some of the patient records used in the current study were also utilized in previous publications.6,7

All radiographs were taken with Orthoceph (model OC100D, General Electric, Tuusula, Finland). The cephalometric radiographs were scanned using an Epson Expression digital scanner (model 1680, Epson America, Long Beach, Calif) at a resolution of 300 dpi with no magnification of the radiographs. They were then printed and numbered randomly for blinding purposes. One individual not involved in the research had the codes. The 498 radiographs (T1 and T2 radiographs of 249 patients) were randomly assigned, and no information was given to the evaluator regarding the age of the patients or whether the radiographs were taken before or after treatment. As explained before, these x-rays were not taken with any orthodontic appliance in the mouth.

Commonly used cephalometric landmarks, reference planes, and measurements were used. Definitions for each landmark (S, N, A, B, Pg, Gn, Go, Ar, Me, ANS, PNS, Po) and plane (MP, SN, OP, FH, Np) are included in Table 1. A list of all cephalometric variables measured with their descriptions is provided in Table 2. All angles and linear variables were measured and rounded off to full degree and millimeter, respectively. All variables were measured at T1 and T2. Measurements for each variable at T1 were utilized in the statistical analysis; however, only lower incisor proclination (ie, L1-MP) was used at T2. For the purpose of consistency, each variable was measured consecutively on all (498) radiographs before quantification of the next variable. One evaluator placed all the landmarks, and the locations of the landmarks were verified by a second evaluator by randomly selecting 40 radiographs. Landmarks were placed on 50 lateral cephalograms per day at the beginning of the study; later only 100 measurements were taken per day to reduce operator fatigue. To assess intrarater reliability, 10 randomly selected radiographs were landmarked and measured three times with 2 days apart between sets. Intraclass correlation coefficients (ICCs) for each variable were above 0.8, suggesting good reliability (Table 3).

Table 1 Definitions of Landmarks and Reference Planes Used

View Fullscreen

Table 2 List of the Cephalometric Variables Measured and Their Descriptions

View Fullscreen

Table 3 Intrarater Reliability (ICC) Values for All Variables

View Fullscreen

RESULTS

Several predictor variables in this study measured similar relationships, meaning they were correlated with one another. Tolerance values were close to zero and the variance inflation factor was greater than 2 for most variables when stepwise regression was run using all cephalometric variables at T1. This confirmed a multicollinearity problem, which would result in exaggerated standard errors of the regression coefficients, making it less likely that the influence of predictor variables on lower incisor proclination could be assessed. With the use of PCA, the 20 predictor variables were grouped into a few uncorrelated components, whereby each component was a linear combination of the original variables. Principal components were then used as predictor variables and lower incisor proclination as a response variable in a multiple linear regression (MLR) model. MLR was chosen because the purpose of the study was to predict the mean lower incisor proclination after Xbow treatment from numerous cephalometric variables/components.

Principal Component Analysis

Extraction communalities were high for most variables (0.9 to 0.7 for all, except Ar-Go-Me [0.64], OB [0.4], U1-UL [0.39]) (Table 4). This proposed that the extracted components represented the original variables favorably. Six components were extracted based on eigenvalue greater than one, accounting for 78% of the cumulative variance (Table 5). Scree plot demonstrated the first four components on the steep part of the slope and then a significant drop in eigenvalue from component four to five (Figure 1). Therefore only four components were retained, explaining 60% of the variance of the original 20 variables based on the scree plot. The critical value for loading was 2*0.163 =  0.33, based on the sample size of 249 patients. Therefore, the factors extracted consisted of variables with factor loadings greater than 0.33. Components from varimax rotation were used for easier interpretation (Table 6). All components were combinations of linear and angular measurements; they represent the vertical and anterior-posterior (AP) distance/relationships of craniofacial structures.

• Component 1, skeletal component: positive correlation with original variables when vertical orientation is considered (SN-SGn, SN-PP, SN-MP, SN-OP, Ar-Go-Me), and negative correlation with AP relationships (SNA, SNB, PgNp, and SGo∶NMe ratio, which is measured anteroposteriorly)

• Component 2, incisal distance: positively correlated with AP measurements (OJ, U1-PP) and negatively correlated with vertical length (OB, L1-U1)

• Component 3, anterior skeletal projection: positively correlated with all AP relationships (SNA, SNB, ANB, PgNp, and ANp)

• Component 4, maxillo-mandibular relation: positively correlated with original variables for ANB, Wits (AP), and OB (vertical)

View Figure

• Download as Powerpoint
• Download Figure

Table 4 Extraction From PCA

View Fullscreen

Table 5 Variance Explained by the Six Extracted Components^a

View Fullscreen

Table 6 Components Extracted From PCA Using Varimax Rotation

View Fullscreen

Multiple Linear Regression

MLR was used with L1-MP at T2 as the response variable and the four principal components as predictor variables. Model assumptions of constant variance, normality, linearity, and independence were reasonably met. Component 3 (anterior skeletal projection) yielded a nonsignificant P value of .929 and therefore was removed and the model rerun. Both incisal and maxillo-mandibular relations (Components 2 and 4) were significant (Table 7), whereas the skeletal component (Component 1) was not significant (P  =  .071). It was still retained because it accounted for the greatest variability. P values of coefficients were β₀ (< .001), β₁ ( =  .071), β₂ ( =  .004), and β₃ (< .001) (significance level, alpha  =  .05). The skeletal coefficient in the linear regression equation was negative, suggesting a decrease of 0.78 degrees in the lower incisor inclination for each unit increase in skeletal distance (Figure 2), with constant values for other components. However, L1-MP at T2 would increase by 1.26 degrees for each unit increase in incisal distance (Figure 3) and by 1.52 degrees per unit increment of maxillo-mandibular distance (Figure 4). The regression coefficients were β₀ (intercept  =  98.51), β₁ ( =  −0.78), β₂ ( =  1.26), and β₃ ( =  1.52) (Table 7). The R squared statistic from regression was 0.091, which means that only 9% of variability in the response variable L1-MP at T2 could be explained by the skeletal, incisal, and maxillo-mandibular components. This would leave 91% of variability unaccounted for.

View Figure

• Download as Powerpoint
• Download Figure

View Figure

• Download as Powerpoint
• Download Figure

View Figure

• Download as Powerpoint
• Download Figure

Table 7 Multiple Linear Regression

View Fullscreen

The estimated regression equation was as follows:

DISCUSSION

In this study, we assessed anterior-posterior (SNA, SNB, ANB, ANp, PgNp, and Wits), vertical basal (SN-SGn, SN-PP, SN-OP, SN-MP, Ar-Go- Me, SGo∶NMe, and NSn∶SnMe), and dentoalveolar (L1-MP, L1-PP, L1-U1, L1-APg, OB, OJ, U1-UL) cephalometric relationships. Although initially the cephalometric variables were grouped into six principal components, only four were finally utilized as described in the Results. At the end of the selection process, the incisal and maxillo-mandibular components were statistically significant in predicting lower incisor proclination, but the obtained correlation value was of little clinical significance. In fact, the magnitude of proclination was more dependent on variation in individual patient factors. This is especially true in a sample of patients in which different degrees of initial dental and skeletal discrepancies affect the progress of each treatment effect. It should be noted that the size of this sample is large enough that it would depict a large spectrum of mild to moderate Class II division 1 clinical scenarios.

The skeletal component was a combination of vertical (positive correlation) and horizontal distances (negative correlation). This could be interpreted as larger vertical cephalometric measurements (longer faces/vertical growth pattern), which would be associated with less incisor proclination. This will be inversed in horizontal growers. Reduced proclination after Xbow treatment in long faces was reported in a previous study7 when compared with patients with short or normal facial height. Alternatively, lower incisor inclination post Xbow treatment could be more likely to occur in individuals who have a greater horizontal distance between anterior teeth as opposed to vertical overlap. This is expected as larger lower incisor proclination will be required to attain incisal contact. This hypothesis will have to be explored in the future.

For the maxillo-mandibular component, AP variables (ANB and Wits) had higher loadings and component coefficients when compared with OB variables (vertical). Therefore, it could be hypothesized that the greater the horizontal distance as opposed to vertical length between the jaws, the larger is the treatment-generated lower incisor proclination.

One potential limitation of our study could be the lack of cephalograms taken immediately after appliance removal. However, we do not perceive this as a drawback because physiologic recovery was allowed to take place before the T2 radiographs were taken. By using this approach, a clinician would know the amount of proclination after relapse and before full fixed appliances are used. Although radiographs taken immediately after removal of the springs could give some extra information, this was not an established protocol at the practice from which the radiographs were retrieved. In fact, most Class II treatment studies8–12 report T2 radiographs taken immediately after completion of treatment without allowing time for any muscle-splinting effect or physiologic relapse to occur.13

Other drawbacks include lack of quantification of appliance breakage, which would affect treatment time and simultaneous maxillary expansion in some patients. Additionally, a few patients received an upper 2 × 4 to create an overjet in cases of retroclined upper laterals. Information about these two factors was not readily available, but they could have been considered confounders. Finally, because of the fact that a previous study6 discussed cephalometric and dental changes rendering from Xbow treatment, we believed that there was no need to describe again the changes that occur in each cephalometric landmark during the course of this type of treatment.

Clinically, the Xbow appliance can be considered an efficient treatment alternative for mild to moderate Class II malocclusion6,7; however, the large variability in final proclination needs to be gauged during treatment planning. We were unable to identify cephalometric factors that could have predicted early identification of clinical cases that would result in significant lower incisor proclination during Xbow treatment. Clinicians have to monitor this treatment effect closely because of the unknown predictability of the magnitude of incisor proclination from the Xbow appliance.

Factors such as reduced thickness of the free gingival margin and the presence of gingival inflammation should also be considered before any orthodontic appliance that would procline incisors is used, as was shown in two recent systematic reviews.14,15 It needs to be emphasized that the amount of initial crowding was not considered in the analysis. Crowding is a key factor in considering whether a nonextraction class II correction is feasible. The clinician in these cases decided to go ahead with Xbow treatment after careful analysis of the malocclusion, including crowding. In some cases, extractions may be required before or after Xbow treatment based on the expected amount of incisor flaring.

CONCLUSION

• The best prediction model could account for only 9% of the total variability. Therefore, the average lower incisor proclination from Xbow treatment cannot be predicted, in a clinically meaningful way, with the use of common cephalometric variables at T1.