Linear Predictors of Facial Rotation Pattern in Croatian Subjects with Skeletal Class III Malocclusion.

OBJECTIVES: The objective of this study was to determine whether the linear measures of the maxilla, mandible and cranial base were predictors of facial growth rotation in a Croatian population with Class III malocclusion by cephalometric radiographic methods.
MATERIAL AND METHODS: The examined sample consisted of pretreatment lateral cephalometric records of 201 (111 females and 90 males) untreated Class III patients of Caucasian Croatian ancestry from the Department of Orthodontics at Zagreb University. The measurements were divided into five categories for analysis: cranial base, skeletal maxillary and skeletal mandibular relationships, sagittal intermaxillary and vertical relationships. Five multiple linear regression models were used to identify predictors of facial rotation pattern.
RESULTS: The effective length of the mandible was the most important predictor of facial rotation pattern, with the increased length largerly predisposing the tendency to the vertical growth pattern. No significant dichotomy was found regarding gender apart from the fourth model in which the posterior rotation of maxilla is related to female gender.
CONCLUSION: These predictors could help orthodontists determine timing and therapy for Croatian patients with Class III malocclusions.

Introduction

Class III malocclusions have been harnessing the attention of clinicians for hundreds of years due to the complexity of therapy, the possibility of relapses conditioned by individual growth and development and due to their great influence on facial esthetics. The incidence of relapse is reported in one article up to 50% (). Additionally, etiologic diversity makes it more difficult to diagnose and plan the therapy. Genetic and environmental factors are considered the cause of Class III malocclusions. Diversity of loci and suspicious genes have been identified using the linkage analysis and association studies (-). The tooth size and dental arch asymmetry were also recognized as important contributing factors to the etiology of malocclusions (-). In their research on fluctuating dental arch asymmetry, Škrinjarić et al. () concluded that the highest fluctuating asymmetry appeared in Class III anomalies, which suggest that patients with this malocclusion are under the greatest influence of genetic and environmental stressors during early growth and development.

The prevalence of Class III malocclusion shows the great variety in different ethnic groups, ranging from 0-26% (), being highest in the Asian population (). Chinese and Malaysian population showed a relatively high prevalence: 16% () and 17% (), respectively. The lowest prevalence is found in the African population, although the higher frequency of Class III malocclusion was found in two countries (). Recent studies showed a range of 2% to 6% among European population ().

For a correct diagnosis, timing and therapy plan in patients with Class III malocclusion, it is very important to look attentively at the vertical growth pattern. Schudy () investigated the interaction of horizontal and vertical facial dysplasias and determined the proportions by which he differentiated the average versus extreme facial types based on the SN-MP angle. He divided his sample of 120 patients into three groups based on their SN-MP angle and coined the phrase “facial divergence” as a method of indicating vertical variation. The two extremes of facial divergence were described as “hyperdivergent” for individuals with increased mandibular plane angle and “hypodivergent” for individuals with decreased mandibular plane angle. Other terms which have been used to describe different vertical facial types include high and low angle, which also refer to the degree of facial divergence, and long or short facial types, based on linear measurements of facial height. The Jarabak ratio shows the relation between the anterior (N-Me) and posterior facial height (S-Go). The formula is obtained by dividing the posterior facial height with anterior facial height, then multiplying by hundred. Values between 62-65% point to an average face, a higher percentage is seen in low angle cases, while smaller percentage points to high angle cases. ().

Considering the therapy plan it is controversial whether to start treatment immediately or to defer it until the growth is completed. Due to different procedures during therapy, it would be easier if we could immediately know whether the case could be resolved only by orthodontists, or by surgeons and orthodontists. Many authors agreed that a favorable outcome of an early treatment of Class III malocclusion is associated with a smaller gonial angle and a hypodivergent growth pattern while an unfavorable outcome is connected with vertical growth pattern (, ).

The aim of this study was to determine whether the linear values of the maxilla, mandible and cranial base could be predictors of facial rotation pattern in a Croatian population with Class III malocclusion by cephalometric radiographic method.

Subjects and methods

Sample

The sample of this retrospective study consisted of 201 patients with Class III malocclusion (90 male and 111 female; aged 12-20 years; mean age 15±3 years) and they were obtained from the archives of the Department of Orthodontics, Dental Clinic, Clinical Hospital Center Zagreb, Croatia. More than one thousand patient files were reviewed.

The inclusion criteria were as follows: 1. high quality of pretreatment lateral cephalograms, 2. age between 12 and 20 years, 3. Croatian ethnicity, 4. ANB angle less than 0.5°, 5. Wits appraisal of less than 0 mm for girls and less than: -1 mm for boys.

The standard of the ANB value was derived from a previous study on subjects of Croatian ethnicity with normal occlusion ().

Patients who exhibited an anterior mandibular shift, craniofacial syndromes, clefts, trauma, hypodontia and those who had already received orthodontic therapy were excluded.

The Ethics Committee of the School of Dental Medicine approved this study, as the patients were examined for routine diagnostic needs and future orthodontic treatment planning. All patients or their parents (if the patients were under 18) signed an informed consent form authorizing the use of their radiograms.

Cephalometric analysis

Lateral cephalograms were obtained under standardized conditions: in the maximal intercuspal position, using ear rods for stabilization (median plane focal distance: 1.55 m; detector to midsagittal distance: 0.125 m). Two devices were used. Twenty cephalograms were taken with a Planmeca PM 2002 CC Proline (Planmeca, Helsinki, Finland). Analog cephalograms were digitized using a Scan Maker i900 (Microtek, Willich, Germany). 181 digital ephalograms were stored on a CD-ROM in digital format and were taken with an Orthopantomograph OP200D (Instrumentarium Oy, Tuusula, Finland) with an average exposure time of 10 seconds and at values of 85 kV – 13 mA.

Cephalometric analysis was performed with DOLPHIN IMAGE software (v.11.0). To prevent magnification error and to calibrate each cephalogram in the DOLPHIN software to obtain real linear values, pictures were taken with a metal calibration ruler incorporated in the cephalostat and two ruler points reproduced on the headfilm.

On each cephalogram, ten cephalometric landmarks, representing hard tissues, were identified (Figure 1). From these landmarks, thirteen angular and linear measurements were recorded and analysed. The measurements were divided into five categories for analysis: cranial base, skeletal maxillary and skeletal mandibular relationships, sagittal intermaxillary and vertical relationships (Table 1). To determine the vertical growth pattern, the Bjork and Jarabak analysis and N-Me/S-Go were used. If at least two parameters indicated the same growth pattern, the patient was classified in that category.

Figure 1

Landmarks used in the study: S indicates Sella; N,nasion; Co, condilion; ar, articulare; A, point A (subspinale); ANS, anterior nasal spine; PNS, posterior nasal spine; Gn, gnathion; Go, gonial intersection; Me, menton;

Table 1

Five categories for the analysis

Cranial base
Linear: S-N, S-ar
Angular: N-S-ar
Maxillary skeletal relationships variables
Linear: Co-A
Angular: S-N:ANS-PNS
Mandibular skeletal relationships variables
Linear: Co-Gn, ar-Go, Go-Gn
Angular: Me-Go-ar, Me-Go:S-N
Intermaxillary relationships variables
Angular: ANS-PNS:Me-Go
Vertical relationships
Linear:N-Me, S-Go
Ratio: S-Go:N-Me (%)

Statistical analysis

Statistical analyses were performed using Statistical Package for Social Sciences software (version 10.0, SPSS, Chicago, SAD). The level of significance was set at P-values of < 0.05.

The normality of all cephalometric variables used for multivariate tests was confirmed using the Shapiro-Wilk test. After deriving descriptive statistics, multiple linear regression analysis was used to study the associations between linear measurements of maxilla, mandible and cranial base and facial growth rotation. As age and gender may influence the results and act as confounders, their significance was also tested for inclusion in the analysis.

To test the measurement error for the cephalometric variables used in this study, the lateral cephalograms of 30 randomly selected patients were redigitized 1 month later by the same examiner and were measured again using intraclass correlation coefficients (ICCs) with their respective 95% confidence intervals, measurement errors (MEs), smallest detectable changes (SDCs), limits of agreement (LoAs) and the relationship between the differences of the two measurements that were within the limits of agreement.

ME was measured according to the procedure described by Bland and Altman as the square root of the mean square error from an analysis of variance ().

Intraexaminer reproducibility was substantial to excellent (ICC=0.65-1.00). Measurement error was low (0.12-3.01) and was always lower than the biological variability of the associated variables.

Results

The multiple linear regression model for prediction of facial rotation pattern estimated by the Jarabak's ratio

Univariate correlations are presented in Table 2. With the control of gender, linear measurements of the maxilla, mandible and cranial base are statistically significant predictors of facial growth rotation estimated by the Jarabak's ratio (p<0.001). The entire regression model accounts for 69.3% of variability of the facial growth rotation. The largest independent contribution to the explanation of rotation pattern variability is the height of the ramus (Ar-Go) and the effective length of the mandible (Co-Gn) accounting for 46.8 and 34.3% of the variability. The length of anterior cranial base (S-N) and the length of maxilla (Co-A) (3.1 and 4.8%) (Table 3) had the smallest contribution. The horizontal growth pattern tendency was associated with shorter S-N, longer posterior cranial base length (S-Ar), longer Co-A, longer Ar-Go, longer length of mandibular corpus (Go-Gn) and shorter Co-Gn. The age positively linearly correlated with each linear variable (range r = 0.29-0.57) and it was creating a problem of multicollinearity in the multiple regression analysis, which is why it was not included in the analysis.

Table 2

Spearman's correlation

		Jarabak	s-n	s-ar	co-a	ar-go	go-gn	co-gn	gender	age
Jarabak	r	1.000	0.112	0.362	0.211	0.418	-0.008	0.005	-0.050	0.049
s-n	r		1.000	0.475	0.820	0.576	0.691	0.708	-0.219	0.289
s-ar	r			1.000	0.597	0.413	0.483	0.549	-0.229	0.315
co-a	r				1.000	0.572	0.759	0.781	-0.205	0.316
ar-go	r						0.495	0.745	-0.172	0.530
go-gn	r						1.000	0.870	-0.135	0.447
co-gn	r							1.000	-0.209	0.570
gender	r								1.000	-0.058
age	r									1.000
Jarabak	p	.	0.056	<0.001	0.001	<0.001	0.456	0.472	0.242	0.243
s-n	p		.	<0.001	<0.001	<0.001	<0.001	<0.001	0.001	<0.001
s-ar	p			.	<0.001	<0.001	<0.001	<0.001	0.001	<0.001
co-a	p				.	<0.001	<0.001	<0.001	0.002	<0.001
ar-go	p					.	<0.001	<0.001	0.007	<0.001
go-gn	p						.	<0.001	0.028	<0.001
co-gn	p							.	0.001	<0.001
gender	p								.	0.205
age	p									.

Table 3

Multivariate linear regression model for prediction of rotational growth pattern estimated by Jarabak's ratio posterior and anterior facial height

	Non-standardized coefficient B	Standard error	Standardized coefficient Beta	Sig.	Correlations
					Zero order	partial	Semipartial
(Constant)	62.066	2.768		<0.001
s-n	-0.296	0.066	-0.323	<0.001	0.112	-0.307	-0.176
s-ar	0.494	0.058	0.428	<0.001	0.362	0.524	0.335
co-a	0.314	0.057	0.466	<0.001	0.211	0.371	0.218
ar-go	0.901	0.052	1.196	<0.001	0.418	0.782	0.684
go-gn	0.346	0.051	0.647	<0.001	-0.008	0.436	0.264
co-gn	-0.720	0.048	-1.822	<0.001	0.005	-0.733	-0.586
gender	-0.135	0.356	-0.016	0.706	-0.050	-0.027	-0.015

R=0.839; R2=0.703; Adjusted R2=0.693; p<0.001.

The multiple linear regression model for prediction of facial rotation pattern estimated by angle between the mandibular plane and cranial base

The mandibular plane angle model accounted for 60.5% of the varibility in facial rotation pattern (p<0.001; Table 4). With the control of gender, the vertical growth pattern tendency was associated with longer S-N and decreased S-Ar, Co-A, Ar-Go,Go-Gn and increased Co-Gn. Ar-Go and Co-Gn contributed the most (29.4 and 51.3%, respectively). The S-N and S-Ar (2.2 and 2.8%) had the smallest contribution. Again, gender was not a significant predictor and age was not included due to multicolinearity problem.

Table 4

Multivariate linear regression model for prediction of rotational growth pattern estimated by inclination of the mandible in relation to the cranial base

	Non-standardized coefficient B	Standard error	Standardized coefficient Beta	Sig.	Correlations
					Zero order	partial	semipartial
(Constant)	37.877	4.290		0.000
s-n	0.339	0.102	0.271	0.001	-0.161	0.232	0.147
s-ar	-0.334	0.090	-0.212	<0.001	-0.222	-0.259	-0.166
co-a	-0.617	0.088	-0.672	<0.001	-0.266	-0.452	-0.313
ar-go	-0.976	0.080	-0.948	<0.001	-0.186	-0.660	-0.542
go-gn	-0.847	0.080	-1.159	<0.001	-0.122	-0.608	-0.473
co-gn	1.201	0.075	2.225	<0.001	0.053	0.757	0.716
gender	0.443	0.551	0.037	0.422	0.018	0.058	0.036

R=0.786; R2=0.618; Adjusted R2=0.605; p<0.001.

Multiple linear regression model for prediction of facial rotation pattern estimated by Bjork polygon

A finding of regression that as an initial variable assumes inclination of the mandibular plane in relation to the cranial base also confirms regression with Bjork polygon as the outcome, accountig for 60.6% of variability (Table 5).

Table 5

Multivariate linear regression model for prediction of rotational growth pattern estimated by Bjork polygon

	Non-standardized coefficient B	Standard error	Standardized coefficient Beta	Sig.	Correlations
					Zero order	partial	semipartial
(Constant)	397.843	4.278		0.000
s-n	0.340	0.102	0.272	0.001	-0.161	0.233	0.148
s-ar	-0.333	0.089	-0.211	<0.001	-0.222	-0.259	-0.165
co-a	-0.619	0.087	-0.674	<0.001	-0.267	-0.454	-0.315
ar-go	-0.977	0.080	-0.950	<0.001	-0.187	-0.661	-0.543
go-gn	-0.844	0.079	-1.158	<0.001	-0.121	-0.608	-0.472
co-gn	1.200	0.074	2.226	<0.001	0.054	0.758	0.716
gender	0.445	0.550	0.038	.420	0.018	0.058	0.036

R=0.787; R2=0.620; Adjusted R2=0.606; p<0.001.

Multiple linear regression model for prediction of facial rotation pattern estimated by inclination of the maxillary plane to the cranial base

The statistically significant predictors of the growth rotation of the maxilla were the S-N and S-Ar and gender (Table 6). The posterior rotation of maxilla, which is related to female gender, increased S-N and reduced S-Ar. The model accounted for 14.3% of variability and the most significant independent contribution gave S-Ar (7.1%).

Table 6

Multivariate linear regression model for prediction of rotational growth pattern estimated by inclination of the maxilla in relation to the cranial base

	Non-standardized coefficient B	Standard error	Standardizirani koeficijent Beta	Sig.	Correlations
					Zero order	partial	semipartial
(Constant)	7.539	3.609		0.038
s-n	0.207	0.086	0.290	0.017	-0.053	0.171	0.158
s-ar	-0.307	0.075	-0.340	<0.001	-0.337	-0.281	-0.266
co-a	-0.018	0.074	-0.035	0.803	-0.151	-0.018	-0.016
ar-go	-0.085	0.067	-0.144	0.210	-0.150	-0.090	-0.082
go-gn	-0.090	0.067	-0.215	0.183	-0.154	-0.096	-0.087
co-gn	0.052	0.063	0.170	0.405	-0.172	0.060	0.055
gender	1.277	0.464	0.188	0.006	0.228	0.194	0.180

R=0.416; R2=0.173; Adjusted R2=0.143; p<0.001.

Multiple linear regression model for prediction of facial rotation pattern estimated by intermaxillary plane angle

Significant predictors of hyperdivergent jaw growth pattern were decreased Co-A, decreased Ar-Go and Go-Gn and increased Co-Gn (Table 7). The entire regression model accounted for 56.8% of the rotational growth variability.

Table 7

Multivariate linear regression model for prediction of rotational growth pattern estimated by intermaxillary angle

	Non-standardized coefficient B	Standard error	Standardized coefficientBeta	Sig.	Correlations
					Zero order	partial	semipartial
(Constant)	30.281	4.371		0.000
s-n	0.133	0.104	0.109	0.203	-0.133	0.092	0.059
s-ar	-0.026	0.091	-0.017	0.775	-0.031	-0.021	-0.013
co-a	-0.599	0.089	-0.668	<0.001	-0.184	-0.435	-0.312
ar-go	-0.891	0.082	-0.888	<0.001	-0.104	-0.618	-0.508
go-gn	-0.756	0.081	-1.061	<0.001	-0.034	-0.557	-0.433
co-gn	1.147	0.076	2.179	<0.001	0.155	0.736	0.701
gender	-0.826	0.562	-0.071	0.143	-0.115	-0.105	-0.068

R=0.764; R2=0.584; Adjusted R2=0.568; p<0.001.

Discussion

Class III malocclusion can contain many combinations of skeletal and dentoalveolar elements. All those morphological features differ from population to population (). The knowledge of various morphological features is important to make the outcome of the therapy as complete and individual as possible. Although some orthodontists recommend early treatment of skeletal Class III malocclusion, relapse can occur due to the postadolesent growth and therefore the case may become surgical. The therapy with facemask can displace maxilla forward and downward and the mandible backward (-). Those rotational and translational movements may lead to unwanted effects in high angle patients (-). It will worsen pre-existing hyperdivergent facial profile. In patients with maxillary retrognathism and small gonial angle, facemask therapy increases the vertical dimension and causes a favorable esthetic change (). That is why the growth pattern is quite important in therapy decision. The patients with Class III malocclusion and vertical growth pattern are considered the most difficult to treat. The aim of this study was to determine whether the linear values of the maxilla, mandible and cranial base could be predictors of facal growth rotation in a Croatian population with Class III malocclusion which would be of great help when making a decison about the treatment timing and choice of therapy (orthodontic, orthopaedic, surgical).

Five prediction models were used.

In the first model, the largest contribution to the explanation of rotation pattern variability was the height of the ramus. It was expected that the longer ramus height is associated with the horizontal growth pattern, which is in agreement with the study of Siriwat and Jarabak (). The second largest contribution was the effective length of the mandible. It is likely that the reduction in the effective length of the mandible is due to reduction of the gonial angle, that is because of the mandibular rotation around the point of the gonion. The anterior cranial base length and the length of maxilla were poor predictors of facial rotation pattern. This can be explained by the fact that there is variability in the position of point A-supraspinale and point N-nasion during growth, which can lead to contradictory findings.

The majority of authors of previous studies assessed predictors of the result of orthodontic treatment in Class III patients where discriminant function or regression analyses were used to identify the variables showing the highest prediction potential (prediction models). Fudalej () made a systematic review based on 14 selected scientific papers and found out that there were no studies having the same predictors of treatment outcome. Despite the extensive cephalometric analyses carried out, only the mandibular ramus length (, ) and total mandibular length (, ) appeared in more than one research. These findings are in line with our study where the largest independent contribution to the explanation of rotation growth pattern variability is the height of the ramus and the effective i.e. total length of the mandible.

In the second and third model as expected, an increased effective length of the mandible was the strongest predictor of vertical rotation pattern. However, it should be pointed out that the reduction of ramus and mandibular corpus is a predictor of mandibular post-rotation. Isaacson et al. () obtained similar results regarding the height of the ramus. The mean height of the ramus was reversly related to the SN-MP angle i.e. ramus was shorter in the high and largest in the low angle group.

A decreased maxillary length was also a predictor of vertical rotation pattern. This may be because the maxilla is positioned more distally, thus causing the post-rotation of the mandible. Similar results was given by Ferrario et al. (). The aim of theit study was to find the relationship between the mandibular size and the shape to skeletal divergency (according to MP-SN angle) and found that hyperdivergent subjects generally have a smaller maxilla and mandible.

In the fourth model is interesting that the length of the maxilla is not at all a predictor of the rotational growth of the maxilla, only anterior and posterior cranial base length (increased S-N and reduced S-Ar are related to the posterior rotation of the maxilla). It my be explained by the fact that the length Co-A affects anteroposterior position of maxilla whereas the growth of cranial base (anterior part) causes the translation of the nasomaxillary complex-secondary displacement and affects also a vertical dimension. Significantly higher mean value of ANS-PNS:S-N angle in females than males means that the maxillary plane is more downward positioned relative to the cranial base; this may relate to a caudal jaw growth rotation in females.

In the fifth model, the statistically significant predictors of divergent jaw growth pattern are a decreased maxillary length, decrease of ramus height and mandibular corpus length, and increase in effective mandibular length. Björk () demonstrated that an open bite is associated with a large ramus while Sassouni () and Schudy () reported that open bite usually goes with shorter ramus. Hellman () suggested that a short ramus and corpus, rather than vertical development in nasomaxillary complex, leads to the development of open bite.

In four out of five models, the vertical growth pattern is assocciated with decreased ramus height and mandibular corpus length and increased effective mandibular length. Mangla et al. () evaluated mandibular morphology in different facial types and found a significantly increased ramus height in hypodivergent and normodivergent groups when compared to hyperdivergent group. These results coincide with conclusions of a study by Sassouni (, ), Muller () and Schudy () who observed a significant reduction in the width and height of ramus in the hyperdivergent group. This finding may be explained by highly significant negative correlations between the ramus height and angles of mandibular rotations (SN-MP, PP-MP, Ar-Go-Me), which compensate the effect of downward mandibular movement with the increase in ramus heigh,t and hence decrease its effect on the anterior facial height.

Conclusion

The effective length of the mandible was the most important predictor of facial rotation pattern, with the increased length largerly predisposing the tendency to the vertical growth pattern. No significant dichotomy regarding gender was found except in the fourth model where the posterior rotation of the maxilla is related to female gender.

A patient with maxillary retrognathism and greater height of mandibular ramus and shorter effective length of mandible is more likely to succeed in early treatment with facemask therapy.