We investigated the long-term success of orthopedic treatment in skeletal Class III malocclusions, established a model to predict its long-term success, and verified previously reported success rates and prediction models.
Fifty-nine patients who underwent successful facemask treatment and were followed until growth completion were evaluated. After completion of growth, the patients were divided into successful and unsuccessful groups according to overjet, overbite, and facial profile. Pretreatment cephalometric measurements were compared between groups, and logistic regression analysis was used to identify the predictors of long-term success. Four previously published articles were selected to verify the success rate and predictability of the prediction models with regard to our patient sample.
The treatment success rate was 62.7%. The AB-mandibular plane angle, Wits appraisal, and the articular angle were identified as predictors. The success rates differed according to success criteria and patient characteristics. The prediction models proposed by the 4 previous studies and our study showed similar predictabilities (61.0%-64.4%) for our patient sample. The predictability for the unsuccessful group was low.
Our results suggest that no particular method or factor can predict the long-term success of orthopedic treatment for skeletal Class III malocclusion.
The success rate of Class III orthopedic treatment was 62.7%.
AB-mandibular plane angle, Wits, and articular angle were selected as predictors.
The success rate differed depending on success criteria and subjects.
Five prediction models showed similar predictabilities of 61.0% to 64.4%.
The predictability for the unsuccessful group was low.
The etiology of skeletal Class III malocclusion includes a retrognathic maxilla, a prognathic mandible, or a combination of both. More than two thirds of all skeletal Class III malocclusions are reportedly caused by a retrognathic maxilla or a combination of a retrognathic maxilla and a prognathic mandible. On the basis of the etiology, maxillary protraction has been considered a major treatment modality for growing patients with skeletal Class III malocclusion.
The prognosis of orthopedic treatment for skeletal Class III malocclusion is favorable when treatment is administered before the pubertal growth peak. However, a Class III problem may worsen during the growth peak if a patient is left untreated. Moreover, the negative psychological effect of having a Class III malocclusion during adolescence and the increase in self-esteem from early intervention can be as important as the visible treatment outcome. Therefore, early treatment is recommended for skeletal Class III malocclusion. However, a discrepancy between maxillary and mandibular growth during the pubertal growth period can result in relapse, and some patients eventually may need corrective jaw surgery. This may be the most frustrating situation, not only for patients but also for clinicians. Therefore, the availability of predictors to determine the eventual prognosis of early orthopedic treatment for skeletal Class III malocclusion can increase the reliability of decisions regarding the timing of treatment and the selection of cases.
Previous studies reported long-term success rates of 50.0% to 71.4% for the orthopedic treatment of skeletal Class III malocclusion. Because treatment success criteria, patient characteristics, and time points of outcome evaluation differed among studies, the reported success rates ranged widely and cannot be generalized to all patients treated for skeletal Class III malocclusion. Furthermore, even though studies have attempted to determine predictors of the long-term success of orthopedic treatment for skeletal Class III malocclusion, information on the prediction of posttreatment stability and suitable conditions for the application of orthopedic appliances remains controversial in each study, resulting in unstable predictability. Therefore, an attempt to verify the accuracy of the various predictors proposed by previous studies using 1 data set would help to minimize the controversy and to obtain consistent findings regardless of the criteria used to judge success or patient characteristics. However, the predictive accuracy of the previous prediction models has rarely been compared using patient records that were different from those that generated each prediction model.
Regarding the time points used for evaluation of outcome, progressive mandibular growth leads to difficulty in maintaining a corrected skeletal relationship. Variations in the timing, amount, and direction of mandibular growth increase the difficulty in predicting the final outcome. These variations also indicate that the success of Class III treatment should be evaluated after completion of craniofacial growth, not after the growth peak. However, most previous studies have determined the success of treatment before completion of growth, and each study used different criteria for success and different patient samples; therefore, the cephalometric predictors of long-term treatment success and their predictability remain controversial.
From these perspectives, we conducted this study to investigate the long-term success of orthopedic treatment for skeletal Class III malocclusion using records obtained after growth completion, to establish a novel method to predict long-term success, and to verify the effects of different success criteria, patient characteristics, and prediction models.
Material and methods
Fifty-nine patients (26 men, 33 women) with skeletal Class III malocclusion and an anterior crossbite who had facemask treatment during the growth period were included in this study. Initially, 73 patients were selected from the archives of the Department of Orthodontics, Yonsei University Dental Hospital, in Seoul, Korea, on the basis of inclusion and exclusion criteria. The inclusion criteria were correction of Class III problems after facemask treatment, initiation of orthopedic treatment at a skeletal maturity index of less than 4, and availability of pretreatment (T0), posttreatment (T1), and postretention (T2) records. T1 referred to the end of facemask protraction, whereas T2 referred to growth completion confirmed by fusion of the epiphysis and metaphysis of the distal radius on a hand-wrist radiograph. The exclusion criteria were second phase treatment with a fixed appliance, centric relation-centric occlusion discrepancy, congenital missing teeth, facial asymmetry with menton deviation greater than 2 mm at T0, systemic disease, craniofacial syndrome, and previous orthodontic treatment. Using facial photographs and dental casts obtained at T2, 3 orthodontists (Y.J.C., J.E.C., K.H.K) with more than 10 years of experience independently classified the patients into successful (n = 37; 14 men, 23 women) and unsuccessful (n = 22; 12 men, 10 women) groups according to our criteria for success: overjet and overbite greater than 1 mm and an acceptable facial profile. Fourteen patients were additionally excluded because the 3 experts did not agree with regard to the facial profile, although these patients showed an overjet and an overbite greater than 1 mm. Eventually, 59 patients were included. The mean age at T0 was 9.18 years, and the mean facemask treatment duration was approximately 1 year ( Table I ). The maxilla was protracted using approximately 350 to 450 g of force per side, directed slightly downward on the basis of the conventional protocol. All patients wore a facemask for at least 12 hours per day until a positive overjet was achieved. Then, they wore a Class III activator as a retainer at night for approximately 2 to 3 years. The mean overjets were –2.2, 3.7, and 2.8 mm at T0, T1, and T2, respectively, in the successful group, and –1.4, 3.2, and –0.6 mm at the same time points in the unsuccessful group. This study was approved by the institutional review board of Yonsei University Hospital (3-2014-0097). Because of the retrospective nature of the study, the institutional review board waived the requirement for written informed consent from the patients.
|Successful (n = 37)||Unsuccessful (n = 22)||Total (n = 59) (y)||P value|
|SMI||Chronologic age (y)||SMI||Chronologic age (y)|
|T0||0 (n = 12)
1 (n = 11)
2 (n = 5)
3 (n = 9)
|8.96 ± 1.77||0 (n = 8)
1 (n = 6)
2 (n = 5)
3 (n = 3)
|9.55 ± 1.58||9.18 ± 1.71||0.203|
|T1||0 (n = 6)
1 (n = 5)
2 (n = 6)
3 (n = 8)
4 (n = 5)
5 (n = 7)
|9.79 ± 1.82||0 (n = 1)
1 (n = 6)
2 (n = 5)
3 (n = 5)
4 (n = 3)
5 (n = 2)
|10.45 ± 1.43||10.03 ± 1.70||0.644|
|T2||11 (n = 37)||19.02 ± 1.85||11 (n = 22)||19.29 ± 2.53||19.10 ± 2.11||0.153|
Cephalometric analysis was performed using lateral cephalograms obtained at T0. For linear measurements, a horizontal reference plane (HRP) was drawn 7° downward from the sella-nasion line at sella, and a vertical reference plane (VRP) was drawn perpendicular to the HRP at sella. To indicate the maxillary and mandibular positions, we measured perpendicular distances between the following 6 landmarks and HRP and VRP, respectively: ANS, PNS, A-point, B-point, pogonion, and articulare. Other linear and angular measurements are shown in Figures 1 and 2 , respectively. Vceph software (version 6.0; Cybermed, Seoul, Korea) was used for the cephalometric analyses, and each lateral cephalogram was calibrated to scale by using a ruler attached to a head holder before the measurements. Using variables with statistically significant differences between the successful and unsuccessful groups, we formulated an equation to predict the long-term success of treatment.
We selected 4 previously published studies that had investigated cephalometric predictors of the long-term success of Class III orthopedic treatment during the growth period and had proposed prediction models. The success criteria and success rates in these studies are summarized in Table II . Three studies divided their patients into successful and unsuccessful groups, and the remaining study proposed 3 categories: good, fair, and poor. For conciseness and convenience, we categorized the good and fair groups as successful and the poor group as unsuccessful in our study.
|Study B||Study G||Study Y||Study M||This study|
|Success rate*||71.4% (30/42)||68.8% (44/64)||50% (16/32)||66.7% (30/45)|
|Success rate†||66.1% (39/59)||44.1% (26/59)||40.7% (24/59)||67.8% (40/59)||62.7% (37/59)|
We investigated the effects of different patient characteristics and success criteria on the success rates and the effects of different patient characteristics and prediction models on the accuracy of prediction (predictability). First, to investigate the effects of patient characteristics, we classified our patients using the T2 records into successful and unsuccessful groups on the basis of the success criteria in each study, and we calculated each success rate. The success rate derived from our patients was compared with the previously reported success rate in the corresponding study. Second, to investigate the effects of success criteria, we compared the 5 success rates derived from our patients on the basis of the success criteria in the 4 studies and our study.
Third, to investigate the effects of different patient characteristics on the predictability of the prediction models, we applied our patient records to the prediction models and success criteria in the previous studies and derived the prediction outcomes and treatment outcomes, respectively. Using the T0 records of our patients, we divided them into favorable and unfavorable (prognosis) groups by the prediction model proposed in each study. Similarly, we divided our patients into successful and unsuccessful (treatment) groups on the basis of the success criteria in each study using the T2 records of our patients. Next, per each study, we classified our patients into 4 subgroups: favorable-successful, favorable-unsuccessful, unfavorable-successful, and unfavorable-unsuccessful. The favorable-successful and unfavorable-unsuccessful groups represented accurate predictions, whereas the other 2 groups represented inaccurate predictions. Afterward, the accuracy of prediction for our patients was calculated using each previous model and compared with the previous results.
Finally, the predictabilities of the 5 prediction models for our patient sample were compared. We applied the T0 records of our patients to each prediction model to obtain the prediction outcome, and we applied the T2 records to the success criteria in our study to obtain the treatment outcome. The accuracy of prediction was then calculated for each model.
All data were statistically analyzed using software (version 8.02; SAS Institute, Cary, NC). The data are presented as means and standard deviations. Independent t tests were used to compare age and cephalometric measurements between the successful and unsuccessful groups. A logistic regression model was used to identify cephalometric variables at T0 that were significant predictors of treatment success or failure. P <0.05 was considered to be statistically significant.
One examiner (J.E.C.) performed all measurements. To determine the intraexaminer reliability, the same examiner reanalyzed 20 randomly selected cephalometric measurements within 2 weeks. The intraclass correlation coefficient (>0.92) indicated high reliability.
The treatment success rate in this study was 62.7% when calculated using our success criteria. The success rates in the 4 previous studies ranged from 40.7% to 67.8%, depending on the success criteria used. The success rates were relatively high (66.1%-67.8%) when judged by the criteria of 2 studies, but relatively low when judged by those of the remaining 2 studies ( Table II ).
Tables III and IV show comparisons of cephalometric measurements before and after treatment between the successful and unsuccessful groups in this study. The successful group had shorter distances between ANS, A-point, B-point, and pogonion from HRP, a higher Wits appraisal value, shorter distances of N-ANS and N-Me, a larger articular angle, and a larger AB to mandibular plane angle compared with the unsuccessful group at T0 ( P <0.05). After treatment, most of the variables were not significantly different ( P >0.05).
|Successful (n = 37)||Unsuccessful (n = 22)||P value||Successful (n = 37)||Unsuccessful (n = 22)||P value|
|ANS||64.4 ± 3.44||63.9 ± 5.47||0.692||66.6 ± 3.52||65.6 ± 5.93||0.412|
|PNS||18.2 ± 2.53||17.4 ± 3.19||0.277||18.5 ± 2.41||17.5 ± 3.32||0.192|
|A point||60.6 ± 3.50||59.9 ± 5.48||0.604||63.3 ± 3.82||62.1 ± 6.22||0.865|
|B point||59.8 ± 4.73||58.9 ± 6.48||0.661||57.6 ± 4.67||57.2 ± 6.45||0.813|
|Pog||59.4 ± 5.10||58.8 ± 5.89||0.771||57.1 ± 5.24||57.0 ± 6.26||0.972|
|Ar||13.1 ± 2.79||13.9 ± 2.50||0.301||13.6 ± 2.73||14.4 ± 2.52||0.272|
|ANS||42.3 ± 3.06||44.2 ± 3.83||0.045 ∗||43.9 ± 3.40||45.2 ± 3.64||0.162|
|PNS||40.8 ± 2.70||41.6 ± 3.46||0.331||42.7 ± 3.11||43.1 ± 3.42||0.623|
|A point||47.4 ± 3.86||49.8 ± 4.40||0.037 ∗||49.3 ± 3.88||50.9 ± 4.39||0.168|
|B point||87.5 ± 6.50||91.1 ± 6.59||0.048 ∗||91.2 ± 6.15||94.3 ± 5.85||0.061|
|Pog||96.4 ± 6.82||100.5 ± 6.89||0.030 ∗||100.3 ± 6.49||103.9 ± 6.41||0.046 ∗|
|Ar||28.1 ± 3.03||26.9 ± 3.44||0.162||29.0 ± 3.30||27.2 ± 3.72||0.066|
|ANS-PNS||46.3 ± 3.16||46.9 ± 4.00||0.492||48.2 ± 2.84||48.2 ± 4.27||0.930|
|Ar-A point||76.5 ± 3.69||77.4 ± 5.36||0.461||80.7 ± 5.77||80.4 ± 6.05||0.842|
|Ar-Pog||99.7 ± 5.40||101.9 ± 7.43||0.208||100.7 ± 5.58||104.2 ± 7.19||0.041 ∗|
|Ar-Go||41.5 ± 3.37||43.3 ± 5.15||0.148||41.6 ± 3.59||43.4 ± 5.13||0.115|
|Go-Me||69.8 ± 3.88||71.3 ± 3.79||0.141||71.4 ± 3.64||72.9 ± 3.78||0.134|
|Wits||–6.9 ± 2.57||–8.9 ± 2.39||0.005 †||–3.0 ± 2.16||–4.6 ± 1.64||0.003 †|
|N-ANS||50.0 ± 3.47||52.1 ± 4.06||0.039 ∗||51.6 ± 3.66||53.2 ± 3.89||0.104|
|ANS-Me||60.4 ± 4.65||62.5 ± 4.58||0.103||62.7 ± 4.39||64.7 ± 4.56||0.103|
|N-Me (ant facial height)||110.4 ± 6.99||114.6 ± 7.31||0.034 ∗||114.4 ± 6.80||118.0 ± 7.18||0.056|
|S-Go (post facial height)||69.8 ± 4.32||70.5 ± 6.91||0.669||70.9 ± 4.46||71.6 ± 6.88||0.672|
|Post/ant facial height (%)||62.7 ± 3.15||61.0 ± 4.81||0.148||61.5 ± 3.26||60.1 ± 4.80||0.186|
|Successful (n = 37)||Unsuccessful (n = 22)||P value||Successful (n = 37)||Unsuccessful (n = 22)||P value|
|SNA||79.0 ± 3.61||77.7 ± 3.48||0.191||80.9 ± 3.75||79.5 ± 4.26||0.193|
|SNB||80.0 ± 3.20||79.4 ± 3.87||0.519||78.6 ± 3.32||78.2 ± 3.79||0.706|
|ANB||–1.0 ± 1.68||–1.7 ± 1.65||0.171||2.4 ± 1.48||1.4 ± 1.46||0.019 ∗|
|PP angle (PP-MP)||1.9 ± 2.83||3.2 ± 3.31||0.109||1.5 ± 3.03||2.6 ± 3.38||0.173|
|MP angle (SN-MP)||29.1 ± 4.14||31.4 ± 5.58||0.073||30.8 ± 4.19||32.9 ± 5.70||0.104|
|Saddle angle (N-S-Ar)||122.7 ± 5.51||125.1 ± 4.43||0.092||123.0 ± 5.30||125.6 ± 4.46||0.053|
|Articular angle (S-Ar-Go)||147.2 ± 5.51||144.3 ± 5.04||0.047 ∗||149.2 ± 5.22||145.9 ± 5.30||0.023 ∗|
|Gonial angle (Ar-Go-Me)||126.0 ± 6.29||126.5 ± 3.92||0.065||125.3 ± 6.25||128.3 ± 4.44||0.054|
|Facial convexity angle (N-A-Pog)||–2.1 ± 4.43||–3.5 ± 4.46||0.230||4.8 ± 3.41||1.9 ± 3.99||0.004 †|
|AB plane angle (AB-NPog)||1.7 ± 2.64||2.2 ± 2.44||0.525||–3.3 ± 1.83||–1.8 ± 1.86||0.003 †|
|AB to MP angle (AB-GoMe)||62.3 ± 4.20||59.6 ± 2.80||0.011 ∗||66.9 ± 3.84||64.2 ± 3.59||0.008 †|
Nine variables—HRP-ANS, HRP–A-point, HRP–B-point, HRP-Pog, Wits appraisal, N-ANS, N-Me, articular angle, and AB-MP angle—were used for discriminant analysis. In a stepwise manner, 3 variables—AB-MP angle, Wits appraisal, and articular angle—were selected to distinguish the 2 groups. The discriminant formula was as follows.