Yang’s Classification of Juvenile TMJ Anterior Disc Displacement Contributing to Treatment protocols

This study aims to establish a new staging system of temporomandibular joint (TMJ) anterior disc displacement (ADD) and evaluate its role in guiding the treatment plan. A consecutive sample of 522 juvenile patients (780 joints) diagnosed as ADD based on magnetic resonance imaging (MRI) was included and analyzed. 674 joints received TMJ treatments according to our staging system, while 106 joints rejected any treatment; only for follow-up. The outcomes were judged according to our success criteria. The prognosis of our staging system was also evaluated in comparison to Wilkes classification. Kaplan–Meier survival analysis showed that significant stratifications of the ameliorative rate were found at all subgroups within any two stages in our staging system, except for subgroups between stages 0 and 1, stages 2 and 3, and stages 3 and 4. After analyzing the interactions between different candidate prognostic factors in a Cox model, the relative risks of deterioration of ADD included treatment methods (HR = 42.94, P < 0.0001), disease course (HR = 0.98, P = 0.0019), stages of ADD (HR = 3.81, 9.62, 12.14, P = 0.016, 0.000,0.000 respectively for stage 2, stage 3 and stage 4) and the interaction between ADD stages and treatment methods. The C index of this model was 0.87. The new staging system of TMJ ADD appears reliable, and benefits to making treatment planning and predicting the prognosis.


Validation of the Predicting Model.
Validation of the predicting model on our dataset was performed at last. The C index of 0.87 indicated 87.0% accuracy in ranking our patients according to the risk of deterioration. Calibration was assessed at 60 months from their first visit to allow a direct comparison with our models. The calibration plot of the predicted versus observed probabilities of amelioration showed good calibration with an accurate risk estimation (Fig. 2).   www.nature.com/scientificreports www.nature.com/scientificreports/

Discussion
Condylar resorption of the TMJ, which predominately occurs in adolescents, can lead to mandibular asymmetry or retrognathia and other dentofacial deformities 25,26 . Despite a number of theories, the causation of condylar resorption has never been exactly demonstrated so far.
ADD, as one of the most common TMJ diseases, has gained an extensive attention not only in the TMJ field but also in the orthodontics and orthognathics fields. The reason might be the potential correlation between ADD and condylar resorption. Many studies have reported that ADD in teenagers is closely related to mandibular asymmetry and mandibular deficiency, particularly retrognathia 27,28 . Orthognathic surgeons have found that 88% of patients who had an orthognathic surgery for a class II malocclusion had TMJ disc displacement 29 . Moreover, surgeons have noted an increased rate of Class II relapse following orthognathic surgeries in young patients with ADD 30,31 . In our clinic, we also found that patients with ADD often had condylar resorption of the TMJ. Besides, the longer ADD history, the more severe condyle resorption and dentofacial deformities 32,33 . All of the above-mentioned results indicate that juvenile ADD is a major causative factor for condylar resorption resulting in mandibular asymmetry and retrognathia.
Although the potential relationship between ADD and condylar resorption has been implied in many studies 34,35 , there are no reports describing the severity of this disease. In the present study, we have designed a new staging system for juvenile ADD and condylar resorption based on MRI images. Since MRI is a noninvasive technique which allows a valid and reproducible assessment for both osseous components and soft tissues of the TMJ, it is used as a primary evaluation method for staging in the present study. The superior imaging sensitivity of MRI allows for detecting cartilage and bony changes at an early stage. In our staging system, we have divided juvenile ADD into five stages according to the degrees of condylar resorption and status of the disc, as well as considering stages 3 and 4 as ADD with severe condylar resorption. The last two stages have been further subdivided into sub-stages A and B, based on the shape and length of the disc, inflammatory changes in the condyle, and the presence of bone marrow in the condylar head. The purpose of such sub-staging is to reflect the ability of condylar regeneration and possibility of disc repositioning. The authors believe that stages 3 and 4 juvenile ADD actually represents what has previously been termed idiopathic condylar resorption (ICR).
Another aim of our staging system is to provide a reasonable treatment protocol for ADD patients with varying severities. Since the length of the disc in stages 0, 1, 2 and 3A is barely enough to cover the top of the condyle, reposition treatments, including functional splint and arthroscopic or open surgeries, are recommended. Actually by many years follow-up, we found that a quite number of juvenile ADD patients had condylar regeneration with new bone formation after disc reposition, which will be presented in our future article. As for stage 3B patients, because of their markedly distorted disc and seriously resorbed condyle, disc reposition is deemed as noneffective. If these patients lacked any TMJ symptoms with a disc-like character in the TMJ bilaminar zone, then MRI was repeated after 6 to 12 months to determine whether condylar resorption had stopped. Once this was ascertained, the dentofacial deformity was corrected using moderate orthodontics and/or orthognathic surgery but without TMJ treatment. If condylar resorption continued, the patient was then treated according to stage 4. In this stage, since condylar bone marrow is markedly reduced (stage 4A) or absent (stage 4B), we believed that little or even no condylar regeneration would be possible. Thus, in this case, we routinely resect the condyle and replace it with costochondral grafts (CCG) or total alloplastic TMJ replacements.
To evaluate our staging system, we compared it with Wilkes's classification. Although Wilkes staging system was widely used in the TMJ field, its predictive value for the TMJ is still unclear. In the present study, all the cases were classified according to both staging systems. By using the Kaplan-Meier survival analysis, significant stratifications in our staging system were observed compared with Wilkes staging system. This implied that our system is superior to Wilkes system in the prognostic value of the disc displacement.
To find which variables were related to the prognosis, we have also used Cox regression analysis. Our results showed that the treatment methods, disease course, our stages and the interaction of our stages and treatment www.nature.com/scientificreports www.nature.com/scientificreports/ methods were associated with the prognosis. The higher stage and absence of treatment method were especially detrimental and correlated with the deterioration. Although the disease course was a risk factor for the deterioration by the univariate analysis, it became protective in the multivariate Cox model. We think this is because the disease course is relevant to our stage. That means the longer disease course, the higher ADD stage 33 . However when ADD patients were processed by suitable treatments, their prognosis became reversed. The predictive accuracy and calibration manifested that this model had a good predictive value.
To our knowledge, the ADD staging system is the first to demonstrate the severity of TMJ disc displacement and its relationship with condylar resorption. The results have shown that this staging system is feasible, and benefits in guiding the treatment protocol and predicting the prognosis.
However we still can not elaborate why ADD in young patients will deteriorate into serious condylar resorption, and its pathogenic mechanism needs to be further explored in the future. Certainly, a longitudinal study of ADD will give us more information regarding this. From our results, we think that it is very important to inhibit or reverse the deteriorating changes of the condyle secondary to ADD at an early stage. We emphasize that the salvageable disc should be repositioned as early as possible.

Conclusion
In conclusion, we believe that the new staging system of TMJ ADD appears reliable, and benefits in guiding the treatment plan and predicting the prognosis.

Patients and Methods
Ethics Statement. This study followed all the tenets of the Declaration of Helsinki for research involving human subjects, and was critically reviewed and approved by the institutional review board of Shanghai Jiao Tong University School of Medicine. An informed written consent was obtained from all participants or their legal guardian (patients under the age of 18 years).

Patients Selection.
A total of 3177 consecutive juvenile patients (age range from 10 to 20 years) who visited the TMJ Clinic at the Department of Oral Surgery in Shanghai Ninth People's Hospital from August 2009 to October 2012 were reviewed in the current study. Among them, those who were diagnosed as ADD of the TMJ based on magnetic resonance imaging (MRI) scanning according to the reported methods 36,37 and had more than one time follow-up MRI scanning in our department, were included in our study. The patients with histories of congenital deformities (e.g. microsomia, Treacher Collins syndrome), systemic disease (e.g. juvenile rheumatoid arthritis), infection, jaw fracture, or other clinically significant pathologies affecting the growth of the condylar were excluded in our study. Demographics and clinical course were collected from the recorded data of all included cases. The duration between first occurrence of clicking, pain of the TMJ or mouth opening limitation and their first visit to our department was considered as the disease course. The new staging system for ADD patients. Primarily based on MRI, juvenile ADD is divided into 5 stages according to the degree of condylar resorption and status of the disc. Stage 0: MRI shows basic shape of the disc with normal condylar shape and height, also the marrow of the subchondral bone has normal volume and quality. Stage 1: the disc has a basic shape and the condyle shows mild or local resorption, however, still with normal height. In addition, the marrow of the subchondral bone is partially reduced on the top. Stage 2: the disc has a basic shape and the condyle appears as moderately resorbed with reduced height, while the marrow of the subchondral bone becomes mildly reduced. Stage 3: the shape of disc is basic or distorted and the condyle has severe resorption. Also the marrow of the subchondral bone is moderately reduced. Stage 4: the conditions of disc and condyle are the same as stage 3, but the marrow of the subchondral bone is moderately reduced with inflammatory changes, severely reduced, or even totally absent (Table 3, Fig. 3). According to the shape and length of the disc, inflammatory changes in the condyle, and the presence of bone marrow in the condylar head, stage 3 and stage 4 were subdivided into sub-stages A and B (Table 3, Fig. 3).
At the same time, the Wilkes stages of all the patients were recorded based on Wilkes staging system 24 . Bone marrow changes are defined based on the degree of intensity of the bone marrow as seen on MRI in relation to the standard brain intensity (constant reference), as follows: Normal: the condyle bone marrow shows hyperintensity in relation to the brain intensity. Mild: the BM is of an equal intensity as the brain.
Moderate: the BM signals on MRI shows hypo intensity in relation to the standard brain intensity. Severe: Condyle BM shows no intensity at all.
Treatment protocols for ADD patients. According to our staging system, patients in stage 0, stage 1, stage 2 and stage 3A were suggested to undergo disc reposition treatments, including functional splint therapy and arthroscopic or open surgeries. As for patients in stage 3B, MRI was repeated after 6 to 12 months to determine whether condylar resorption had stopped. Once condylar resorption continued, it was treated as stage 4. In this stage, the condyle was routinely resected and replaced with a costochondral graft (CCG) or total alloplastic TMJ replacement. However all the ADD patients had their right to accept our treatment suggestion or give up the treatment protocols. No matter what selection they chose, All of them were required for regular follow-ups.
Evaluation of the outcomes of ADD. The outcomes of ADD were evaluated using clinical examination, a questionnaire about jaw function and pain, and imaging studies. In the present study, we only focused on the imaging of the TMJ. As for disc repositioning or no treatment patients, MRI was requested regularly. The following criteria were used to define the outcomes: www.nature.com/scientificreports www.nature.com/scientificreports/ 1. Excellent: There is normal disc position, with condylar regeneration. 2. Good: There is normal disc position with little or no condylar regeneration, but with no progression of resorption for disc repositioning patients; or there is no progression of condylar resorption for patients with no treatment. 3. Poor: There is recurrent disc displacement and/or progression of condylar resorption for disc repositioning patients; or condylar resorption continues for patients with no treatment.
As for condylar reconstruction cases, we focused on CT and panoramic radiographs. MRI was performed occasionally. The following criteria were used to define the outcomes: 1. Good: The graft is well incorporated, with little or no resorption. 2. Poor: There is resorption of the graft.
In the present study, excellent and good outcomes were considered as ameliorative, while poor was deemed as deteriorative.
Statistics analysis. Statistical analysis was performed with R version 3.3.2. The risk of deterioration was estimated with Cox proportional hazard regression modeling and patients without deterioration were censored at the time of their last available follow-up. The initial assessment was made using the univariate analysis, with   www.nature.com/scientificreports www.nature.com/scientificreports/ continuous variables evaluated as non-transformed, log-transformed and best-fit fractional polynomial transformations. Transformation did not significantly improve the model fit for any variable, so non-transformed data were used. Multivariate model development involved covariate assessment by statistical significance and biological/clinical importance 38 . Variables and interaction terms were entered into multivariate models if the univariate P value was <0.05 or of biological/clinical importance. For each added covariate, the likelihood-ratio (LR) test was used to evaluate the effect of removal on model fit. Covariates were included in the model if P < 0.05, the LR test indicated a significant improvement to model fit (P < 0.05) and/or there was an evidence of biological/clinical relevance. Model discrimination was assessed using Harrell's concordance index (c-index) 39 , which is similar to an area under the receiver operating characteristic curve but is suitable for the censored data. For the C-index, a value of 0.5 indicates no discrimination while 1.0 indicates perfect discrimination. For the final model, the C-index and 95% confidence interval (CI) were calculated with bootstrap resampling (100 replications) to reduce overfit bias. Model calibration was assessed with plots of the predicted versus observed probabilities of deterioration, using 100 bootstrap resample. The validation and calibration plots were produced with the R software "rms" package. All P values were two-sided, with P < 0.05 considered as a statistically significant.