TMJ ID is one of the most common temporomandibular disorders, or TMD. The term generally denotes an abnormal positional relationship of the articular disk to the mandibular condyle and the articular eminence. The IDs have been associated with characteristic clinical findings such as pain, joint sounds and abnormal jaw function.^1,2 With MRI techniques that depict the articular disk, the condition of ID has been considered to be an underlying mechanism in the pathogenesis of TMJ pain and dysfunction.^3,4 Mechanical disturbances were regarded as etiologic in the production of an imbalance between anabolic and catabolic processes, progressive cartilage degradation and secondary inflammatory components,⁵ while therapeutic procedures such as splint therapy,^6,7 arthrocentesis,^8,9 arthroscopic lysis and lavage^10,11 and arthrotomy^12,13 were used primarily to reduce TMJ loading and restore normal TMJ function and structure.

TMJ OA is a common noninflammatory joint disease that is characterized by a deterioration of the articular surfaces and a simultaneous remodeling of the underlying bone.^14,15 TMJ OA and ID are closely related. Several studies have supported the assertion that IDs are associated with the progressive development of radiographically detectable degenerative changes.^15–20 These changes include deviation in shape, surface irregularities, subchondral cyst formation, flattening, sclerosis and exophyte formation, and they are accepted as radiologic signs of TMJ OA.^15,20,21

TMJ effusion describes the MRI finding of a hyperintense signal within the joint space.^22–28 Westesson and Brooks²³ emphasized the importance of effusion in MRI of the TMJ and suggested that TMJ effusion represents an inflammatory response to a dysfunctional diskcondyle relationship. Sano and Westesson²⁵ compared retrodiskal tissue intensity in painful and painless TMJs and found a significant association between TMJ pain and increased signal intensity. In an article based on a sample of 123 TMJs, however, Adame and colleagues²⁸ failed to relate TMJ pain to TMJ effusion. They described TMJ effusion as being related to the MRI finding of ID and OA. Further, Murakami and colleagues²⁶ reported that the MRI finding of high-signal intensity in the closed locked TMJ did not directly relate to the presence of TMJ pain.

Most research has been conducted as simple paired tests of magnetic resonance imaging factors, which assumes an ‘all or none’ role for the factors and ignores the simultaneous contribution of multiple factors.

Several MRI studies have described edema and necrosis as bone marrow signal abnormalities of the TMJ.^29–35 A 1999 study analyzed biopsies from 50 mandibular condyles and found that edema occurs frequently without histologic evidence of necrosis, suggesting that edema may be a precursor to osteonecrosis as it is for other joints.³⁶ However, the significance of this finding in the TMJ is uncertain, and the reported etiologic concepts are controversial.

Most research has been conducted as simple paired tests of MRI factors; however, this assumes an "all or none" role for MRI factors and ignores the simultaneous contribution of multiple factors, which inevitably occur in biological systems. Further, most studies have relied on symptom-based data rather than a discrete disorder classification, while they also frequently have had inadequate control groups. We conducted this study to evaluate whether TMJ pain disorder subgroups may be related to MRI diagnoses of TMJ ID, OA, effusion and bone marrow edema, as well as to analyze whether common MRI features such as disk displacement, or DD; OA; effusion; and bone marrow edema may predict the presence of TMJ pain.

	MATERIAL AND METHODS

TOP MATERIAL AND METHODS RESULTS DISCUSSION CONCLUSION

 
Subjects. The study group consisted of 118 consecutively seen patients with TMJ pain and 46 consecutive patients who did not have TMJ pain; all of the patients were referred from medical practitioners and dentists in the community to the Orofacial Pain and Temporomandibular Disorder Clinic in the Department of Oral and Maxillofacial Surgery at the University of Innsbruck, Austria. This clinic is the primary referral center for TMD at the university, and it offers both conservative and surgical treatments. Patients who had reported pain and dysfunction of the temporomandibular region as their chief complaints were referred to the center for treatment. The group with TMJ pain included 102 female patients and 16 male patients, who had a mean age of 36.9 years (range 12–69 years). The group of patients who did not have TMJ pain (the control group) consisted of 31 female and 15 male patients, who had a mean age of 38.3 years (range 15–64 years). We informed the subjects about the study procedure and obtained their informed consent.

Criteria for including a patient with TMJ pain were the presence of a TMJ pain side-related, single clinical diagnosis of ID type I (DD with reduction); ID type III (DD without reduction); degenerative joint disease, or DJD; or capsulitis/synovitis, or C/S. Other criteria were a report of orofacial pain referred to the TMJ and the presence of TMJ pain during palpation, function or unassisted or assisted mandibular opening. Criteria for including a patient with no TMJ pain were the presence of at least one normal TMJ without a diagnosis of ID type I, ID type III or DJD; the absence of a TMJ with a diagnosis of ID type III on either side; and the absence of a TMJ with pain during palpation, function, and unassisted or assisted mandibular opening. After we applied the criteria, our study group of 118 subjects with painful TMJs included 35 ID type I TMJs, 39 ID type III TMJs, 18 DJD TMJs and 26 C/S TMJs. We also had 46 subjects with no TMJ pain and without a clinical disorder.

We clinically diagnosed TMJ disorders according to the Clinical Diagnostic Criteria for Temporomandibular Disorders, or CDC/TMD.³⁷ Criteria for each diagnostic group were specified as guidelines to be used in the clinical diagnostic process. Clinical evidence of absence of TMD was used in the absence of clinical inclusion criteria defining one of the TMD subgroups. We did not include patients who had been received a TMD diagnosis of myalgia or collagen vascular disease nor those with trauma histories.

To determine whether TMJ disorders subgroups were related to MRI diagnoses of TMJ ID, OA, effusion and bone marrow edema, the subjects underwent clinical and MRI investigation. One clinician (R.E.) clinically evaluated each subject; immediately after each clinical examination, the subject underwent MRI. The double-blind study ensured that the clinician and radiologist (A.R.) interpreted the clinical records and images independently without knowledge of the other’s results.

The clinical assessment consisted of a standardized evaluation of mandibular range of motion, joint pain and auscultation of joint sounds.

One clinician (R.E.) performed the clinical evaluation following a structured protocol.^38,39 Reliability scores were determined by administering the CDC/TMD in a blind test-retest method on a set group of 70 consecutive subjects, thereby allowing for intrarater comparison. We evaluated the statistical significance of the diagnostic percentage agreement between the interpretations using the statistical test. We found that the intraobserver reliability was satisfactory ( > 0.75) to excellent ( = 1.00) for all of the CDC/TMD diagnoses, with the majority having a value of 1.00. The intraobserver reliability regarding pain during palpation was lower than for other clinical items but was still acceptable ( > 0.70).

The clinical assessment consisted of a standardized evaluation of mandibular range of motion, joint pain and auscultation of joint sounds. We evaluated mandibular range of motion for maximum opening and lateral movements. Maximum opening was measured from a central maxillary incisor to the opposing mandibular incisor on a millimeter ruler. The clinician measured lateral movements relative to the maxillary midline with the teeth slightly separated. The TMJs were auscultated with a stethoscope, with the subject performing three openings and three lateral and protrusive movements. The joint sounds were described as single and reciprocal clicks.

The examiner assessed TMJ pain on palpation through bilateral manual palpation of the lateral aspect of the condyle. He then assessed the parameter of TMJ pain during unassisted mandibular opening by asking the subject to open his or her jaw as far as possible. The examiner performed assisted opening by applying force to the lower and upper incisors with his middle fingers and thumbs. The examiner recorded a positive pain score if a subject experienced a distinct painful sensation in the TMJ during the procedure. He assessed muscle pain as positive or negative using a bilateral manual palpation technique on the anterior, posterior and middle temporalis; the tendon of temporalis, the superficial and deep masseters, the lateral pterygoid; and the anterior and posterior digastric muscle.

The clinician assigned a diagnosis of myalgia if palpation produced a clear reaction from the subject (that is, if the subject experienced a distinctly tender or painful sensation of two or more on a severity scale of 0 to 3 in two or more muscle sites).

The TMJ ID type I subgroup was defined as subjects who had a click in the TMJ during vertical mandibular range of motion and lateral or protrusive excursion, as well as normal closing with or without clicking reproducible on two of three occasions.

Subjects in the TMJ ID type III subgroup were characterized by a history of a sudden reduction in mandibular opening, an unassisted mandibular opening of less than 35 mm and a mandibular opening with assistance increased by 3 mm or less. Subjects with a history of click had to show that the click disappeared and that they had sudden decrease in mouth opening.

Clinical criteria for a TMJ diagnosis of DJD included the presence of hard grating or crepitus during mandibular range of motion. We did not have access to tomographic films to evaluate the osseous structure of the TMJ.

Criteria for including a patient with a TMJ disorder of C/S were the presence of TMJ pain during palpation, function and assisted mandibular opening.

MRI. We carried out MRI with a 1.5 tesla MRI scanner and a dedicated circular-polarized transmit-and-receive TMJ coil. The data were collected on a 252 x 256 dots-per-inch matrix with a field of view of 145 mm, giving a pixel size of 0.60 x 0.57 mm.

With the subject in a supine position, we obtained 15 paracoronal slices and eight parasagittal slices of each TMJ using a turbo-spin-echo–proton density sequence (repetition time, 2,800 milliseconds; echo time, 15 ms) and a turbo-inversion-recovery-magnitude sequence (repetition time, 4,000 ms; echo time, 30 ms; inversion time, 150 ms) with thin 3-mm slices. Magnetic resonance, or MR, images were corrected to the horizontal angulation of the long axis of the condyle.

The study showed that subjects with temporomandibular pain and control subjects were associated with a high rate of disk displacement and osteoarthrosis.

Each subject received a nonferromagnetic intermaxillary device to achieve the different mouth opening positions. We made sequential bilateral T1- and T2- weighted MR images at the closed-mouth and the respective maximum open-mouth positions. We selected T1-weighted MR images for analysis of the disk-condyle relationship that depicted the disk, condyle, articular eminence and glenoid fossa. We defined normal disk position as the posterior band of the disk’s being at the superior, or 12 o’clock, position relative to the condyle, whereas we defined DD as the posterior band of the disk’s being in an anterior, anteromedial, anterolateral, medial or lateral position relative to the superior part of the condyle. We categorized diagnosis of the TMJ diskcondyle relationship as normal (absence of ID) and DD with and without reduction. We based our categorizations on the finding of a closed–mouth-related diagnosis of absence or presence of DD associated with or without an open–mouth-related interposition of the disk between the condyle and the articular eminence (Figure).^40,41

View larger version (211K): [in this window] [in a new window]   Figure. Bilateral closed–mouth-related magnetic resonance, or MR, images in a subject with left–side-related temporomandibular joint, or TMJ, pain. A–B. Left TMJ with presence of disk displacement, or DD; osteoarthrosis, or OA; and bone marrow edema. A. Sagittal MR image shows disk displaced anteriorly and fragmented (arrows), condyle with flattening and erosion and hyperintensity in bone marrow (arrowheads), indicating the edema pattern. B. Coronal MR image shows condyle with flattening and erosion and increased signal from bone marrow (arrowheads), indicating the edema pattern. C–D. Right TMJ with presence of anterolateral DD and absence of OA and bone marrow edema. C. Sagittal MR image shows disk (arrows) anterior to condyle. D. Coronal MR image showing lateral DD (arrows).

We identified joint effusion on the T2-weighted MR images as an area of high signal intensity in the region of the joint space. When more than one high-signal line was evident in at least two consecutive sections, we considered it to be positive for TMJ effusion.²³

We identified bone marrow edema by the presence of a hypointense signal on T1-weighted MR images and a hyperintense signal on T2-weighted MR images (Figure).³⁶

We determined reliability scores by administering the imaging criteria in a blind test-retest method on a set of 50 images in 25 randomly selected subjects, thereby allowing for intrarater comparison. Statistical significance of the diagnostic percentage agreement between the interpretations was evaluated using the statistical test. The intraobserver reliability was satisfactory ( > 0.75) to excellent ( = 1.00) for all of the MRI diagnoses.

Data analysis. We performed ² analysis to analyze the relationship between the clinical subgroups TMJ ID type I, ID type III, DJD and C/S and the MRI findings of TMJ ID, OA, effusion and bone marrow edema. We performed a multiple logistic regression analysis to simultaneously assess the relative odds of each potential MRI diagnosis, adjusted for age and sex. The outcome was always presented as TMJs with no pain versus TMJs with pain.

We used odds ratios, or OR, to describe the proportionate risk that a subject with a certain MRI feature could belong to the TMJ pain disorder group. We defined a significant OR as an upper and lower 95 percent confidence interval not containing the value of zero. For the OR to be clinically relevant or even clinically noticeable, we assumed that it would need to be larger than 2. We set significance at P .05. For all statistical analysis, we used a specialized software package (SPSS 7.5.2G, SPSS, Chicago).

	RESULTS

TOP MATERIAL AND METHODS RESULTS DISCUSSION CONCLUSION

 
The MRI factor of DD with reduction was the most prevalent in the ID type I group (37.1 percent) and only slightly less prevalent in the control group (32.6 percent) (Table 1, page 710). DD without reduction occurred most frequently in the DJD group (83.3 percent) and ID type III group (79.5 percent) and occurred less frequently in the control group (19.6 percent). Comparing the TMJ pain group–related data with those of the control group, we found that the results of ² analysis showed a significant relationship between the clinical finding of TMJ pain and the MRI diagnosis of TMJ ID type for the ID type I group (P = .041), the C/S group (P = .024), the ID type III group (P = .000) and the DJD group (P = .000) (Table 1).

Effusion was relatively common in all diagnostic groups (43.6–72.2 percent) and only slightly less in the control group (30.4 percent) (Table 1). Using ² analysis, we found a significant relationship between the clinical finding of TMJ pain and the MRI diagnosis of TMJ effusion for the DJD group (P = .002) (Table 1).

Bone marrow edema occurred most frequently in the ID type III group (48.7 percent), was minimally represented in the control group (4.4 percent) and was reduced in the other diagnostic groups (20.0–33.3 percent) (Table 1). Using ² analysis, we found a significant relationship between the clinical finding of TMJ pain and the MRI finding of TMJ bone marrow edema for the ID type I group (P = .026), the C/S group (P = .015), the DJD group (P = .002) and the ID type III group (P = .000) (Table 1).

Of the MRI findings considered in the multiple logistic regression analysis, OA and effusion dropped out as nonsignificant in the test groups compared with the control group. The number of factors remaining in the regression equations was two for the ID type III group (DD without reduction and bone marrow edema) and one each for the ID type I group (OA) and DJD group (DD without reduction) (Table 2).

View this table: [in this window] [in a new window]   TABLE 2 SUMMARY OF LOGISTIC REGRESSION ANALYSIS FOR MRI* VARIABLES PREDICTING TMJ PAIN IN SUBJECTS WITH SPECIFIC CLINICAL TMJ DISORDERS.

	DISCUSSION

TOP MATERIAL AND METHODS RESULTS DISCUSSION CONCLUSION

 
The MRI results in our study showed that subjects with TMJ pain and control subjects were associated with a high rate of TMJ DD (75.8 percent versus 52.2 percent) and OA (85.7 percent versus 76.1 percent) (Table 1). This observation compares favorably with the results of other studies that reported prevalences of TMJ DD in TMJ disorder populations ranging from 77 to 89 percent,^41,42 while those for asymptomatic subjects have ranged from 30 to 39 percent.^41,42 Regarding TMJ OA, the findings correspond to other studies that describe frequencies of imaging signs of TMJ OA in TMJ disorder patients as ranging from 11.4 to 58 percent^3,20,43,44 and in patients with asymptomatic TMJs as ranging from 50 to 90 percent.^43,45,46 MRI features of DD and OA, traditionally believed to be influential, contributed only minor amounts to the change in risk in our multiple factor analysis (Table 2). We assume that OA did not enter the regression model for the ID type III group because of the strength of the variable bone marrow edema and DD without reduction (Tables 1 and 2), which explained much of this feature. The relationship of OA to ID type I was very weak (1:5.6) and may have made only a minimal contribution to the clinical picture (Table 2). Based on our study and supported by other studies, MRI features of DD and OA may not be considered the unique and dominant factors in the definition of TMJ pain disorders.^20.47.48 The contribution of these MRI features, however, was not zero, and the minor elevations in OR may indicate that they are probably making some contribution biologically and should not be ignored.

The MRI results in our study showed that subjects with TMJ pain were associated with a high rate of bone marrow edema (range, 20.0–48.7 percent). This observation is inconsistent with those of previous studies, which showed prevalence rates of bone marrow edema in pain and dysfunction patients ranging from 2.5 to 6.5 percent.^31,33–35 The results, however, may be not directly comparable, as the studies were conducted without the use of strict inclusion and exclusion criteria to define homogeneous TMJ disorder groups. Bone marrow edema had the highest OR and was predominantly associated with the ID type III diagnosis (Table 1 and Table 2). For bone marrow edema to be an etiologic factor for a subject with ID type III, some evidence of it should be present in some of the preceding DD without reduction instances that eventually would go on to develop bone marrow edema. The prevalence of DD without reduction in subjects with ID type III was 79.5 percent and carried an increased OR for that group (OR = 10.2:1). Therefore, we hypothesized that the tendency of subjects in the ID type III group to develop bone marrow edema was a consequence of and secondary to DD without reduction instead of being part of the TMJ derangement etiology. Studies also supported an association of bone marrow edema with the MRI diagnosis of TMJ DD without reduction.^33–35

Using magnetic resonance imaging to make diagnoses may not be considered the unique or dominant factor in defining temporomandibular joint disorder populations.

DD with reduction and effusion did not persist in any of the regression equations. Both variables were not associated with either subjects with TMJ pain or the control subjects. The OR that a subject in the study would have been classified as a TMJ pain patient based on DD with reduction or effusion was minor and would not have attained a 2:1 risk level in the multiple regression analysis for any of the TMJ pain groups. Because the contribution of DD with reduction is minor, surgical elimination of DD with reduction through arthroscopy or arthroplasty is difficult to support. Even in the presence of symptoms, surgical intervention could be inappropriate, because the painful clicking, locking or both joint may be the consequence of osseous, ligamentous, synovial or both disorders rather than being a disk-related etiologic factor.

For a clinically perceptible influence to be achieved, we hypothesized that an MRI feature would need to have at least a 2:1 mean OR for the disorder. Only two MRI conditions reached this threshold, namely DD without reduction and bone marrow edema. These features should remain part of the investigation’s database of patients with TMJ pain, but their clinical value must not be overstated. The utility of these features is limited, because these parameters still described only a small proportion of the variation in each pain group and did not explain the majority of the pain occurrences. A new definition of "normal" may arise, namely, those features without a significant elevated risk for pain. Common to both subjects with TMJ pain and control subjects were DD with reduction, OA and effusion (Table 1). Therefore, these parameters alone may not define either test or control subjects.

Further multiple-factor studies are warranted, including those that encompass additional morphological and psychological variables. It may be presumed that additional factors with increased relative risk of developing the disorder may be revealed, but it may be considered doubtful that a single etiologic factor for pain dysfunction problems may appear. Further, regression analysis as in our study gives no information on etiology and states only the empirical relative odds of TMJ disorder. From a methodological point of view, a clearer answer to the etiologic contribution of MRI factors to TMJ disorders may have to await an incidence study rather than the present case-control study model.

Observer variations are common in any clinical experience and could influence significantly the diagnostic process. The level of observer performance is dependent on several factors such as training observers, image quality and the definition of specific criteria for interpretation. Our study used high-quality MR images and included MRI variables interpreted according to well-defined criteria. The intraobserver reliability in reporting MR images of the TMJ was within acceptable limits for diagnostic study. Interobserver reliability, however, has not been measured. Thus, we acknowledge that some of the variation of OA, effusion or both was probably a consequence of overrating. Consequently, it is possible that the relevance of these features to selected disorder groups may have been underestimated.

The possibility of rater bias by the radiologist who assessed the MRI variables also must be considered. With regard to the diagnosis of bone marrow edema, the radiologist may be influenced by the state of the disk and other joint components. The most effective way to control for this type of error is to blind the rater in some way; however, as most radiological measurements are observational, such a technique often is unreasonable. The major protections against rater bias are to develop grading criteria that are as objective as possible, to train the observers how to use the instruments and to document reliability across raters.

	CONCLUSION

TOP MATERIAL AND METHODS RESULTS DISCUSSION CONCLUSION

 
While the contribution of MRI variables to the TMJ pain subgroups was not zero, most of the variation in each TMJ pain population was not explained by MRI parameters. Thus, using MRI to make diagnoses may not be considered the unique or dominant factor in defining TMJ disorder populations. Therapy for patients with TMJ based on the evaluation of concomitant morphological abnormalities, whether prophylactically or as treatment for TMJ disorders, may be unwarranted.