The aims of the study were to evaluate the diagnostic accuracy and utility of the mean region of interest (ROI) and mean and maximum volume of interest (VOI) analysis methods for 99m Tc MDP SPECT scintigraphy in the diagnosis of active unilateral condylar hyperplasia (UCH). Inactive UCH ( n = 43) and active UCH ( n = 8) patients, and patients without condylar hyperplasia (controls, n = 41) were analyzed. Inter-observer agreement was good for all methods. Condylar uptake was not normally distributed, with a longer right tail in UCH patients compared to control patients. Receiver operating characteristic curve analysis indicated that the ROI method was slightly superior to both VOI methods for the diagnosis of active UCH (area under the curve = 0.866, 0.811, and 0.817, and J = 0.642, 0.596, and 0.573, respectively). The ‘traditional’ 55% cut-off value proved optimal for ROI and mean VOI methods, but a cut-off of 56.125% was optimal for maximum VOI. Sensitivity was 88% for all three methods using these cut-off values, while specificity was 77%, 65%, and 70% for mean ROI, mean VOI, and maximum VOI, respectively. These results indicate that corrective surgery for negative scan patients can be performed without delay, with an error rate of only 3%, but not in positive scan patients.
Condylar hyperplasia is a pathology of the temporomandibular joint (TMJ) resulting in a progressive unilateral non-neoplastic growth, involving both the size and configuration of the neck and the mandibular condyle. Condylar hyperplasia is a unilateral condition characterized by generalized enlargement of the condyle, the condylar neck, the ramus and body of the mandible, leading to facial asymmetry and occlusal alterations. Unilateral condylar hyperplasia (UCH) usually presents as an excessive growth of one condyle. It is the most common postnatal growth abnormality of the TMJ. Women tend to be predisposed to UCH, with a female to male ratio of approximately 2:1. However, large heterogeneity between populations has been observed. Although it has previously been suggested that there is also a sex difference in the laterality of UCH, this was not substantiated in a recent meta-analysis. Left-sided UCH seems to be slightly more prevalent than right-sided UCH, regardless of sex.
The main complaint of UCH patients is progressive facial asymmetry, but almost one-third of the patients complain of swelling on the contralateral side, pain, and dysfunction. Therefore, attention must be paid to facial asymmetry even when it is not among the patient complaints.
The treatment of mandibular asymmetry is primarily surgical, with or without orthodontics, and consists of two types of intervention depending on the condylar activity. A high condylectomy of the affected side is indicated to limit progressive asymmetry during the active phase of UCH. Secondary correction by mandibular or maxillary osteotomies or both (orthognathic surgery) is appropriate to correct any residual occlusal and facial asymmetry. However, if orthognathic surgery is performed while condylar activity persists, then further asymmetry may develop. Consequently, accurate assessment of the cessation of excess activity in the condyle is warranted. Conversely, condylectomy in a ‘burnt-out’ condyle causes undue and unnecessary disruption of the TMJ and can affect occlusion.
Cisneros and Kaban were the first to use bone scintigraphy to study patients with mandibular asymmetry in 1984. Bone scintigraphy offers an instant method of comparing the differential activity between normal and abnormal condyles, and this reflects the relative growth rates at the time of the investigation. Studies published in the literature, most of them using planar imaging, have demonstrated a 5% to 12% difference in bone activity between normal and abnormal condyles.
However, there is no standard method for diagnosing a patient with UCH by skeletal scintigraphy. No uniform method for quantification of bone activity was used in any of the planar or single photon emission computed tomography (SPECT)-based studies included in a recent review and meta-analysis. Although many studies have been published since 1984, only seven of these were found to be sufficiently detailed and large enough to be included in the recent meta-analysis, and SPECT was used in only three of these publications.
Only one prospective study using SPECT scintigraphy on a normal population has been published. Kajan et al. studied 38 patients, ranging in age from 13 to 34 years, who were undergoing skeletal scintigraphy for a variety of conditions. They concluded that the variation in growth activity in normal right and left mandibular condyles was less than 6.2%, which is lower than the 10% difference that is widely employed.
The lack of a standardized and detailed method for skeletal scintigraphy in these patients and the lack of two-by-two contingency tables limit the ability to assess the diagnostic accuracy of the test and to use it effectively in UCH patients.
The objectives of this study were to describe and evaluate three different quantification methods, to describe and compare a population without condylar hyperplasia to a population of inactive UCH patients, and to evaluate the usefulness of SPECT scintigraphy in the management of UCH patients.
Normal population and TMJ patients
Fifty-one patients (24 male, 27 female) with a mean age of 21.5 years (range 8–66 years) suspected of having mandibular condylar hyperplasia were referred for bone scintigraphy during the years 2008 to 2011. These patients underwent a combined total of 60 scans (seven patients had two scans and one patient underwent three scans). The clinical records of all of these patients were assessed by a maxillofacial surgeon.
Forty-one patients (23 male, 18 female) with a mean age of 33.3 years (range 18–73 years) who were referred to the nuclear medicine department for bone scintigraphy for reasons unrelated to a TMJ pathology agreed to participate in the study (control population). All of these patients provided signed informed consent and completed a questionnaire designed to ensure that the patient had no TMJ pathology. Six of these patients were also examined by a maxillofacial surgeon according to surgeon availability. No TMJ abnormality was found in any of the patients.
This study was approved by the institutional review board and all normal participants signed an informed consent agreement.
The patient evaluation included the administration of a questionnaire to collect demographic information and obtain a comprehensive history, including primary complaints, initial symptoms, duration of symptoms, presence of joint noise, limitation in mouth opening, and prior treatment. Each patient self-assessed his or her level of pain and the extent of dysfunction using a visual analogue scale (VAS) and indicated the location of the pain on a facial diagram.
Deviation was evaluated in the transverse, vertical, or both planes. The evaluation was based on clinical signs, including occlusion, occlusal plane, and deviation of the mandibular midline. The clinical examination included the determination of maximum mouth opening, range of lateral and protrusive mandibular movements, characteristics of the limitation in jaw movement, determination of joint noise on palpation, and the evaluation of pain on palpation of the head and neck muscles and both TMJs. The severity of the occlusal plane inclination was evaluated by the angle between the occlusal plane and the inter-pupil line. Deviation of the dental midline, cross-bite, and open-bite were also recorded.
The radiological evaluation included preoperative transpharyngeal and transcranial radiographs of the TMJ in closed-mouth and open-mouth positions, along with panoramic and cephalometric X-rays in anterior–posterior and lateral views. The condylar head was classified as normal, enlarged, deformed, or enlarged and deformed, and the condylar neck was classified as normal, elongated, or enlarged.
TMJ SPECT imaging
Skull bone SPECT scans were obtained using either an Infinia Hawkeye 4 (GE Healthcare, Tirat Carmel, Israel) or Varicam Hawkeye (GE Healthcare, Tirat Carmel, Israel) dual-head gamma camera equipped with a lower-energy general-purpose parallel hole collimator using 740 MBq of technetium 99m methylene diphosphonate ( 99m Tc MDP) for adults. One hundred and twenty projection images were acquired over 360 degrees during 20 s per projection as a 128 × 128 matrix. Transaxial, coronal, and sagittal tomograms were reconstructed using a Butterworth filter (power 10, critical frequency 0.48) and ordered-subset expectation maximization (OSEM) iterative reconstruction (two iterations, 10 subsets).
Interpretation of SPECT images
All images were assessed in a blinded manner; the interpretation was performed by two independent nuclear medicine physicians (ST and MK). The volume of interest (VOI) of 1.8 cm 3 , including the TMJ, was measured on a Xeleris 3 workstation (GE Healthcare, Tirat Carmel, Israel), and both the mean and maximum values were recorded. A fixed size region of interest (ROI) of 1.7 cm 2 was drawn on the summed transaxial slices that included the right TMJ and the mean value was recorded (only by ST). The ROI was mirrored and placed on the left side to assure a fixed size.
The statistical analysis was performed using IBM SPSS Statistics version 21.0 software (IBM Corp., Armonk, NY, USA). The control population and the inactive UCH population were compared using a one-tailed Mann–Whitney U -test. For all other analyses, appropriate two-tailed tests were used. Pearson’s correlation coefficient ( r ) and Cohen’s kappa test with a cut-off value of 55% condylar uptake were used to assess inter-observer agreement for each measurement method. The area under the curve (AUC) of the receiver operating characteristic (ROC) curve, Youden’s index ( J ), Cohen’s kappa, and the odds ratio (OR) were calculated to compare scintigraphy results against active or inactive UCH. A P -value of less than 0.05 was considered statistically significant.
Inter-observer agreement in VOI calculation methods
The summed ROI method, at least for a fixed ROI size, has been reported previously as having very good inter-observer agreement. However, only one article describing the use of VOIs could be identified in the literature, and the interpretation was performed by a single observer. Therefore, inter-observer agreement was evaluated by assessing Pearson’s correlation coefficient ( r ) for both the mean VOI method ( r (48) = 0.913, P < 0.001) and maximum VOI method ( r (48) = 0.948, P < 0.001) in inactive UCH patients. These results were highly significant; however, Pearson’s correlation coefficient ( r ) only indicates that there is a linear relationship between observers.
In order to further assess inter-observer differences, the difference between the two observers for each method was also plotted against the mean of the measurements of the two observers (Bland–Altman plots; Fig. 1 ). The difference values were largely clustered around the zero line for both methods. However, for a minority of the patients, large differences were noted, and the impact of such differences on the decision regarding whether a condyle was active or not cannot be ascertained from Bland–Altman plots. Therefore, Cohen’s kappa was calculated using the generally accepted cut-off value of 55% for the active condyle. The kappa score was 0.662 for mean VOI and 0.877 for maximum VOI ( Table 1 ).
|Method||Pearson’s r||P -value||Cohen’s kappa||P -value|
These results, together with those reported in the previous literature, indicate that there is very good inter-observer agreement for all three of the methods used in this study.
Comparison of the normal condyle population and the UCH populations
Forty-one patients without condylar hyperplasia referred to the department for unrelated indications were analyzed (control population). The lack of condylar hyperplasia was verified using questionnaires; some patients also underwent a physical examination depending on the availability of a maxillofacial surgeon.
The condylar uptake values did not seem to follow a normal distribution ( Fig. 2 ). The distribution was skewed with a long right tail, which was even longer in the inactive UCH population. Both the median and the interquartile range were increased in the inactive UCH population compared to the control population ( Table 2 ).
|Method||Normal||Inactive UCH||P -value|
|Mean ROI||51.1 (50.6–52.2)||52.2 (51.1–54.7)||0.003|
|Mean VOI||52.0 (50.7–53.9)||53.0 (51.1–56.7)||0.041|
|Max VOI||53.3 (51.5–54.6)||53.7 (50.9–57.2)||0.240|
The statistical analysis was performed for all three methods using the non-parametric Mann–Whitney U -test ( Table 2 ). In patients with inactive UCH, the percentage maximum activity of the abnormal condyle was significantly higher than the corresponding condyle activity in the control group for both the mean ROI ( P = 0.003) and the mean VOI ( P = 0.041) methods, but not for the maximum VOI method ( P = 0.240). This is in accordance with the results of AlSharif et al., who also found uptake in the inactive UCH population to be significantly increased with almost all of the methods utilized.
It should be noted that if the traditional 55% cut-off value is used, the resulting false-positive rate for the mean ROI, mean VOI, and maximum VOI methods is 23.3%, 34.9%, and 39.5%, respectively.
The percentage activity of the active condyle was significantly higher in the active UCH population ( n = 8) than in the inactive UCH population ( n = 43) for all methods, as expected ( P = 0.001, 0.006, and 0.004 for the mean ROI, mean VOI, and maximum VOI methods, respectively).
Comparison of the three methods in the diagnosis of active UCH
The sensitivity and specificity of the three methods were calculated. The ROC curve provides a useful tool to compare the different methods of analysis. The ROC curves ( Fig. 3 ) illustrated that all three methods can help in the diagnosis of active vs. inactive UCH, although the mean ROI method (AUC = 0.866, J = 0.642) consistently provided slightly better results than both the mean VOI (AUC = 0.811, J = 0.596) and maximum VOI (AUC = 0.817, J = 0.573) methods ( Table 3 ). This interpretation is also supported by calculating Cohen’s kappa and the OR using a cut-off value of 55% for all methods.