Is Single-Photon Emission Computed Tomography/Computed Tomography Superior to Single-Photon Emission Computed Tomography in Assessing Unilateral Condylar Hyperplasia?
,
Open Access
Article InfoArticle Outline
Purpose
Single-photon emission computed tomography (SPECT) with technetium-99m diphosphates plays an important role in assessing unilateral condylar hyperplasia (UCH). The aim of this study was to evaluate whether quantification methods of SPECT plus CT (SPECT/CT) based on precise region-of-interest (ROI) drawings made under the guide of CT images were more accurate than conventional SPECT methods in the assessment of UCH growth.
Materials and Methods
This study is a nonblinded retrospective case series. Patients with UCH who had undergone SPECT/CT were enrolled. CT images were used to guide ROI drawings around the anatomic contour of the affected and contralateral condyles on SPECT/CT images versus fixed ROIs on conventional SPECT images. Mean and maximum values within the ROIs were recorded to compute percentile ratios. Sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), and receiver operating characteristic (ROC) curves were calculated separately for SPECT-based methods (SPECTaver, SPECTmax) and SPECT/CT methods (SPECTCTaver, SPECTCTmax). The area under the ROC curve of each method was calculated and compared pairwise.
Results
Fifty-six patients (30 patients with progressive and 26 patients with nonprogressive mandibular asymmetry) were evaluated. SPECTmax had the highest sensitivity of 83.3%, followed by SPECTCTmax, SPECTaver, and SPECTCTaver. In contrast, SPECTaver, SPECTCTmax, and SPECTmax had similar specificities, PPVs, and NPVs. Nonetheless, SPECTCTaver had the lowest specificity, PPV, and NPV among all methods. ROC analysis also showed similar diagnostic performances among SPECTaver, SPECTmax, and SPECTCTmax (P > .05) and poorer diagnostic performance of SPECTCTaver compared with the other 3 methods (P < .05).
Conclusions
The method of using ROIs drawn around the contour of the condyle on SPECT/CT images does not show improved accuracy over conventional SPECT-fixed ROI methods in assessing UCH.
Asymmetric facial deformities can be caused by condylar hyperplasia (CH), defined by the excessive growth of the unilateral condyle. CH characterized by excessive bone growth usually occurs unilaterally in growing patients, especially during adolescence. Occlusal discrepancies and temporomandibular joint (TMJ) disorders have been associated with facial asymmetry in many cases. Women present with a higher prevalence than men.1 However, the etiology and pathogenesis of CH are not clear, and genetic factors, trauma, or hormonal conditions can contribute to the disease.2
Multiple classification systems have been proposed to better characterize the pathology. Obwegeser and Makek3classified CH into 3 categories based on asymmetry and predominant growth vector. Type 1 was defined as hemimandibular elongation, type 2 was defined as hemimandibular hyperplasia, and type 3 was defined as a combination of types 1 and 2. Wolford et al4 updated the classification system by including the pathologies causing CH. Four different categories based on clinical imaging, growth, and histologic characteristics were developed to provide optimal treatment to patients.
The assessment of growth status is important to treatment protocols. Treatments differ considerably according to the affected structures, the patient's age, the severity of the asymmetry, and the disease stage. The diagnosis of condyle growth is currently made from a combination of facial, intraoral, and radiographic findings over a period of time.2, 5 Technetium-99m methylene diphosphonate (99mTc-MDP) bone scintigraphy has been considered a useful tool to detect active CH. 6 However, improved resolution can be obtained with single-photon emission computed tomography (SPECT),7, 8, 9 because previous studies have shown that the SPECT technique is more sensitive than planar scanning for detecting abnormal condylar uptake. A meta-analysis conducted by Saridin et al6 calculated a sensitivity of 0.90 for SPECT and a sensitivity of 0.71 for planar bone scans. Furthermore, with the addition of computed tomography (CT) to SPECT, functional and morphologic information can be acquired.10
Hybrid SPECT integrated with multidetector CT (SPECT/CT) offers detailed anatomic information overlaid on the radioactivity uptake image, with the CT portion of SPECT/CT being used to correct attenuation and evaluate the anatomic change of the condyles. Using the CT component of the SPECT/CT images to guide anatomic contouring, precise regions of interest (ROIs) are drawn over the condyles on the image slice, thereby decreasing interobserver variations in ROI placement. A case report found that precise ROIs drawn under the guide of anatomic contours of condyles on CT images of SPECT/CT scans can improve the diagnostic accuracy of actively hyperplastic condyles.11 However, no study has validated this claim with a larger study population.
Is SPECT/CT with precise ROI delineation under the guide of anatomic structure CT images superior to SPECT in the diagnosis of active CH? This study evaluated and compared the diagnostic performance of SPECT/CT based on ROI delineation guided by anatomic CT structure images of condyles with conventional SPECT for the detection of active CH.
Materials and Methods
Patients and SPECT/CT Protocols
The present study was nonblinded, retrospective, and approved by the institute's ethics committee. All procedures performed in this study involving human participants were in accordance with the ethical standards of the institutional and national research committees and with the 1964 Declaration of Helsinki and its later amendments or comparable ethical standards. Informed consent was obtained from each participant included in the study. Patients with suspected progressive asymmetry of the mandible who were referred for regional SPECT/CT for the assessment of active CH from July 2012 through January 2013 were evaluated retrospectively. Inclusion criterion was lower facial plane asymmetry that was evaluated clinically and radiologically by experienced maxillofacial surgeons. Patients with previous trauma or surgery to the TMJ, previous mandibular fractures, neoplastic pathology of the TMJ, and systemic diseases or congenital conditions that could affect the TMJ were excluded from the evaluation. The final diagnosis was derived from clinical studies, serial conventional imaging, and clinical and imaging follow-up.
Patients were intravenously injected with 99mTc-MDP 555 to 851 MBq (15 to 23 mCi) for imaging, depending on their body weight (14.8 MBq/kg). Four hours later, images were acquired on a hybrid SPECT/CT dual-head gamma camera (Inifinia Hawkeye 3, GE Healthcare, Chicago, IL) equipped with parallel-hole, low-energy, high-resolution collimators. The photo peak was set at 140 keV and a 20% symmetrical window, and emission data were acquired in the supine position with a 128 × 128 matrix with circular obits. Each head circulated 180° with 30 stops, and each stop was at 6°. Transaxial, coronal, and sagittal tomograms were reconstructed using a Butterworth filter (power, 10; critical frequency, 0.48 cycles/pixel) and ordered-subset expectation maximization iterative reconstruction (2 iterations, 10 subsets). CT images were obtained using tube voltage of 140 keV, tube current of 2.5 mA, tube rotation velocity of 2.6 rpm, helical scan per slice of 4.42 mm, pitch of 1.9, and matrix of 512 × 12. Attenuation correction was applied to SPECT/CT images using CT-based attenuation maps.
Interpretation of SPECT and SPECT/CT Images
Reconstructed images were assessed by experienced nuclear medicine physicians. For SPECT/CT images, precise ROIs were drawn over the condyles under the guide of anatomic contouring on the CT component of the SPECT/CT images. Contoured ROIs of the bilateral condyle were drawn on the transaxial CT image and mirrored on the SPECT image (Fig 1), and the process was repeated on adjacent transaxial image slices and centered on the image slice with the highest radiotracer count; mean and maximum values were recorded. Mean radiotracer count ratios (SPECTCTaver) and maximum radiotracer count ratios (SPECTCTmax) between the condyles were calculated. For SPECT images, ROIs were drawn over 1 condyle on the transaxial image and then copied and placed on the contralateral condyle to ensure a fixed ROI size on the SPECT image (Fig 2). Mean radiotracer count ratios (SPECTaver) and maximum radiotracer count ratios (SPECTmax) between the condyles were calculated. Condylar uptake of at least 55%, generating differences of at least 10% between condyles, was considered active unilateral CH (UCH).12 All image interpretations and ROI analyses were performed on a Xeleris 3 workstation (GE Healthcare).
Figure 1
Technetium-99m methylene diphosphonate single-photon emission computed tomographic plus computed tomographic images of the skull base, including the mandibular condyles, in the transaxial plane. Regions of interest (circled areas) were drawn precisely around the condyles on A, unenhanced computed tomographic views and were mirrored to B, single-photon emission computed tomographic views and C,fused single-photon emission computed tomographic plus computed tomographic views.
Figure 2
Technetium-99m methylene diphosphonate single-photon emission computed tomographic image in transaxial plane. Fixed regions of interest (circled areas) were drawn over the bilateral condyles.
Statistical Analysis
Data are presented as mean ± standard deviation. Continuous variables were analyzed using Student t test. Categorical variables were analyzed using Pearson χ2 test. Sensitivity, specificity, and positive and negative predictive values (PPV and NPV) were calculated separately for SPECT/CT and SPECT. The area under the curve (AUC) of the receiver operating characteristic (ROC) curve was calculated to compare the diagnostic performance of the different ROI methods. All statistical analyses were performed with IBM SPSS Statistics 21.0 (IBM Corp, Armonk, NY) or MedCalc 18.11 for Windows (MedCalc Software, Ostend, Belgium). A P value less than .05 was considered statistically significant.
Results
Patient Demographics
Fifty-six patients (24 men and 32 women) were retrospectively evaluated. Patient demographics are presented in Table 1. Patients' mean age was 19.8 years (range, 15 to 28 yr). Ten patients (18%) had chin deviation toward the left and 46 patients (82%) had chin deviation toward the right. Based on the reference standard, 30 patients had active CH and 26 had stable CH. No significant differences between the active and stable CH subgroups were found for age (active CH, 19.54 ± 2.18 yr; stable CH, 20.1 ± 2.47 yr; P = .37 > .05) or gender (active CH, 14 men and 16 women; stable CH, 10 men and 16 women; P = .53 > .05).
| Active CH | Inactive CH | |
|---|---|---|
| Age | 19.54 ± 2.18 | 20.1 ± 2.47 |
| Gender | ||
| Male | 14 | 10 |
| Female | 16 | 16 |
| Total | 30 | 26 |
Abbreviation: CH, condylar hyperplasia.
Diagnostic Performance of SPECT/CT and SPECT-Based Methods
The sensitivity, specificity, PPV, and NPV of SPECT- and SPECT/CT-based methods are listed in Table 2. Of the 4 methods, SPECTmax had the highest sensitivity of 83.3%, followed by SPECTCTmax with 76.7%, and SPECTaver and SPECTCTaver had the same sensitivity of 66.7%; specificities of SPECTaver, SPECTCTmax, SPECTmax, and SPECTCTaver were 88.5, 84.6, 80.8, and 61.5%, respectively. The results also showed similar PPVs for SPECTaver (87.0%), SPECTCTmax (83.3%), and SPECTmax (85.2%), whereas SPECTCTaver showed the lowest PPV (66.7%) and NPV (61.5%) among all methods evaluated.
| Sensitivity, % | Specificity, % | PPV, % | NPV, % | |
|---|---|---|---|---|
| SPECTCTaver | 66.7 | 61.5 | 66.7 | 61.5 |
| SPECTCTmax | 76.7 | 84.6 | 85.2 | 75.9 |
| SPECTaver | 66.7 | 88.5 | 87.0 | 69.7 |
| SPECTmax | 83.3 | 80.8 | 83.3 | 80.8 |
Abbreviations: NPV, negative predictive value; PPV, positive predictive value; SPECTaver, mean radiotracer count ratios for single-photon emission computed tomography; SPECTCTaver, mean radiotracer count ratios for single-photon emission computed tomography plus computed tomography; SPECTCTmax, maximum radiotracer count ratios for single-photon emission computed tomography plus computed tomography; SPECTmax, maximum radiotracer count ratios for single-photon emission computed tomography.
ROC Curve Analysis
ROC curve analysis showed that the SPECTaver, SPECTmax, and SPECTCTmax curves were similar and closer to the left upper quadrant, which indicates these 3 methods had a better diagnostic performance than SPECTaver (Fig 3). The AUCs were 0.899 for SPECTaver, 0.896 for SPECTmax, 0.864 for SPECTCTmax, and 0.720 for SPECTCTaver. No significant difference was found among SPECTaver, SPECTmax, and SPECTCTmax (P > .05); however, the AUC of SPECTCTaver showed poorer diagnostic performance compared with the other 3 methods (P < .05).
Figure 3
ROC curves of different analysis methods. ROC, receiver operating characteristic; SPECTaver, mean radiotracer count ratios for single-photon emission computed tomography; SPECTCTaver, mean radiotracer count ratios for single-photon emission computed tomography plus computed tomography; SPECTCTmax, maximum radiotracer count ratios for single-photon emission computed tomography plus computed tomography; SPECTmax, maximum radiotracer count ratios for single-photon emission computed tomography.
Discussion
The purpose of this study was to evaluate whether the quantification based on SPECT/CT was superior to that based on SPECT in the diagnosis of active CH. The results showed that SPECTaver, SPECTmax, and SPECTCTmax had similar sensitivity, specificity, and AUC of ROC curves. However, SPECTCTaver showed poorer diagnostic performance compared with the other 3 methods. The results did not show better diagnostic performance of quantifications with ROIs drawn based on the anatomic contours of condyles with SPECT/CT than that with conventional SPECT in the diagnosis of active CH.
The results also showed that quantification methods using maximum count per pixel showed higher sensitivity compared with quantification methods using average counts per pixel with SPECT or SPECT/CT in the diagnosis of active CH.
The mean age of the present study population was 19.8 years (range, 15 to 28 yr), which is consistent with previous studies finding that UCH usually affects teenagers and young adults 10 to 30 years old.2, 5, 6 In the present study, the number of female patients was 32 (57%), which also is consistent with a systematic review that found a higher incidence of CH in female than in male individuals, according to a meta-analysis (n = 275 patients) conducted by Raijmakers et al,1 in which 64% of patients with CH were women (95% confidence interval, 0.58-0.70).
In the present study, of all 4 methods evaluated, SPECTmax had the highest sensitivity (83.3%) and NPV (80.8%) for detecting active UCH. The method also had relatively high specificity (80.8%) and PPV (83.3%). The finding is similar to previous reports using the maximum count per pixel of ROI methods in SPECT.10, 13 In a prior study performed with the maximum count per pixel of ROI on SPECT images of 33 patients, investigators found that the sensitivity, specificity, PPV, and NPV were 82.4, 81.3, 82.4, and 81.3%, respectively.13 The advantage of using the ROI with the maximum count per pixel is the relative independence of the position or size of the ROI; hence, it has good reproducibility, which allows for easier drawings.14 SPECTCTmax had the second highest sensitivity (76.7%) and NPV (75.9%) and slightly higher specificity (84.6%) and PPV (85.2%) compared with SPECTmax. The discrepancy between SPECTCTmax and SPECTmax might be due to the ROI drawing differences. Using ROI drawings based on the contour of the condyle rigorously limits the count of pixels within the boundary, regardless of the location of the maximum count per pixel, meaning that the maximum count within the ROI might not be the true maximum count per pixel of the radioactivity uptake area if the real maximum count per pixel is at the border of or outside the ROI. In contrast, SPECTmax always places the ROI around the maximum count pixel.
The results also showed that average counts per pixel of the ROI methods seemed to be less sensitive compared with the maximum count per pixel of the ROI methods in the SPECT and SPECT/CT methods. Compared with the maximum count of the ROI methods, methods using the average count of ROIs in SPECTCTaver or SPECTaver showed a relatively lower sensitivity of 66.7%. However, SPECTaver showed the highest specificity and PPV among all methods evaluated, whereas SPECTCTaver yielded the lowest sensitivity, specificity, NPV, and PPV. This could be explained by the fact that ROIs based on the contour of the condyle of the CT image of SPECT/CT arbitrarily and improperly truncate the activity distribution, which could severely underestimate the average counts per pixel if the maximum count is at or outside the border of the condyles. Furthermore, this could be attributed to the variation of average counts per pixel in the ROIs, which is dependent on the size and position of the ROI drawings. Moreover, maximum ROIs are not dependent on the position or size of the ROI as long as the maximum count per pixel is within the ROI and therefore are less operator dependent.13, 14
ROC analysis showed that the SPECTaver had the highest AUC of 0.899, which is consistent with a prior study using the same ROI method in which the AUC was 0.866 with the mean ROI method.8 Moreover, the present study also showed that SPECT/CT and SPECT had similar diagnostic performance in the SPECTaver, SPECTmax, and SPECTCTmax methods (P > .05), but not in SPECTCTaver, which showed a significantly smaller AUC compared with the others (P < .05).
The present results did not illustrate that SPECT/CT with precise ROI drawings is superior to SPECT in the assessment of CH, which could mainly relate to differences in ROI drawing techniques. In addition, the registration differences of radioactivity uptake over the condyles might have contributed to the results. Furthermore, the mandibular condyle is a morphologically small structure relative to the SPECT spatial resolution; thus, the radioactivity detected in the condyle is mostly governed by the point spread function of the imaging method rather than by the shape of the condyle.14 The present results also were confirmed by a previous study using SPECT/CT, which showed that SPECT and SPECT/CT had similar sensitivities (80%).10
This study has a few limitations. First, the retrospective study has its own biases. Second, histopathology was not available for the present patients. Third, SPECT/CT is costly and time consuming, and potential problems can arise from the image fusion misregistration between the SPECT and CT components. Potential image fusion misregistration also could lead to an erroneous radiotracer count for the ROIs. An additional radiation dose from the CT component also is incurred to the patient. Furthermore, because the TMJ is subjected less to attenuation artifacts compared with other deep-seated organs, the application of quantitative SPECT/CT might not lead to different results from conventional SPECT. Another limitation of this study is that it was conducted at a single center with a sample population composed mainly of patients of Asian descent.
In conclusion, 99mTc-MDP hybrid SPECT/CT provides new techniques to evaluate active CH. The method using ROIs drawn around the contour of the condyle on SPECT/CT does not show improved accuracy over conventional SPECT-fixed ROI methods in the assessment of active CH.
References
- Raijmakers, P.G., Karssemakers, L.H., and Tuinzing, D.B. Female predominance and effect of gender on unilateral condylar hyperplasia: A review and meta-analysis. J Oral Maxillofac Surg. 2012; 70: e72
- Almeida, L.E., Zacharias, J., and Pierce, S. Condylar hyperplasia: An updated review of the literature.Korean J Orthod. 2015; 45: 333
- Obwegeser, H.L. and Makek, M.S. Hemimandibular hyperplasia—Hemimandibular elongation. J Maxillofac Surg. 1986; 14: 183
- Wolford, L.M., Movahed, R., and Perez, D.E. A classification system for conditions causing condylar hyperplasia. J Oral Maxillofac Surg. 2014; 72: 567
- Rodrigues, D.B. and Castro, V. Condylar hyperplasia of the temporomandibular joint: Types, treatment, and surgical implications. Oral Maxillofac Surg Clin North Am. 2015; 27: 155
- Saridin, C.P., Raijmakers, P.G., Tuinzing, D.B. et al. Bone scintigraphy as a diagnostic method in unilateral hyperactivity of the mandibular condyles: A review and meta-analysis of the literature. Int J Oral Maxillofac Surg. 2011; 40: 11
- Saridin, C.P., Raijmakers, P.G., Al Shamma, S. et al. Comparison of different analytical methods used for analyzing SPECT scans of patients with unilateral condylar hyperactivity. Int J Oral Maxillofac Surg. 2009; 38: 942
- Rushinek, H., Tabib, R., Fleissig, Y. et al. Evaluation of three analysis methods for 99mTc MDP SPECT scintigraphy in the diagnosis of unilateral condylar hyperplasia. Int J Oral Maxillofac Surg. 2016; 45: 1607
- Lopez, B.D. and Corral, S.C. Comparison of planar bone scintigraphy and single photon emission computed tomography for diagnosis of active condylar hyperplasia. J Craniomaxillofac Surg. 2016; 44: 70
- Agarwal, K.K., Mukherjee, A., St, A. et al. Incremental value of single-photon emission computed tomography/computed tomography in the diagnosis of active condylar hyperplasia. Nucl Med Commun. 2017; 38: 29
- Kao, Y.H., Magsombol, B.M., and Ng, D.C. The potential of hybrid SPECT/CT fusion imaging to improve diagnostic accuracy in the scintigraphic quantitative functional assessment of suspected unilateral mandibular hyperactivity. Oral Maxillofac Surg. 2012; 16: 89
- Hodder, S.C., Rees, J.I., Oliver, T.B. et al. SPECT bone scintigraphy in the diagnosis and management of mandibular condylar hyperplasia. Br J Oral Maxillofac Surg. 2000; 38: 87
- AlSharif, A.A., Tarawneh, E.S., AlKawaleet, Y.I. et al. Standardization of quantitative single photon emission computed tomography in control individuals and in patients with condylar hyperplasia.Nucl Med Commun. 2014; 35: 1268
- Fahey, F.H., Abramson, Z.R., Padwa, B.L. et al. Use of (99m)Tc-MDP SPECT for assessment of mandibular growth: development of normal values. Eur J Nucl Med Mol Imaging. 2010; 37: 1002
Δεν υπάρχουν σχόλια:
Δημοσίευση σχολίου