Differentiation of Cervical Spine Osteoradionecrosis and Bone Metastasis After Radiotherapy Detected by Bone Scan in Patients With Nasopharyngeal Carcinoma: Role of Magnetic Resonance Imaging

Background: Osteoradionecrosis (ORN) of the cervical spine is a serious complication after radiotherapy (RT), which may show increased radiotracer uptake on a bone scan (BS) and be mistaken as metastasis. We aimed to assess the value of magnetic resonance imaging (MRI) in the differentiation of cervical spine ORN from bone metastasis after RT detected by BS in nasopharyngeal carcinoma (NPC). Methods: In this retrospective study, 35 NPC patients who had undergone RT were enrolled, of whom 21 patients showed cervical spine ORN and 14 showed bone metastasis. New areas of increased radiotracer uptake in the cervical spine on a BS were noted in all patients, following which the patients underwent neck MRI for further assessment. Two radiologists independently reviewed two sets of images including a BS set and an MRI set (MRI with BS) and reached a consensus. The diagnostic sensitivity, specificity, and accuracy for ORN detection were calculated, and interobserver agreement was evaluated using the kappa test. Results: A total of 75 cervical spine lesions were identified (44, ORN; 31 metastases). The BS set analysis showed that the diagnostic sensitivity, specificity, and accuracy were only 38.6, 48.3, and 42.7%, respectively, for differentiation of cervical spine ORN from bone metastasis. On the other hand, the MRI set analysis showed that the diagnostic sensitivity, specificity, and accuracy increased to 86.4, 90.3, and 88.0%, respectively. The interobserver agreement for the MRI set was determined to be very good (κ = 0.92). Conclusion: MRI is a reliable technique for the further discrimination of emerging cervical spine lesions after RT detected by BS. Furthermore, it could be a better differential diagnosis technique for distinguishing ORN from metastasis and may help avoid a wrong assignment of the patient to a metastatic stage with indication for treatment with supplemental toxicity and a subsequent palliative strategy.


INTRODUCTION
Nasopharyngeal carcinoma (NPC) is a malignant tumor with a very unique geographic distribution; it is mainly prevalent in southern China and Southeast Asia (1,2). In the regions with NPC prevalence, NPC has been reported in approximately 50 per 100,000 persons annually and this incidence is almost 50 times higher than that in western countries (2). Radiotherapy (RT) has been identified as the primary therapeutic method for NPC, and post-RT adverse events have drawn a great deal of attention.
Post-RT adverse events in patients with NPC involving the central nervous system, including radiation encephalopathy, diffuse white matter injury, and optic neuritis, have been welldescribed (3). Osteoradionecrosis (ORN), with an incidence of 10.1%, is also a common complication after RT; it frequently develops in the mandible, maxilla, and skull base (4)(5)(6). Nevertheless, ORN in the cervical spine has been regarded as a rare complication, although the cervical vertebrae are often included in the irradiation field (6)(7)(8)(9). Sometimes, cervical spine ORN may be misdiagnosed as bone metastasis, which may lead to patients accepting an unnecessary biopsy or harmful chemoradiotherapy (10)(11)(12).
Modern medical imaging has played a great role in the detection and characterization of bone disorders. Bone scan (BS) is considered as a sensitive enough technique for the detection of bone marrow diseases and has been widely applied for monitoring bone metastasis in patients with a malignancy (13). However, BS has a limitation in terms of specificity, because RT-induced bone complications, acute osteoporotic fractures, and the metastases may all result in increased uptake (14)(15)(16). MRI is quite sensitive for the detection of reactive marrow edema and fatty changes, and very helpful for the detection of soft-tissue masses. MRI has been proven to be valuable for distinguishing benign and malignant compression spinal fractures (17,18), and differentiating RT-induced insufficiency fractures from metastases (14). However, few studies have assessed the MRI findings of ORN (10)(11)(12), and the role of MRI in the distinction of cervical spine ORN and metastasis remains unknown. Thus, the purpose of this study was to assess the additional value of MRI in the identification of cervical spine ORN and metastasis after radiotherapy in patients with NPC detected by BS.

Patient Samples
The ethics committee of Affiliated Cancer Hospital & Institute of Guangzhou Medical University approved this retrospective study, and the requirement of patients' informed consent was waived. Between January 2013 and December 2016, data were reviewed for 2046 consecutive patients with pathology-proven NPC after RT. We found 73 patients showed new cervical spine lesions at follow-up BS. The following patients were included in this study: (1) patients with emerging areas of abnormal uptake in the cervical spine; (2) patients who underwent neck MRI for further assessment; and (3) patients for whom sufficient clinical and MRI follow-up data were available to resolve the diagnosis of lesions. The following patients were excluded from the study: (1) patients with an abnormal signal in the cervical spine on pretreatment MRI, (2) patients for whom MRI data were unavailable after lesion detection by BS; and (3) patients in whom the diagnosis of lesions could not be confirmed. Finally, 35 NPC patients who had undergone RT were enrolled, of whom 21 patients showed cervical spine ORN and 14 showed bone metastasis. The patient selection flowchart is shown in Figure 1.

BS Acquisition
Whole-body BS was performed using a Philips SPECT scanner (Netherlands). Both anterior and posterior whole-body bone images were obtained 3 h after intravenous injection of 15-25 mCi 99 mTc-MDP; the scan parameters were as follows: matrix, Frontiers in Oncology | www.frontiersin.org 256 × 1,024; scanning speed, 15 cm/min; and acquisition energy window, 140 keV (±7.5%).

Image Analysis
All images were independently reviewed by two radiologists (observer 1, J.S.L, with 12 years of experience; observer 2, B.G.L, with 15 years of experience) and an agreement was reached. Two sets of images were designed to differentiate cervical spine ORN and metastasis, including a BS (BS images alone) set and an MRI set (MR images and BS images). Observers were only aware of the RT history but were blinded to patients' other clinical records, other imaging examination (CT, PET/CT) data, and the final diagnosis. Regarding the qualitative diagnosis, all cervical spine lesions were classified on a three-point scale as benign, malignant, or equivocal.
In the BS analysis, the observers were requested to record the abnormal uptake location and number of the cervical spine, as well as whether abnormal uptake was noted in other anatomical locations except for the cervical spine. If focal radiotracer uptake was greater than that in the anterior iliac spine, the lesion was classified as malignant (13,17). A lesion was classified as benign if radiotracer uptake was equal to or lower than that in the anterior iliac spine and no abnormal uptake was noted in other anatomical locations except for the cervical spine. If a lesion was not classified as malignant or benign, it was classified as equivocal.
In the MRI set analysis, the number, location, MR signal characteristics (bone marrow edema change, degree, and pattern of enhancement), vertebral destruction, and body collapse of cervical spine lesions were recorded; any associated paravertebral soft-tissue masses were also recorded. In addition, other ancillary features were also documented, including abnormal enhancement or necrosis of the paravertebral muscle, abnormal neck lymphadenopathy (short axis larger than 1 cm), and radiation encephalopathy (REP). Regarding the qualitative diagnosis, the criteria for lesion classification as malignant were as follows: (1) presence of a distinct soft-tissue mass with abnormal enhancement on MRI (10,17); and (2) absence of a distinct soft-tissue mass but presence of marked vertebral and heterogeneous enhancement on MRI and involvement of abnormal neck lymphadenopathy or meeting of the BS criteria for malignancy. The criteria for lesion classification as benign were as follows: (1) presence of only reactive marrow edema change with or without paravertebral muscle edema (19,20); and (2) presence of an abnormal signal change on unenhanced images and homogeneous enhancement on contrast-enhanced images, and meeting of the BS criteria for benign lesions. If a lesion was not classified as malignant or benign, it was classified as equivocal.

Reference Standard
The final diagnosis of a cervical spine lesion was based on all available imaging investigations, clinical data, and followup MRI data for at least 6 months. The reference standard was as follows (10,19): (1) lesions that shrank or remained stationary at MRI for more than 6 months without radiotherapy and/or chemotherapy were interpreted as ORN; and (2) lesions with progressive enlargement that presented as soft-tissue masses or distinctly regressed after radiotherapy and/or chemotherapy were identified as metastases. If a lesion's nature could not be confirmed by a follow-up procedure, it was excluded.

Statistical Analysis
Categorical data were expressed as numbers and frequencies (%), and continuous data were expressed as median and range. Interobserver agreement between observers was assessed using the kappa test and defined as follows (21): 1 ≥ k > 0.8, very good; 0.8 ≥ k > 0.6, good; 0.6 ≥ k > 0.4, moderate; 0.4 ≥ k > 0.2, fair; and 0.2 ≥ k > 0, poor. The diagnostic sensitivity, specificity, and accuracy were calculated based on the diagnostic scale. Pearson chi-square test (or Fisher test) was used for comparing differences in the incidence of imaging features between ORN and bone metastasis. Statistical tests were performed using SPSS 16.0 (SPSS Inc., Chicago, IL, USA); P < 0.05 indicated statistical significance.

BS Findings and Diagnostic Performance
As shown in Table 3, in the patient-wise analysis, the incidence of involvement with abnormal uptake in other anatomical locations except for the cervical spine in the metastasis group (Figures 5, 6) was significantly higher than that in the ORN group (Figure 2  in metastases (Figures 5, 6) was significantly higher than that in ORNs (Figure 3) [48.4% (15/31) vs. 11.4% (5/44), P < 0.001].
The qualitative assessment showed that 17 lesions of the 27 ORNs detected on BS were interpreted as benign, 5 lesions were classified as malignant, and the remaining 5 lesions showed equivocal findings; 15 lesions of the 25 metastases detected on BS were interpreted as malignant, 2 lesions were classified as benign, and the remaining 8 lesions showed equivocal findings. The diagnostic sensitivity, specificity and accuracy in the discrimination of ORN from metastasis were 38.6% (17/44), 48.3% (15/31), and 42.7% (32/75), respectively. The interobserver agreement between the two observers in the BS set was determined to be good (k = 0.77).
For qualitative assessment, in the 44 cases showing ORNs, 38 lesions were interpreted as benign, 3 lesions were classified as malignant, and the remaining 5 lesions were considered to show equivocal findings. Of the 31 metastases, 28 lesions were interpreted as malignant, 3 lesions were classified as equivocal, and no lesion was classified as benign. For the discrimination of ORN from metastasis, with the addition of MRI, the diagnostic sensitivity, specificity, and accuracy increased to 86.4%   (38/44), 90.3% (28/31), and 88.0% (66/75), respectively. The interobserver agreement in the MRI set was determined to be very good (k = 0.92).

DISCUSSION
Accurately distinguishing cervical spine ORN from metastasis is crucial in clinical practice. This study demonstrated that both ORN and metastasis could show increased radiotracer uptake on a BS, and BS showed comparatively low detection sensitivity and classification efficiency in the discrimination of ORN and metastasis. MRI showed additional value when used along with BS; the combined approach showed a better lesion detection rate and improved diagnostic sensitivity and specificity.
BS has been considered as one of the most common and accessible modern imaging procedures for monitoring bone metastasis in patients with malignancy. BS shows enough sensitivity for detecting metastasis, but it shows several limitations in specificity because many benign bone tumors, infections, and degenerative disease may also show abnormal radiotracer uptake (13,15,16,22). Recently, BS has been applied for detecting RT-induced bone complications and has shown variable sensitivity in the detection of insufficiency fractures after RT for pelvic malignancy, with a detection rate ranging from 40 to 87.5% (14,15,23). In this study, BS showed a detection sensitivity of 61.4% for detecting ORN. Thus, novel imaging techniques need to be applied to improve the detection sensitivity of ORN.
MRI is an important alternative imaging modality for further assessment of vertebral benign and malignant diseases. However, the value of MRI in the characterization of cervical spine ORN has not been well-described. To date, only few studies have discussed the MRI features of ORN that occurred in the upper (C1/C2) cervical spine (10)(11)(12), and ORN in other anatomical locations of the cervical spine was also only reported in some case reports (24)(25)(26). In this study, we found that ORN more frequently developed in C1/C2, which may be attributable to the fact that C1/C2 is adjacent to the nasopharynx and may be more susceptible to RT. Our study also demonstrated that 66.7% of patients with ORN showed devolution with multiple vertebras, which was consistent with the findings of previous studies (10,11). We found some MRI findings that could facilitate the differential diagnosis of cervical spine ORN from metastasis. First, patients with metastasis tended to show the involvement of cervical lymphadenopathy, with almost 80% of the patients showing cervical spine metastasis combined with cervical lymphadenectasis, which was also consistent with the results obtained by Wu et al. (10). Second, reactive vertebral marrow edema was the most common feature for ORN, with up to 95.5% of ORNs showing vertebral marrow edema change, and the bone marrow edema in ORN was frequently asystematic and homogeneous, but the marrow edema in metastasis was more likely to be circumscribed and asymmetrical. Third, reactive paravertebral muscle edema was also a distinctive ancillary feature that may be associated with ORN. In addition, a vertebral soft-tissue mass was a reliable symptom for detecting metastasis. Although, King et al. (11) indicated that some cervical ORN patients may also have soft-tissue masses adjacent to the cervical spine, we found that these lesions were usually less bulky, and the signal intensity of edema tended to be extensive and rarely showed an occupied effect. These findings exhibited by ORN likely reflect localized infectious and inflammatory processes (19,20,27). Meanwhile, the soft-tissue involvement in ORN may be more symmetric than that in metastasis. In summary, accounting for these meaningful findings, MRI may be a reliable technique for distinguishing ORN from bone metastasis, and the possibility of aggressive radiotherapy for patients due to misdiagnosis may be effectively avoided with the involvement of MRI.
With respect to the diagnostic efficiency in the identification of cervical spine ORN and metastasis, BS showed limited value  for differential diagnosis. In our study, the diagnostic sensitivity, specificity, and accuracy using BS alone were only 38.6, 48.3, and 42.7%, respectively. This may be attributed to the relatively low lesion detection sensitivity on BS and the fact that a large number of lesions detected were classified as equivocal. We found that the diagnostic sensitivity increased to 86.4% and specificity increased to 90.3% with the addition of MRI. Meanwhile, 38.6% of ORNs and 19.4% of metastases that were not detected on BS were found on MRI, and the frequency of equivocal lesions was also reduced. These results were similar to those of several previous studies that assessed the added value of MRI or SPECT/CT to BS alone in the discrimination of RT-induced pelvic insufficiency fracture and pelvic metastasis in cervical cancer patients (13)(14)(15)22). Therefore, the results of our study indicated that MRI could be a noninvasive technique for further discrimination of new cervical spine lesions detected by BS.
This study had several limitations. First, puncture pathologic evaluation was not available due to the high probability of fracture or hemorrhage; thus we used standardized clinical and MRI follow-up to confirm ORN or metastasis. Second, this was a retrospective study with a limited sample performed in a single center. Further prospective studies with larger sample sizes are needed to validate the results. Third, only routine MRIs were used to assess the diagnostic efficacy, and the potential performance of some functional MR imaging techniques, such as diffusion-weighted imaging (DWI), and perfusion-weighted imaging (PWI), need to be Friday, January 10, 2020 1:47 pm explored further.
In conclusion, both cervical spine ORN and metastasis can show increased radiotracer uptake on a BS in patients with NPC after RT, and MRI showed additional value in identifying these cervical spine lesions. MRI could be a better differential diagnosis technique for distinguishing ORN from metastasis and may avoid wrong assignment of patients to a metastatic stage with indications for treatment modalities with supplemental toxicity and a subsequent palliative strategy.

DATA AVAILABILITY STATEMENT
The datasets generated for this study are available on request to the corresponding author.

ETHICS STATEMENT
The studies involving human participants were reviewed and approved by Affiliated Cancer Hospital & Institute of Guangzhou Medical University. Written informed consent for participation was not required for this study in accordance with the national legislation and the institutional requirements. Written informed consent was not obtained from the individual(s) for the publication of any potentially identifiable images or data included in this article.