Comparison of PET/CT and MRI for the detection of bone marrow invasion in
patients with squamous cell carcinoma of the oral cavity
Yasser G. Abd El-Hafez a,b , Chien-Cheng Chen c , Shu-Hang Ng c , Chien-Yu Lin d , Hung-Ming Wang e , Sheng-Chieh Chan a , I-How Chen f , Shiang-Fu Huan f , Chung-Jan Kang f , Li-Yu Lee g , Chih-Hung Lin h , Chun-Ta Liao f, , Tzu-Chen Yen a, a Nuclear Medicine Department, Molecular Imaging Center, Chang Gung Memorial Hospital, Chang Gung University, Taoyuan, Taiwan, ROC b Radiotherapy and Nuclear Medicine Department, South Egypt Cancer Institute, Assiut University, Egypt c Department of Diagnostic Radiology, Chang Gung Memorial Hospital, Chang Gung University, Taoyuan, Taiwan, ROC d Department of Radiation Oncology, Chang Gung Memorial Hospital, Chang Gung University, Taoyuan, Taiwan, ROC e Department of Medical Oncology, Chang Gung Memorial Hospital, Chang Gung University, Taoyuan, Taiwan, ROC f Department of Otorhinolaryngology, Head and Neck Surgery, Chang Gung Memorial Hospital, Chang Gung University, Taoyuan, Taiwan, ROC g Department of Pathology, Chang Gung Memorial Hospital, Chang Gung University, Taoyuan, Taiwan, ROC h Department of Plastic and Reconstructive Surgery, Chang Gung Memorial Hospital, Chang Gung University, Taoyuan, Taiwan, ROC a r t i c l e i n f o Article history: Received 24 December 2010 Received in revised form 11 February 2011 Accepted 11 February 2011 Keywords: Bone invasion Oral cancer Mandible Maxilla Squamous cell carcinoma Positron emission tomography (PET)/ computed tomography (CT) Fluorodeoxyglucose Magnetic resonance imaging s u m m a r y Our aim was to retrospectively assess the diagnostic performance from combined positron emission tomography/computed tomography (PET/CT) and magnetic resonance imaging (MRI) for the detection of bone marrow invasion of the mandible or maxilla in patients with oral cavity squamous cell carcinoma (OCSCC). A total of 114 patients with OCSCC, arising from or abutting the upper or lower alveolar ridge, under- went staging PET/CT and MRI studies before surgery. The possibility of bone marrow invasion on PET/CT and MRI was graded retrospectively on a 5-point score. Histopathology was taken as the reference stan- dard. Sensitivity, specicity, predictive values and likelihood ratios were calculated. Clinical factors affecting the performance, like tumor origin and dentate status were also explored. PET/CT was found to be more specic than MRI (83% vs. 61%, respectively, p = 0.0015) but less sensitive (78% vs. 97%, respectively, p = 0.0391). Dentate status and tumor origin affected the diagnostic perfor- mance of PET/CT. In patients with positive MRI, sensitivity and specicity of PET/CT were 78% and 100% in dentate patients with alveolar ridge tumors, 75% and 80% in dentate patient with buccal tumors, 90% and 33% in edentulous patients with alveolar ridge tumors and 0% and 63% for edentulous patients with buccal tumors, respectively. PET/CT is more specic than MRI and can be used to complement the role of MRI. A negative MRI result can condently exclude the presence of bone marrow invasion, while in patients with positive MRI nd- ings, a negative PET/CT may be useful to rule out bone marrow invasion in dentate patients. 2011 Elsevier Ltd. All rights reserved. Introduction The preoperative detection of bone marrow invasion involving both the maxilla and mandible in patients with squamous cell carcinoma of the oral cavity (OCSCC) is critical to the planning of surgery and postoperative management. It is accepted that when the mandibular bone marrowis free of tumor, mandibular continu- ity can usually be preserved and a marginal mandibulectomy may be oncologically sufcient for minimal cortical erosion. 1,2 However, once bone marrow invasion has occurred, a much more extensive and lengthy operation involving segmental mandibulectomy plus bone reconstruction is necessary to provide an adequate clear bony margin. 3,4 When tumor invades the maxillary marrow spaces, greater resection margin and extensive reconstruction may be required. Computed tomography (CT) and magnetic resonance imaging (MRI) are the standard modalities used to evaluate primary tumor status in patients with OCSCC. In general, MRI is preferred because of its superior soft-tissue contrast resolution, 5 but its diagnostic accuracy in the detection of bone marrow invasion is still a matter of debate. 611 Indeed, few past studies have shown high sensitivity 1368-8375/$ - see front matter 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.oraloncology.2011.02.010
Corresponding authors. Address: Department of Otorhinolaryngology, Head and
Neck Surgery, Head and Neck Oncology Group, Chang Gung Memorial Hospital, Chang Gung University, 5 Fu-Hsin Street, Gweishan, Taoyuan 333, Taiwan, ROC. Tel.: +886 3 328 1200x8466; fax: +886 3 211 0052 (C.-T. Liao). Department of Nuclear Medicine, Molecular Imaging Center, Chang Gung Memorial Hospital, Chang Gung University, 5 Fu-Hsin Street, Gweishan, Taoyuan 333, Taiwan, ROC. Tel.: +886 3 328 1200x2744; fax: +886 3 211 0052 (T.-C. Yen). E-mail addresses: liaoct@adm.cgmh.org.tw (C.-T. Liao), yen1110@adm.cgmh. org.tw (T.-C. Yen). Oral Oncology 47 (2011) 288295 Contents lists available at ScienceDirect Oral Oncology j our nal homepage: www. el sevi er . com/ l ocat e/ or al oncol ogy and specicity, 10,11 whereas others demonstrated high sensitivity but low specicity. 69 The preoperative detection of bone marrow invasion may be theoretically enhanced with the use of combined positron emission tomography (PET)/CT imaging that detects metabolic anomalies via differences in tissue glucose uptake. However, relatively few studies have addressed this question. 1215 The purpose of our study was therefore to assess and compare the diagnostic accuracy of PET/CT and MRI in detecting bone marrow invasion involving both the maxilla and mandible in patients with OCSCC and also to ex- plore possible clinical factors, which might affect the performance of these modalities. Materials and methods Patients This retrospective study was approved by the Institutional Review Board at our hospital, and all patients provided written in- formed consent. The study population consisted of 114 patients with newly diagnosed OCSCC who were referred to our hospital between June 2006 and December 2009. Inclusion criteria were a diagnosis of SCC, originating from the alveolar ridge (upper or low- er) or other oral cavity subsites but involving the alveolar ridge, preoperative PET/CT and MRI staging studies and surgical manage- ment. All patients underwent marginal or segmental mandibulec- tomy, either with or without inferior maxillectomy. Segmental mandibulectomy was performed in the presence of: (a) bone mar- row invasion detected by conventional modalities (CT/MRI and orthopantogram); (b) an edentulous atrophic mandible; and (c) tu- mors close to the mandible precluding a marginal resection. Tumor margins were sent for frozen section. If frozen section margins were positive, additional tissue was excised and sent for frozen section in order to ensure that margins were tumor-free. The sur- gical defects were repaired with primary closure or reconstructed by free or local aps. Patients were classied as edentulous if they lost one or more tooth at the site of the tumor, either the upper or lower alveolar ridge, as seen in the immediate post-surgical specimens. Histopathology The surgical specimens were xed in 10% formalin solution and decalcied for a period of one day. Specimens were inked and soft tissue was removed to grossly evaluate the bone. In the presence of gross bone invasion, a representative 5 mm section of the bonetumor interface was prepared. In the absence of gross inva- sion, 2 sections (5 mm each) were taken from the site with the deepest tumor invasion. All sections were H&E stained in 5 lm thickness and reviewed by a single pathologist who was aware of the clinical staging. Table 1 General characteristics of the study participants. Parameter n % Sex Male 2 (1.8) Female 112 (98.2) Age, years Median (range) a 50 (2977) Anatomical subsites Tongue 6 (5.3) Floor of the mouth 6 (5.3) Buccal 47 (41.2) Alveolar ridge 39 (34.2) Hard palate 2 (1.8) Retromolar trigone 14 (12.3) Dentate status Dentate 64 (56.1) Edentulous 50 (43.9) Metal-related artifacts No 53 (46.5) Yes 61 (53.5) Clinical T-status T1 3 (2.6) T2 34 (29.8) T3 15 (13.2) T4 62 (54.4) Pathological T-status T1 3 (2.6) T2 38 (33.3) T3 14 (12.3) T4 59 (51.8) Pathological bone marrow invasion No 77 (67.5) Yes 37 (32.5) Site of marrow invasion Mandible 34 (29.8) Maxilla 2 (1.8) Both 1 (0.9) Bone management MM b 16 (14) SM c 36 (31.6) MM + MX d 36 (31.6) SM + MX 26 (22.8) a The numbers in parentheses indicate the range of the data. Unless otherwise indicated, values are given as numbers of patients (percentages in parentheses). b Marginal mandibulectomy. c Segmental mandibulectomy. d Maxillectomy. Table 2 Diagnostic performances of PET/CT and MRI for the detection of bone marrow invasion in patients with squamous cell carcinoma of the oral cavity. Imaging modality FN a TP b TN c FP d Total Sensitivity (95% CI e ) Specicity (95% CI e ) PPV f (95% CI e ) NPV g (95% CI e ) Accuracy (95% CI e ) LR+ h (95% CI e ) LR i (95% CI e ) PET/CT 8 29 64 13 114 78 (7186) 83 (7690) 69 (6178) 89 (8395) 82 (7489) 4.6 (0.88.5) 0.3 (0.71.2) MRI 1 36 46 29 112 j 97 (94100) 61 (5270) 55 (4665) 98 (95101) 73 (6581) 2.5 (0.4 to 5.4) 0.0 (0.3 to 0.4) a False negative. b True positive. c True negative. d False positive. e Condence interval. f Positive predictive value. g Negative predictive value. h Positive likelihood ratio. i Negative likelihood ratio. j Two cases had severe metal artifacts and could not be interpreted on MRI. Y.G. Abd El-Hafez et al. / Oral Oncology 47 (2011) 288295 289 PET/CT imaging Patients were asked to fast for at least 6 h before the start of the PET study. Serum glucose level was determined at the time of intravenous injection of 370 MBq (10 mCi) of 18 F-FDG and all the patients had glucose concentrations <200 mg/dL. PET/CT images were acquired using a combined PET/CT scanner (Discov- ery ST 16, GE Healthcare). To minimize FDG physiological uptake, speech or swallowing was forbidden unless strictly necessary from the intravenous injection of FDG to the completion of the study. No muscle relaxants were used. Before PET acquisition, helical CT images were obtained for anatomical localization/atten- uation correction (AL/AC) from the head to the proximal thigh according to a standardized protocol. The following settings were used: transverse 10-mm collimation, 16-slice mode, 120 kVp, auto mAs (range 10300), 0.5 s tube rotation, 17.5 mm/s table speed and pitch 1.75. No intravenous contrast materials were adminis- tered for the CT scan. CT data was resized from a 512 512 ma- trix to a 128 128 matrix to match the PET data to allow for image fusion and generation of CT transmission maps. Emission data were acquired from the head to the proximal thigh at 50 min after intravenous injection of the tracer. Data were ac- quired in two-dimensional mode (3 min per table position). The study was performed with the arms positioned along the side of the body and head holder was used. We did not perform a dedi- cated regional study for the head and neck region in all partici- pants following the whole-body acquisition. After CT-based AC, PET images were reconstructed with an ordered subset expecta- tion maximization iterative reconstruction algorithm (4 iterations and 15 subsets). Slice thickness was 3.75 mm. All data were then transferred to the processing workstation (Xeleris Workstation, GE Healthcare). CT scans were displayed with a bone algorithm and fused PET/CT images were evaluated in the axial, sagittal, and coronal planes in the default window level. No threshold was applied. Standardized uptake value (SUV) was dened as the highest activity concentration per injected dose per body weight (in kilograms) after correction for radioactive decay. Bone marrow invasion was considered to be present if there was focal FDG uptake within the marrow spaces contiguous with the pri- mary tumor. MRI imaging All patients were examined using either a 1.5-T MRI (Intera, Philips Medical Systems) or a 3.0-T (Magnetom Trio, Siemens Table 3 Differences in sensitivity, specicity and accuracy between PET/CT and MRI for the detection of bone marrow invasion in patients with squamous cell carcinoma of the oral cavity. Modality PET/CT p FN a TP b TN c FP d MRI FN a 0 1 0 0 0.0391 e TP b 8 28 0 0 0.0015 f TN c 0 0 42 4 0.214 g FP d 0 0 20 9 a False negative. b True positive. c True negative. d False positive. e Difference in sensitivity by McNemars test. f Difference in specicity by McNemars test. g Difference in area under the curve (AUC) by ROC curve analysis. Table 4 Association of different clinical characteristics with sensitivity, specicity and positive likelihood ratios of PET/CT in 65 patients with squamous cell carcinoma of the oral cavity and positive MRI for marrow invasion. Characteristic n FN a TP b TN c FP d Sensitivity (95% CI e ) p Specicity (95% CI e ) p LR+ f (95% CI e ) Anatomical subsite g Buccal 20 4 3 8 5 43 (2165) 0.0014 * 62 (4083) 0.8259 1.1 (3.5 to 5.7) Alveolar ridge 33 3 22 5 3 88 (7799) 63 (4679) 2.3 (2.8 to 7.5) Differentiation Well/moderate 57 7 26 16 8 79 (6889) 0.7572 67 (5479) 0.7375 2.4 (1.6 to 6.3) Poor 8 1 2 4 1 67 (3499) 80 (52108) 3.3 (9.1 to 15.8) Metal-related artifacts No 28 3 11 11 3 79 (6394) 0.9129 79 (6394) 0.1739 3.7 (3.3 to 10.6) Yes 37 5 17 9 6 77 (6491) 60 (4476) 1.9 (2.5 to 6.4) Dental status Dentate 28 3 10 13 2 77 (6193) 0.8379 87 (7499) 0.0043 * 5.8 (2.9 to 14.4) Edentulous 37 5 18 7 7 78 (6592) 50 (3466) 1.6 (2.4 to 5.6) Primary tumor size (cm) 64.8 (ROC) 42 3 18 14 7 86 (7596) 0.3700 67 (5281) 0.6970 2.6 (2.2 to 7.4) >4.8 23 5 10 6 2 67 (4786) 75 (5793) 2.7 (3.9 to 9.3) SUV of the primary tumor 613.43 30 2 12 12 4 86 (7398) 0.3284 75 (6090) 0.3934 3.4 (3.1 to 9.9) >13.43 35 6 16 8 5 73 (5887) 62 (4578) 1.9 (2.6 to 6.4) Patient groups g Dentate alveolar ridge 8 1 4 3 0 80 (52108) 0.0025 * 100 0.1165 # i Dentate buccal 10 1 3 4 2 75 (48102) 0.0027 * 67 (3796) 1.0000 2.3 (6.9 to 11.4) Edentulous alveolar ridge 25 2 18 2 3 90 (78102) <0.0001 * 40 (2159) 0.5910 1.5 (3.3 to 6.3) Edentulous buccal h 10 3 0 4 3 0 57 (2688) 0.0 (0.00.0) a False negative. b True positive. c True negative. d False positive. e Condence interval. f Positive likelihood ratio. g Only the two major subsites (buccal and alveolar ridge) were analyzed for this parameter. h Reference group. i False-positive rate equals zero, consequently positive likelihood ratio (LR+) is innity. * Statistically signicant difference. 290 Y.G. Abd El-Hafez et al. / Oral Oncology 47 (2011) 288295 Healthcare, Erlangen, Germany) MRI scanner. The MRI examina- tions were performed using a head-and-neck synergic coil to cover the entire neck fromthe skull base to the thoracic inlet. T2- and T1- weighted fast/turbo spinecho sequences were applied in the axial plane. Post gadolinium contrast enhanced T1-weighted fast/turbo spinecho sequences with fat saturation were applied in the axial, coronal, and sagittal planes at 4-mm slice thickness. All data were archived in DICOM format and transferred to a stand-alone work- station for processing. All of the imaging analysis was performed on a picture archiving and communication system (PACS) worksta- tion (Centricity 1.0; GE Healthcare, Milwaukee, WI, USA). Tumor invasion into the bone marrow was diagnosed by replacement of the marrow fat of the involved mandible or maxilla by contiguous tumors with abnormal T1 hypointensity, T2 hyperintensity, and denite contrast enhancement. Bone erosion was considered to be present if there was any irregular bone thinning while bone destruction was diagnosed in the presence of complete cortical bone loss. MR image distortions or artifacts close to the maxilla or mandible were carefully recorded. Image interpretation PET/CT images were reviewed by one nuclear physician (6 years experience in general nuclear medicine, 2 years in PET/CT) and one head-and-neck radiologist (20 years experience), and a consensus was reached via joint reevaluation of the images. All MRI data were read by an independent head-and-neck radiologist (7 years experi- ence), who was blinded to PET/CT ndings. The readers were aware of tumor origin and side. The images were evaluated qualitatively for the presence of bone marrow invasion on a 5-point score, in which a score of 0 indicated that bone marrow invasion was de- nitely absent; a score of 1, bone marrow invasion was probably ab- sent; a score of 2, ambiguous cases characterized by cortical bone erosion in the absence of overt bone destruction; a score of 3, bone marrow invasion was probably present; and a score of 4, bone marrow invasion was denitely present. A score of 2 or less was considered negative. SUV was not considered in the image interpretation. Statistical analysis Receiver operating characteristic (ROC) curve analysis was per- formed to assess discriminative power of PET/CT and MRI for bone marrow invasion. Histopathology was taken as the reference stan- dard; sensitivity, specicity, predictive values and likelihood ratios were calculated for expressing test performance. Categorical data and paired readings were analyzed using the Chi-squared (v 2 ) test and the McNemars test, respectively. We determined the optimal cutoff values for tumor size and maximum SUV by ROC analysis based on the presence of bone marrow invasion. SUVmax was compared in two sub-groups using Students t-test for independent samples. All statistical analyses were two-sided and the signi- cance level was xed at 0.05. Table 5 False-negative and false-positive results of PET/CT for the detection of bone marrow invasion in patients with squamous cell carcinoma of the oral cavity. No. Anatomical subsite Dental status Metal- related artifacts Differentiation cT a pT b Tumor size (cm) SUV c primary tumor Bone management Marrow invasion PET/CT score PET/CT result MRI score MRI result 1 AR d Dentate + MD f T4 T4 3.5 8.1 SM i + 2 FN l 4 TP n 2 Buccal Edentulous + MD f T4 T4 5 33.4 SM i + MX j + 2 FN l 4 TP n 3 AR d Edentulous + MD f T4 T4 3.2 16.9 SM i + 0 FN l 4 TP n 4 Buccal Edentulous + WD g T4 T4 9 10.7 SM i + MX j + 2 FN l 4 TP n 5 Buccal Edentulous + PD h T4 T4 6 16.8 SM i + MX j + 2 FN l 4 TP n 6 Buccal Dentate MD f T4 T4 5.2 13.5 SM i + MX j + 2 FN l 4 TP n 7 AR d Edentulous MD f T4 T4 2.8 25.3 SM i + 0 FN l 4 TP n 8 RMT e Dentate MD f T4 T4 5.3 20.9 SM i + 0 FN l 4 TP n 9 RMT e Edentulous WD g T4 T4 3.5 14.3 SM i + MX j 4 FP m 4 FP m 10 Buccal Edentulous + MD f T4 T4 4.8 12.6 MM k + MX j 4 FP m 4 FP m 11 AR d Edentulous + MD f T4 T3 5 12.6 SM i 4 FP m 4 FP m 12 Buccal Edentulous + MD f T4 T4 4.2 20.2 SM i 3 FP m 4 FP m 13 Buccal Edentulous MD f T4 T4 6.5 22 SM i + MX j 3 FP m 4 FP m 14 Buccal Dentate + MD f T4 T4 4.5 13.2 MM k + MX j 4 FP m 4 FP m 15 Buccal Dentate + WD g T4 T3 4.5 11.3 MM k + MX j 3 FP m 4 FP m 16 AR d Edentulous WD g T4 T3 3.8 19.8 SM i + MX j 3 FP m 4 FP m 17 AR d Edentulous + PD h T4 T2 2.5 14.6 MM k + MX j 4 FP m 4 FP m 18 Buccal Dentate + WD g T2 T2 3.5 23.4 MM k + MX j 4 FP m 2 TN o 19 AR d Edentulous + WD g T2 T2 2.3 10.3 MM k 3 FP m 2 TN o 20 AR d Edentulous + PD h T3 T3 5 4.4 SM i + MX j 4 FP m 2 TN o 21 Buccal Edentulous MD f T2 T2 3.5 22.3 MM k + MX j 3 FP m 1 TN o a Clinical tumor status. b Pathological tumor status. c Standardized uptake value. d Alveolar ridge. e Retromolar trigone. f Moderately differentiated. g Well differentiated. h Poorly differentiated. i Segmental mandibulectomy. j Maxillectomy. k Marginal mandibulectomy. l False negative. m False positive. n True positive. o True negative. Y.G. Abd El-Hafez et al. / Oral Oncology 47 (2011) 288295 291 Results General characteristics One hundred fourteen patients with squamous cell carcinoma of the oral cavity were enrolled (Table 1). The median time interval between diagnostic studies (PET/CT and MRI) and surgery was 2 days (range: 121 days for PET/CT; 156 days for MRI). The two studies (PET/CT and MRI) were done within 4 weeks of each other, except for one patient, who underwent MRI 54 days before PET/CT. Histopathological examination revealed the presence of marrow invasion in 37 patients (32.5%). More than 85% of the patients had the habits of cigarette smoking and betel quid chewing. The predominant tumor origins were buccal mucosa (41.2%) and alve- olar ridge (34.2%). Sixty-one patients had metal-related artifacts; in two of them, the MRI was severely distorted and uninterpretable. These two pa- tients were excluded from any subsequent comparative analyses. The presence of metal artifacts did not affect the performance of PET/CT. However, we encountered a single false negative result due to misregistration between PET and CT. The FDG activity seemed to be displaced by a few millimeters fromthe site of CT-de- tected cortical erosion (scored 2 by consensus). Diagnostic performance MRI showed the highest sensitivity and negative predictive val- ues (97% and 98%, respectively) with only one false negative result (Table 2). However, low specicity and positive predictive values were encountered (61% and 55%, respectively). Twenty-nine false- positive results were seen. Sensitivity and specicity of PET/CT were 78% and 83% respectively with 8 false negative and 13 false- positive results. Combined PET/CT was signicantly more specic (83% vs. 61%, p = 0.0015) and less sensitive than MRI (78% vs. 97%, p = 0.0391). The overall accuracy was comparable (Table 3). Clinical factors affecting PET/CT performance Tumor origin and dentate status were found to affect the diag- nostic performance of PET/CT. Bone marrow invasion was signi- cantly higher in edentulous patients than in dentate patients (23/ 50 vs. 14/64, p = 0.009). In dentate patients, PET/CT had higher Figure 1 Imaging ndings of a 56-year-old man with a diagnosis of squamous cell carcinoma of the left buccal mucosa, (A) coronal fused PET/CT image of the tumor mass without FDG uptake within the marrow cavity of the mandible. The corresponding non-contrast-enhanced CT image (B) demonstrated the presence of cortical bone erosion only. Post gadolinium contrast enhancement coronal T2 (C) and T1 (D) MR weighted images showed a left buccal-gum cancer, with hyperintensity and enhancement in the bone marrow of left mandible, suggesting bone marrow invasion. Pathological results were positive for marrow invasion. 292 Y.G. Abd El-Hafez et al. / Oral Oncology 47 (2011) 288295 specicity (94% vs. 63%, p = 0.0001) and positive likelihood ratio ([LR+], 13.1 vs. 2.1). A total of 13 false-positive results were encountered, 10 of them were seen in edentulous patients. Inter- estingly, all the false-positive results were in patients who had the habit of betel quid chewing. Alveolar ridge and buccal mucosa represented the two major tumor origins in our study (n = 39 and 47, respectively). Tumors originating in the alveolar ridge were signicantly associated with more bone marrow invasion (25/39 vs. 7/47, p < 0.0001). Consider- ing only these two subsites, PET/CT offered signicant advantages in terms of sensitivity (88% vs. 43%, p < 0.0001) and positive predic- tive value (81% vs. 30%, p < 0.0001) in alveolar ridge tumors com- pared with buccal tumors. There was a trend toward higher sensitivity, specicity, and LR+ when the size of the tumors was less than 4.8 cm or the SUV values less than 13.43 mg/mL. MRI was positive in 65 patients; 29 of them were falsely posi- tive. Given this high false positive rate, we tried to identify a sub- group of patients with MRI-positive ndings in whom PET/CT may be useful (Table 4). Among these 65 patients, MRI gave false-posi- tive results in 53.6% of dentate patients (15/28). PET/CT success- fully excluded the presence of bone marrow invasion in 13 of these 15 cases. In alveolar ridge tumors (n = 33), PET/CT also ex- cluded disease in 5 out of 8 falsely positive MRI results. False-neg- ative and false-positive results of PET/CT for the detection of bone marrow invasion in this study are shown in Table 5. Discussion Assessment of facial bones invasion by OSCC is important for head and neck surgical oncologists before treatment planning. In the past, we largely relied on the ndings from physical examina- tion plus conventional images, such as panorex X-ray, CT and/or MRI with unsatisfying results, especially when patients had a cer- tain degree of dental problems. Studies from other groups have shown that both PET/CT and SPECT/CT may be clinically useful, with varying degrees of sensitivity (58.3100%, 92%, 41.7100%, 39.1100%) and specicity (85100%, 86%, 57.1100%, 4097.1%) for PET/CT, SPECT/CT, CT, and MRI, respectively. 1215 In this study, we found a sensitivity and specicity of 78% and 83% for PET/CT, and 97% and 61% for MRI, respectively. Our data suggest that PET/CT may complement the role of MRI for diagnosing bone mar- row invasion in patients with oral cavity cancer. A negative MRI Figure 2 Imaging ndings of a 41-year-old man with a diagnosis of squamous cell carcinoma of the right buccal mucosa, (A) transaxial fused PET/CT image of the tumor mass that shows intense FDG uptake (SUVmax = 22.3) and it violates the bone boundary of the maxilla. The corresponding coronal non-contrast-enhanced CT image (B) demonstrated the presence of cortical discontinuity, and also evidence of some periodontal disease. The lesion scored 4 on PET/CT by consensus. Post gadolinium contrast enhancement axial T2 (C) and T1 (D) MR weighted images showed right upper gum cancer without abnormal signal intensity at right maxilla, suggestive of no bone marrow invasion. Pathological result was negative for marrow invasion. Y.G. Abd El-Hafez et al. / Oral Oncology 47 (2011) 288295 293 scan could condently exclude the presence of marrow invasion in patients with OCSCC. However, positive results obtained with this modality cannot always be considered true positive and PET/CT scan may help to make the diagnosis. We encountered a single false negative MRI result. Our ndings concerning the high nega- tive predictive value of MRI were well in line with the results of previous studies. 611 In our experience, there are two possible reasons to explain the higher false-positive results. First, Taiwan is an endemic area for betel quid chewing, and oral cancers in association with trismus with subsequent poor oral hygiene are commonly seen in clinical practice. 16,17 Betel quid chewing was observed in 87% of patients in this study. There were 28 false-positive results out of 99 patients chewing betel quid compared to 1/15 patients not having this habit (positive predictive value = 51% vs. 88%, p = 0.02). Second, all the studied tumors were involving the alveolar ridge, which could eli- cit a local inammatory response. It has been reported that MRI may yield falsely positive results in inammatory odontogenic disease. 18 We encountered 8 false-negative PET/CT ndings (Table 5); in 4 of them, the non-optimized CT intended for AC/AL was unable to detect sites of minimal bone breakthrough. We assume that a more optimized CT protocol for the head and neck region may add to the diagnostic yield of PET/CT. The other 4 false-negative PET/CT results were either due to low tracer uptake in the bone marrow (patients # 1 and # 6), misregistration of the CT and PET images (Patient # 5) or retraction of the edentulous alveolar ridge (Patient # 2), which mislead the localization of FDG uptake (Fig. 1). PET/CT showed signicantly higher false positive rate in edentulous com- pared to dentate patients (37% vs. 6%, p = 0.0001). The reason for that may be explained by: (a) Repeated episodes of periodontitis are common in edentulous patients which may show avidity to FDG (Fig. 2). 19 (b) The mean SUV of the primary tumor was signif- icantly higher in edentulous compared to dentate patients (15.2 vs. 12.6, p = 0.036). High FDG uptake may produce a spillover effect, leading to overestimation of the exact tumor extension (Fig. 3). 20 In this study, the sensitivity and positive predictive value of PET/CT were signicantly lower for buccal compared to alveolar ridge tumors; however, the negative predictive value, LR+ and overall accuracy remained comparable. Of note, among all the false-positive results encountered by PET/CT, the mean SUV within buccal tumors was higher, though not signicantly, than that seen in alveolar ridge tumors (17.8 vs. 12.3, p = 0.1). Our study is not without important limitations. A head and neck PET/CT study was not performed in all participants following the whole-body acquisition. No contrast media were used routinely for the CT portion of PET/CT. Moreover, the CT portion of PET/CT was not optimized for diagnostic purpose. In addition, the inevita- ble partial volume effect in PET/CT should be considered. Finally, this study is a retrospective single-reader analysis, which reects a single center experience. Strengths of our report include the large sample size, the use of histopathological examination as the gold standard for comparison, and the homogeneity of diagnosis and treatment plans. Figure 3 Imaging ndings of a 43-year-old man with a diagnosis of squamous cell carcinoma of the right buccal mucosa, (A) coronal fused PET/CT image of the tumor mass that shows intense FDG uptake (SUVmax = 22) and it violates the bone boundary of the mandible. The corresponding coronal non-contrast-enhanced CT image (B) demonstrated the presence of cortical erosion in the edentulous alveolar socket and possible breakthrough. Post gadolinium contrast enhancement coronal T2 (C) and T1 (D) MR weighted images showed right buccal-gum cancer with mild bone erosion on the edentate alveolar socket. No abnormal signal intensity at right mandible, indicating bone cortex invasion without bone marrow involvement. Pathological result was positive for both periosteal and cortical invasion but negative for marrow invasion. 294 Y.G. Abd El-Hafez et al. / Oral Oncology 47 (2011) 288295 Conclusion PET/CT is more specic than MRI and can be used to comple- ment the role of MRI. A negative MRI result can condently exclude the presence of bone marrow invasion in patients with squamous cell carcinomas of the oral cavity. In dentate patients with positive MRI ndings, a negative PET/CT may be useful to rule out bone marrow invasion in dentate patients. PET/CT had a signicantly higher sensitivity and positive predictive value in alveolar ridge tu- mors compared to buccal mucosa tumors. Conict of Interest None declared. Acknowledgements This work was supported in part by a Grant-in-Aid for FDG PET Research in Oral Cancer from Chang Gung Memorial Hospital-Link- ou (CMRPG370062). References 1. Barttelbort SW, Bahn SL, Ariyan SA. Rim mandibulectomy for cancer of the oral cavity. Am J Surg 1987;154(4):4238. 2. Ord RA, Sarmadi M, Papadimitrou J. A comparison of segmental and marginal bony resection for oral squamous cell carcinoma involving the mandible. J Oral Maxillofac Surg 1997;55(5):4707. discussion 78. 3. Politi M, Costa F, Robiony M, Rinaldo A, Ferlito A. Review of segmental and marginal resection of the mandible in patients with oral cancer. Acta Otolaryngol 2000;120(5):56979. 4. Wax MK, Bascom DA, Myers LL. Marginal mandibulectomy vs segmental mandibulectomy: indications and controversies. Arch Otolaryngol Head Neck Surg 2002;128(5):6003. 5. Ahmad A, Branstetter BF. CT versus MR: still a tough decision. Otolaryngol Clin North Am 2008;41(1):122. 6. Rajesh A, Khan A, Kendall C, Hayter J, Cherryman G. Can magnetic resonance imaging replace single photon computed tomography and computed tomography in detecting bony invasion in patients with oral squamous cell carcinoma? Br J Oral Maxillofac Surg 2008;46(1):114. 7. Tsue TT, McCulloch TM, Girod DA, Couper DJ, Weymuller Jr EA, Glenn MG. Predictors of carcinomatous invasion of the mandible. Head Neck 1994;16(2):11626. 8. Van Cann EM, Koole R, Oyen WJ, de Rooy JW, de Wilde PC, Slootweg PJ, et al. Assessment of mandibular invasion of squamous cell carcinoma by various modes of imaging: constructing a diagnostic algorithm. Int J Oral Maxillofac Surg 2008;37(6):53541. 9. van den Brekel MW, Runne RW, Smeele LE, Tiwari RM, Snow GB, Castelijns JA. Assessment of tumour invasion into the mandible: the value of different imaging techniques. Eur Radiol 1998;8(9):15527. 10. Bolzoni A, Cappiello J, Piazza C, Peretti G, Maroldi R, Farina D, et al. Diagnostic accuracy of magnetic resonance imaging in the assessment of mandibular involvement in oral-oropharyngeal squamous cell carcinoma: a prospective study. Arch Otolaryngol Head Neck Surg 2004;130(7):83743. 11. Vidiri A, Guerrisi A, Pellini R, Manciocco V, Covello R, Mattioni O, et al. Multi- detector row computed tomography (MDCT) and magnetic resonance imaging (MRI) in the evaluation of the mandibular invasion by squamous cell carcinomas (SCC) of the oral cavity. Correlation with pathological data. J Exp Clin Cancer Res 2010;29:73. 12. Babin E, Desmonts C, Hamon M, Benateau H, Hitier M. PET/CT for assessing mandibular invasion by intraoral squamous cell carcinomas. Clin Otolaryngol 2008;33(1):4751. 13. Goerres GW, Schmid DT, Schuknecht B, Eyrich GK. Bone invasion in patients with oral cavity cancer: comparison of conventional CT with PET/CT and SPECT/ CT. Radiology 2005;237(1):2817. 14. Gu DH, Yoon DY, Park CH, Chang SK, Lim KJ, Seo YL, et al. CT, MR, (18)F-FDG PET/CT, and their combined use for the assessment of mandibular invasion by squamous cell carcinomas of the oral cavity. Acta Radiol 2010;51(10): 11119. 15. Pentenero M, Cistaro A, Brusa M, Ferraris MM, Pezzuto C, Carnino R, et al. Accuracy of 18F-FDG-PET/CT for staging of oral squamous cell carcinoma. Head Neck 2008;30(11):148896. 16. Liao CT, Chang JT, Wang HM, Ng SH, Hsueh C, Lee LY, et al. Analysis of risk factors of predictive local tumor control in oral cavity cancer. Ann Surg Oncol 2008;15(3):91522. 17. Lin CY, Wang HM, Kang CJ, Lee LY, Huang SF, Fan KH, et al. Primary tumor site as a predictor of treatment outcome for denitive radiotherapy of advanced- stage oral cavity cancers. Int J Radiat Oncol Biol Phys 2010;78(4):10119. 18. Imaizumi A, Yoshino N, Yamada I, Nagumo K, Amagasa T, Omura K, et al. A potential pitfall of MR imaging for assessing mandibular invasion of squamous cell carcinoma in the oral cavity. Am J Neuroradiol 2006;27(1):11422. 19. Shimamoto H, Tatsumi M, Kakimoto N, Hamada S, Shimosegawa E, Murakami S, et al. (18)F-FDG accumulation in the oral cavity is associated with periodontal disease and apical periodontitis: an initial demonstration on PET/CT. Ann Nucl Med 2008;22(7):58793. 20. Soret M, Bacharach SL, Buvat I. Partial-volume effect in PET tumor imaging. J Nucl Med 2007;48(6):93245. Y.G. Abd El-Hafez et al. / Oral Oncology 47 (2011) 288295 295