Preoperative radiological evaluation of the extent of local invasion in maxillary squamous cell carcinoma (SCC) is very important in planning curative surgery. The aim of this study was to examine the accuracy of preoperative radiological evaluation with magnetic resonance imaging (MRI) for the extent of local invasion in maxillary SCC. A retrospective study was conducted of 33 patients who underwent a maxillectomy for maxillary SCC. We compared the MRI findings for 18 structures around the maxillary sinus with intraoperative or postoperative pathological findings. Discrepancies were found between preoperative MRI findings and intraoperative or postoperative pathological findings for 22 patients (66.7%). Overall, the specificity, sensitivity, positive predictive value, and negative predictive value of MRI were 83.4%, 83.0%, 64.5%, and 90.4%, respectively. The receiver operating characteristic curve showed that MRI evaluation of the posterolateral structures including the pterygoid plate, pterygoid muscle, and infratemporal fossa had a lower area under the curve (0.614) and a significantly lower accuracy when compared with the other structures ( P = 0.294, 95% confidence interval 0.405–0.822). In conclusion, as the accuracy of preoperative MRI evaluation of the posterolateral structures is low, careful evaluation of local extension to the posterolateral structures is needed when planning curative surgery for maxillary SCC.
The majority of maxillary squamous cell carcinomas (SCCs) are diagnosed at a locally advanced stage. As the anatomical structures surrounding the maxillary sinus are complex and many vital vessels and cranial nerves pass through it, it can be difficult to obtain sufficient surgical resection margins when performing curative surgery.
Several groups have reported that the local extension of maxillary cancer to the pterygoid plates and infratemporal fossa is a poor prognostic factor. Nazar et al. reported the significant association of the skull base and orbital involvement with poor survival in maxillary sinus malignancies. Also, the free resection margin has been reported to be the most important prognostic factor in reducing local recurrence and increasing the survival rate. To realize an adequate surgical margin, preoperative evaluation of local extension is important. However, the precise radiological evaluation of local extension in maxillary SCC is relatively difficult due to the anatomical complexity of the area. For this reason, magnetic resonance imaging (MRI) is usually used to evaluate the extent of local invasion in maxillary cancer, along with computed tomography (CT).
Ariyoshi and Shimahara compared MRI findings with those of CT in detecting the bone destruction and soft tissue infiltration of maxillary cancer. They reported that MRI was able to show bone destruction of each bony part to a nearly identical degree as CT, and that soft tissue infiltration of the tumour was more clearly detected on MRI compared with CT. Rajesh et al. reported a sensitivity of 100% and specificity of 75% for MRI in detecting bone involvement of maxillary cancer. However, there has been no report of the predictive value of MRI for each structure in detail around the maxillary sinus.
In this study, we calculated the sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV) of MRI using the 18 structures surrounding the maxillary sinus to estimate the accuracy of preoperative MRI evaluation for local invasion in maxillary SCC. Further, we plotted the receiver operating characteristic (ROC) curve to compare the accuracy of MRI in each direction.
Materials and methods
The medical records of 76 consecutive patients with maxillary SCC who underwent a total or radical maxillectomy, with or without craniofacial resection, between 1993 and 2011, were retrospectively reviewed. We only included cases for which preoperative MRI findings, intraoperative frozen biopsy results, and final pathology reports were available. All preoperative MRI findings assessed in this study were interpreted by one radiologist, who re-reviewed the MRI after surgical treatment without any information on the pathology reports. The MRI systems used in this study are shown in Table 1 . We excluded palliative cases, revision cases, and cases with second primary or distant metastasis. Finally, a total of 33 cases of maxillary SCC were included in this study.
|Intela Release 8||1.5 T||Philips||2001–2010|
|Achieva X-Series||2.0 T||Philips||2010–present|
We compared the preoperative radiological findings of local extension with the intraoperative frozen biopsy results or the final pathology reports and calculated the sensitivity, specificity, PPV, and NPV of MRI in maxillary SCC. The sensitivity was determined as the proportion of cases with positive MRI findings among pathologically positive cases. The specificity was determined as the proportion of cases with negative MRI findings among pathologically negative cases. PPV corresponds to the proportion of cases with positive pathological findings among cases with positive MRI findings. NPV corresponds to the proportion of cases with negative pathological findings among cases with negative MRI findings. We also plotted the ROC curve and calculated the area under the curve (AUC) for each direction from the maxillary sinus to estimate the accuracy of MRI in the preoperative evaluation of local extension. Due to the retrospective nature of this study, it was granted exemption in writing by the institutional review board.
The male to female ratio was 22:11 and the mean age of patients was 63 years (range 47–79 years). The characteristics of the cases enrolled are shown in Table 2 . The T and N stage distribution is shown in Table 3 .
|Age, years, range (mean)||47–79 (63)|
|Surgical procedure, n (%)|
|Total maxillectomy||14 (42.4)|
|Radical maxillectomy||15 (45.5)|
|Radical maxillectomy + CFR||4 (12.1)|
|Adjuvant therapy, n (%)|
|Radiation therapy||15 (45.5)|
|Concurrent chemoradiation therapy||3 (9.0)|
|Oncologic outcome, n (%)|
|No evidence of disease||21 (63.6)|