IMPACT OF MAGNETIC RESONANCE IMAGING VERSUS CT ON NASOPHARYNGEAL CARCINOMA: PRIMARY TUMOR TARGET DELINEATION FOR RADIOTHERAPY Na-Na Chung, MD, 1 Lai-Lei Ting, MD, 1,2 Wei-Chung Hsu, MD, 3 Louis Tak Lui, MD, 1,2 Po-Ming Wang, MD 3 1 Division of Radiation Oncology, Department of Oncology, National Taiwan University, 1, Chang-Te Street, Taipei, 100, Taiwan, Republic of China. E-mail: llt@ha.mc.ntu.edu.tw 2 Department of Radiology, College of Medicine, National Taiwan University, Taipei, Taiwan 3 Department of Radiation Oncology, Cheng-Ching Hospital, Taichung, Taiwan Accepted 17 August 2003 Published online 8 January 2004 in Wiley InterScience (www.interscience.wiley.com). DOI: 10.1002/hed.10378 Abstract: Background. Our aim was to assess the capacity of CT versus MRI for delineating to the primary tumor extent of nasopharyngeal carcinoma (NPC) in treated patients. Methods. From December 1997 to April 2000, 258 patients with NPC were enrolled. We focused on the primary tumor extension and the discrepancy between CT and MRI. The delineation of tumor invasion was crucial for determination of the gross tumor volume (GTV) before radiation therapy. Results. A total of 104 patients (40.3%) had intracranial infiltration detected by MRI, whereas CT showed negative findings ( p = 6.879 10 11 ). Once the pterygopalatine fossa was involved, the chance of intracranial invasion was increased (96.1%). The detectable percentage of pterygopalatine fossa involvement accompanying intracranial invasion was higher with MRI than with CT (96.1% vs 56.9%). Conclusions. More detailed information about T and N classification of NPC was provided by MRI than by CT, which led to better target delineation for radiotherapy. A 2004 Wiley Periodicals, Inc. Head Neck 26: 241 246, 2004 Keywords: nasopharyngeal carcinoma; radiation therapy; CT; MRI Correspondence to: L.-L. Ting B 2004 Wiley Periodicals, Inc. With the advent of new imaging modalities, assessment of the tumor extension of head and neck malignancies becomes more precise. Nasopharyngeal carcinoma (NPC) is common in southern China and is known to have a tendency for submucosal spread by perineural invasion or direct skull base bony erosion into the intracranium. The incidence of intracranial extension of this infiltrative lesion is reported to be 12.2% based on CT findings. 1 Compared with conventional imaging studies such as plain radiographic imaging and polytomography, CT has been the most reliable and well-established imaging tool for NPC primary tumor evaluation. 2,3 However, it is difficult to evaluate invasion of the cavernous sinus or intracranium by tumor, especially for subtle lesions adjacent to the skull base. This limitation may cause understaging of the disease and lead to suboptimal treatment planning for radiotherapy and a poor treatment outcome. In recent years, MRI has become more valuable in both staging and treatment planning for radiotherapy because of its superior capability in demonstrating Impact of MRI versus CT on NPC Treatment HEAD & NECK March 2004 241
primary soft tissue invasion and subtle intracranial invasion. This study focused on 258 patients with NPC for primary tumor extension and assessment of the capacity of CT versus MRI for delineating the primary tumor extent, which was crucial to the determination of the gross tumor volume (GTV) before radiation therapy. MATERIALS AND METHODS From December 1997 to April 2000, a total of 327 consecutive patients with NPC visited the Division of Radiation Oncology at National Taiwan University Hospital, Taiwan. All patients with histologic confirmation were enrolled in this study. They underwent CT and/or MRI for evaluation of primary tumor extension. The CT scans were obtained with a Siemens Somatom DRH (Siemens Medical Systems, Forchheim, Germany), or Picker PQ 6000 ( Picker International, Highland Height, OH), in axial and coronal planes after the injection of contrast medium (Ultravist 370, Schering, Berlin, Germany) using 8-mm section thickness and a 5-mm section for cavernous sinus; 50 ml of Ultravist was administered by intravenous push before imaging and 50 ml during the image acquisition. MRI was performed with GE Signa 1.5 Tesla (GE Medical Systems, Milwaukee, WI). The spin-echo (SE) technique was used in the axial, coronal, and sagittal planes. T1-weighted (TR/TE 500-600/10-20/1, repetition time [TR] msec/echo time [TE] msec/excitation) and T2- weighted (TR/TE 3500-5000/80-110/1) images were obtained in the axial planes. The axial and coronal images were repeated with similar parameters after administration of a bolus injection of the contrast agent gadolinium-diethylenetriaminepentacetate, Gd-DTPA (Magnevist, Schering, Berlin, Germany), at a concentration of 0.1 mm/kg body weight. Slice thickness was Table 1. T classification of American Joint Committee on Cancer for NPC (6th ed, 2002). T1 T2 T2a T2b T3 T4 Tumor confined to nasopharynx Tumor extends to soft tissues Tumor extends to the oropharynx and/or nasal cavity without parapharyngeal extension* Any tumor with parapharyngeal extension* Tumor involves bony structures and/or paranasal sinuses Tumor with intracranial extension and/or involvement of cranial nerves, infratemporal fossa, hypopharynx, orbit, or masticator space *Parapharyngeal extension denotes posterolateral infiltration of tumor beyond the pharyngobasilar fascia. Table 2. Detection rate of intracranial lesion in NPC patients (CT vs MRI). No. of patients MRI ( ) MRI (+) p value CT negative 59/258 (22.9%) 104/258 (40.3%) CT equivocal 0 50/258 (19.4%) CT positive 0 45/258 (17.4%) Total 59/258 199/258 (22.9%) (77.1%) 2.09 10 10 2.09 10 10 2.09 10 10 performed as follows: 5-mm section thickness with 1.5-mm interslice gap in both axial and coronal sections. The image matrix was 256 192 in all scanning planes and sequences. CT alone was performed in nine patients, and MRI alone was performed in 60 patients. Both imaging modalities were used with 258 patients. There were 187 men and 71 women (the malefemale ratio was 2.63:1), and the patients ages ranged from 15 to 81 years (mean, 47.2 years). All the images were reviewed and assessed by two of the authors independently. CT images were first reviewed and recorded, followed by MR images the day after. Cases with variable interpretation or disagreement in T classification between the observers (CT and MRI) were reevaluated side by side, and the differences were confirmed to reach a final consensus. Interobserver variability in the CT scans was noted in 23 of 258 cases; in the MR images, in nine of 258 cases. The kappa test for interobserver concordance was 0.837 for CT and 0.952 for MRI. We focused on the comparison of the discrepancy between these two imaging modalities in the primary tumor extension in terms of intracranial infiltration (including nodular dural thickening and cavernous sinus invasion) and pterygopalatine fossa involvement. Intracranial infiltration is established if either dural enhancement had focal Table 3. Distribution of patients by T stage, sex, and age in 258 NPC patients. No. of patients Male female Mean age F SD Median age (range) T1 26 19:7 45.1 F 8.73 45 (15 81) T2 22 13:9 46.3 F 11.26 45.5 (26 64) T3 11 6:5 41.7 F 16.61 43 (15 67) T4 199 149:50 47.9 F 12.46 47 (16 81) Overall 258 187:71 47.2 F 12.31 46 (15 81) 242 Impact of MRI versus CT on NPC Treatment HEAD & NECK March 2004
sections of variable width or there was narrowing and displacement of the venous wall. The log-rank test was used to assess statistical significance of detection rate for intracranial infiltration, and Fisher s exact test was used for pterygopalatine fossa involvement. For evaluation of the impact of changes in T classification of NPC by MRI versus CT, the T classification was defined in accordance with the criteria in the American Joint Committee on Cancer (AJCC) Staging Manual, 6th edition (2002) (Table 1). RESULTS Both CT and MRI were performed in 258 patients. MRI showed intracranial invasion in 199 patients (77.1%); CT, in 50 patients (19.4%) (questionable subtle lesion) and 45 patients (17.4%) (well-demonstrated lesion), for a total of 95 patients (36.8%). These results revealed obvious statistically significant differences ( p = 2.09 10 10 ) by log-rank test. While comparing CTequivocal and -positive intracranial invasion cases with MRI-positive cases, MRI disclosed more identified results than did CT; the statistical significance was obvious (p = 6.879 10 11 ). The MRI findings increased the percentage of T4 patients from 36.87% (95 of 258 patients) to 77.1% (199 of 258 patients), with 104 cases (40.3%) of intracranial infiltration being detected by MRI, but not CT. There was not a single case in which intracranial invasion was seen on CT but not on MRI (Table 2). We also did not find equivocal tests or discrepancies in MRI interpretation for the delineation of tumor, particularly on the lesion adjacent to the skull base, because of good resolution of tumor by MRI with contrast enhancement. Distribution of the primary tumor (T classification) and the patients characteristics are shown in Table 3. We also found that once the tumor had invaded the pterygopalatine fossa (51 patients), the incidence of intracranial invasion was 56.9% (29 of 51 patients) detected by CT and 96.1% (49 of 51 patients) by MRI. The detectable Table 4. Detection rate of NPC patients with pterygopalatine fossa involvement and association of intracranial lesion by CT vs MRI (p = 0.181). MRI negative No. of patients MRI positive CT negative 2/51 (3.9%) 20/51 (39.2%) CT positive 0 (0%) 29/51 (56.9%) Total 2/51 (3.9%) 49/51 (96.1%) percentage of pterygopalatine fossa involvement accompanying intracranial invasion was higher by MRI than by CT. A higher detectable rate of intracranial invasion in the pterygopalatine fossa involvement group was demonstrated. However, there were no significant differences between CT and MRI in demonstrating the involvement of the pterygopalatine fossa and intracranium by Fisher s exact test (p =.181) (Table 4). DISCUSSION NPC is a common head and neck tumor in Taiwan and is endemic in southern China. Because NPC tends to progress upward along the fascial planes to the base of the skull, imaging studies with both axial and coronal views must be evaluated to rule out bony erosion of skull base and/or intracranial spread by the tumor. Intracranial lesions frequently appear as contrast-enhanced lesions adjacent to the cavernous sinus. MRI defines the extent of NP disease better than CT does by its good soft tissue contrast. More detailed image information about tumor infiltration can be provided. Its ability to detect early intracranial invasion is important for NPC staging and upstaging the T stage, as well as the treatment policy and target localization in radiotherapy treatment planning. It is clear that MRI will certainly have roles both in determining T and N classifications and radiotherapy treatment for NPC patients. The superior extension to the intracranial region is commonly seen in patients with NPC. The overall frequency of intracranial extension is reported as 12.2% by CT as reported by Sham et al 1 and 31% by MRI as reported by Chong et al. 4 In our series, the incidence of intracranial invasion demonstrated by CT was 36.8% (composite of 19.4% as equivocal lesions and 17.4% as welldemonstrated intracranial lesions) and by MRI was 77.1%. By CT, 40.3% (104 patients) were defined as no intracranial invasion, but enhanced lesions at intracranial region were finally identified by MRI. MRI increased the T classification in 40.3% to 59.7% (the latter considered equivocal lesion revealed by CT as negative) of cases. This result showed that MRI had an advantage in evaluating intracranial extension, especially for subtle lesions (including nodular dural thickening or enhancement and cavernous sinus invasion) that might lead one to contemplate more aggressive and precise treatment planning in radiotherapy or incorporate multimodality treatments such as induction chemotherapy followed by con- Impact of MRI versus CT on NPC Treatment HEAD & NECK March 2004 243
FIGURE 1. CT versus MRI for evaluation of equivocal intracranial lesion and bony destruction (case 1 [a, b, c] and case 2 [d, e, f]). A suspicious lesion at the right cavernous sinus (white arrow) shown by CT (b, e) was clarified by MRI with T1-weighted Gd contrast enhancement (c, f). The right skull base (in case 1) and the right sphenoid floor (in case 2) showed no apparent bony destruction (black arrowhead) in the bone window of the CT (a, d) and corresponded to the high signal in MRI (c, f) with tumor infiltration. current chemotherapy 5 or concurrent chemoradiotherapy plus adjuvant chemotherapy. 6 A higher intracranial extension rate is noted in our cases (77.1% by MRI) than in other studies. 4 This may result from the primary tumor of our patients being much more advanced and bulky. Besides, we also believed that nodular dural thickening and/or cavernous sinus invasion should indicate evidence of intracranial invasion, 7 although without pathologic proof. In our experience, it was meaningful and important to cover all these lesions in treatment planning. CT and MRI have respective specific advantages and disadvantages. In general, CT displays obvious lesions with cortical bony defect or cortical erosion at the skull base, especially with the bone window. Contrary to CT, MRI may also show an equal ability to identify subtle invasion of the basal skull region by its better identification of subtle marrow space infiltration of skull base erosion (eg, pterygoid bone, sphenoid wing, clivus, petrous); as well, intracranial extension can be recognized, especially in coronal sections with T1-weighted contrast enhancement using fatsuppression techniques because of its greater contrast resolution between normal structure and tumor. 8 MRI adds significant information that is not apparent on CT. In our study, results were obtained similar to those reported by Ng et al 9 and Chong et al. 10 The criterion of bony involvement on MRI was diagnosed as high signal intensity of cortex, marrow replacement by tumor, and contrast enhancement in bone. Early alteration in soft tissue composition through subtle destruction and erosion of the basal skull is characterized by MRI coronal views that allow better understanding of the route of skull base invasion by NP tumor, as well as the extent of intracranium (Figure 1). MRI better assesses different tissue components, particularly the differentiation of cavernous sinus infiltration, nodular dural thickening, and the adjacent normal structure 11 (Figure 2). Excellent imaging quality and contrast between normal and pathologic anatomy in this region can be well demonstrated, particularly with T1- weighted gadolinium-dtpa contrast-enhanced coronal views using the fat-suppression technique. Coronal scans provide valuable information in assessing skull base involvement by the NP tumor. It allowed us to clarify questionable intracranial lesions (Figure 3). 244 Impact of MRI versus CT on NPC Treatment HEAD & NECK March 2004
FIGURE 2. (A) NPC tumor with right pterygopalatine fossa invasion (black arrowhead) in axial view by CT (A-a) and MRI (A-b). Upward extension to right cavernous sinus (white arrow) is shown by T1-weighted contrast-enhanced MRI in coronal view (A-d) but not by CT (A-c). (B) Series of coronal MRI sections of the same patient in (A) with right intracranial invasion. Further clarification of the (A-d) intracranial lesion. Nodular dural thickening and enhancement (B-a, B-b, B-d) with soft tissue tumor mass and venous encasement (B-c) is seen (white arrow). In addition, MRI can clarify intracranial infiltration in terms of nodular dural thickening or enhancement and cavernous sinus lesion. As our data showed, nodular dural thickening or enhancement alone is less frequent in patients FIGURE 3. Obvious intracranial invasion (white arrow) in two NPC patients demonstrated by CT (a, c) and T1-weighted contrast-enhanced MRI (b, d) (in coronal view). with intracranial invasion, occuring in only seven (3.5%) of 199 cases. Most of the patients initially had a cavernous sinus lesion plus nodular dural thickening or enhancement (cavernous sinus lesion alone, 79 of 199, 39.7%; cavernous sinus lesion together with nodular dural thickening or enhancement, 113 of 199, 56.8%) (Figure 4). This might be due to the nasopharynx neighboring the basal skull; the nasopharyngeal tumor tends to progress upward along the fascial planes to the base of the skull, resulting in skull base destruction and cavernous sinus infiltration. The pattern of spread from the pterygopalatine fossa along the maxillary nerve into the cavernous sinus of patients with NPC has been reported by Chong et al. 12 This should be a caution in clinical staging and evaluation, although the incidence is only 15%. The identification of pterygopalatine fossa invasion with intracranial extension is important because it serves as a guide to define the treatment field of radiation therapy so that better target coverage can be designed and eventually affect the treatment outcome. According to the AJCC staging system for NPC (Table 1), T4 lesions include tumor with intracranial extension and/or involvement of cranial nerves, infratemporal fossa, hypopharynx, orbit, and masticator space. Intracranial invasion involves the cavernous sinus more frequently Impact of MRI versus CT on NPC Treatment HEAD & NECK March 2004 245
primary tumor extension in NPC, and we support the greater use of MRI as the preferable imaging modality for the staging workup and evaluation of primary tumor extension. FIGURE 4. Patient distribution of intracranial infiltration detected by MRI. than the cranial fossa (20% vs 5%). 13 The spread of tumor along the vessels or nerves, through the foramina or even by direct invasion, is believed to be the route of intracranial spread. By MR images, the pterygopalatine fossa invasion is recognized by widening of the fossa and replacement of the fat to enhance the soft tissue mass. In our series, once the pterygopalatine fossa had been invaded, the incidence of intracranial invasion was very high (96.1%), although no statistical significance was disclosed ( p =.181 by Fisher s exact test). There was an observable trend for the NPC to display increased risk for intracranial invasion once the pterygopalatine fissure had been involved. It reminded us to pay attention to the intracranial region that might be ignored. Otherwise, the target would be missed and lead to a poor treatment outcome. In addition, the detection accuracy of intracranial invasion was more discernible by MRI in those cases with pterygopalatine fossa invasion (CT, 56.9% vs MRI, 96.1%). Accuracy in pretreatment target delineation is very important in NPC staging and treatment because radiation portals depend on the welldemonstrated tumor extent. It is also the main principle for achieving successful treatment outcome of NPC patients. MRI generally demonstrates the anatomy of NP and extent of NP tumor in greater detail than does CT. We believed that MRI was more sensitive than CT in identifying REFERENCES 1. Sham JST, Cheung YK, Choy D, Chan FL, Leong L. Nasopharyngeal carcinoma: CT evaluation of patterns of tumor spread. AJNR Am J Neuroradiol 1990;12:265 270. 2. Yamashita S, Kondo M, Hashimoto S. Conversion of T-stages of nasopharyngeal carcinoma by computed tomography. Int J Radiat Oncol Biol Phys 1985;11: 1017 1021. 3. Yu ZH, Xu GZ, Huang YR, Hu YH, Su XG, Gu XZ. Value of computed tomography in staging the primary lesion (T-staging) of nasopharyngeal carcinoma (NPC): an analysis of 54 patients with special reference to the parapharyngeal space. Int J Radiat Oncol Biol Phys 1985;11: 2143 2147. 4. Chong VFH, Fan YF, Khoo J BK. Nasopharyngeal carcinoma with intracranial spread: CT and MRI characteristics. J Comput Assist Tomogr 1996;20:563 569. 5. Hong RL, Ting LL, Ko JY, et al. Induction chemotherapy with mitomycin, epirubicin, cisplatin, fluorouracil, and leucovorin followed by radiotherapy in the treatment of locoregionally advanced nasopharyngeal carcinoma. J Clin Oncol 2001;19:4305 4313. 6. Lin JC, Jan JS, Hsu CY, Liang WM, Jiang RS, Wang WY. Phase III study of concurrent chemoradiotherapy versus radiotherapy alone for advanced nasopharyngeal carcinoma: positive effect on overall and progression-free survival. J Clin Oncol 2003;21:631 637. 7. Eisen MD, Yousem DM, Mointone KT, et al. Use of preoperative MR to predict dural, perineural, and venous sinus invasion of skull base tumors. AJNR Am J Neuroradiol 1996;17:1937 1945. 8. Modder U, Lenz M, Steinbrich W. MRI of facial skeleton and parapharyngeal space. Eur J Radiol 1987;7:6 10. 9. Ng SH, Chang TC, Ko SF, et al. Nasopharyngeal carcinoma: MRI and CT assessment. Neuroradiology 1997;39: 741 746. 10. Chong VFH, Fan YF. Skull base erosion in nasopharyngeal carcinoma: detection by CT and MRI. Clin Radiol 1996;51:625 631. 11. Curran WJ, Hackney DB, Blitzer PH, Bilaniuk L. The value of magnetic resonance imaging in treatment planning of nasopharyngeal carcinoma. Int J Radiat Oncol Biol Phys 1986;12:2189 2196. 12. Chong VFH, Fan YF. Pterygopalatine fossa and maxillary nerve infiltration in nasopharyngeal carcinoma. Head Neck 1997;19:121 125. 13. King AD, Lam WWM, Leung SF, Chan YL, Teo P, Metreweli C. MRI of local disease in nasopharyngeal carcinoma: tumour extent vs tumour stage. Br J Radiol 1999; 72:734 741. 246 Impact of MRI versus CT on NPC Treatment HEAD & NECK March 2004