Neuroradiology (1997) 39: 741–746  Springer-Verlag 1997

S. H. Ng T. C. Chang S. F. Ko P. S. Yen Y. L. Wan L. M. Tang M. H. Tsai

Received: 18 November 1996 Accepted: 22 January 1997

)

S. H. Ng ( ) ⋅ T. C. Chang ⋅ S. F. Ko ⋅ P. S. Yen ⋅ Y. L. Wan Department of Diagnostic Radiology, Chang Gung Memorial Hospital, 5 Fu-Shing Street, Kwei Shan, Tao Yuan, Taiwan R. O. C. Tel. +8 86 3 3 28 12 00 ext. 25 74; fax +8 86 3 3 97 19 36; e-mail r2201257 @ tpts1.seed.net.tw L. M. Tang Department of Neurology, Chang Gung Medical College and Chang Gung Memorial Hospital at Taipei, Taiwan M. S. Tsai Department of Otorhinolaryngology, Chang Gung Medical College and Chang Gung Memorial Hospital at Taipei, Taiwan

H EA D A N D N ECK R A DIO LO G Y

Nasopharyngeal carcinoma: MRI and CT assessment

Abstract Precise assessment of the extent of nasopharyngeal carcinoma (NPC) represents the basic step towards optimal treatment. We compared the capacity of CT and MRI in assessing the extent of NPC in 67 patients. MRI was superior to CT in demonstrating lesions in the retropharyngeal node, skull base, intracranial area, carotid space, longus colli muscle and levator palatini muscle. Of 25 cases in which retropharyngeal adenopathy was recognised only on MRI, seven had been reported as showing oropharyngeal involvement and 18 as primary extension to the carotid space on CT. MRI showed skull-base involvement in 40 patients compared with 27 on CT and intracranial involvement in 38 patients versus 24 on CT. There was not a single case in

Introduction With advances in magnetic resonance imaging (MRI) technology, its role in the assessment of head and neck lesions has been increasingly important. MRI has the advantages of excellent tissue contrast, multiplanar capacity, and lack of radiation and bone beam-hardening artefact. It may supplant computed tomography (CT) in the assessment of the head and neck malignancy [1–4]. However, MRI is inferior to CT for delineating cortical bone detail [3, 5–7]. In one report [7], it failed to reveal skull base invasion in four patients with nasopharyngeal carcinoma (NPC). The decision to use primarily MRI rather than CT in the investigation of NPC is controversial.

which skull base invasion was seen on CT but not on MRI. MRI enabled improved recognition of tumour infiltration of longus colli muscles (34 cases compared with 15 on CT). It allowed us to clarify 12 questionable sinonasal opacities on CT. Overall, T-staging was changed in 18 of 67 patients (26.9 %), including upstaging in 15 cases and downstaging in 3 cases, after comparing CT with MRI. The nodel status was changed from negative on CT to positive on MRI in 4 of 67 patients (6 %). We believe that MRI allows more accurate evaluation of the extent of NPC than CT and should be the primary mode of investigation. Key words Computed tomography ⋅ Magnetic resonance imaging ⋅ Nasopharyngeal carcinoma

In this study, we evaluated CT and MRI for staging of NPC and compared their ability to show involvement of the nasopharynx and its adjacent spaces.

Materials and methods From 1993 to 1996, CT and MRI were used to study 67 patients with pathologically-proven NPC. They were 48 men and 19 women, aged 14 to 74 years (mean, 47.8 years). CT was performed after intravenous injection of 100 ml of iodinated contrast medium, with the patient supine and the sections parallel to the hard palate. The scan extended from the cavernous sinus to the soft palate in 5 mm thick contiguous sections, then to the level of the sternal notch in 10 mm thick contiguous sections. Direct coronal scans were also performed whenever possible, with 5 mm contiguous

742

Table 1 Sites involved in 67 patients with nasopharyngeal carcinoma (NPC); NS, not significant

a

equivocal cases test

b

Pearson’s c2

P valueb

Number of patients (%) Lateral wall Vault/posterior wall Levator palatini muscle Choanae Nasal cavity Parapharyngeal space Carotid space Longus colli muscle Oropharynx Sphenoid sinus Ethmoid sinus Maxillary sinus Infratemporal fossa Orbit Skull base Intracranial Retropharyngeal node Cervical node

sections, perpendicular to the hard palate. Images were displayed with both the soft tissue and bone algorithms. MRI was performed at 1.5 Tesla units, using a standard head coil. A spin-echo (SE) technique was used. All patients underwent MRI before and after Gadolinium-DTPA (Gd-DTPA) injection. T1-weighted images were acquired in sagittal, axial and coronal planes with an SE 500/20 sequence, a 20-cm field of view, and a 192 × 256 matrix. Section thickness was 5 mm with a 2.5-mm intersection gap in the axial plane and 4 mm with a 1-mm gap in the sagittal and coronal planes. Axial T2-weighted fat-suppressed fast SE images (3000/85 [effective], 16 echo-train length) were also obtained. After Gd-DTPA, 0.1 mmol per kg body weight, T1weighted fat-suppressed axial and coronal sequences were performed, with similar parameters as before injection. Fat suppression was achieved by presaturation. All films were reviewed individually by two radiologists. The spread of NPC and the involvement of adjacent anatomic structures were examined in detail with a checklist. Any discrepancies between the two radiologists were resolved by consensus. The CT and MRI findings were compared, to assess concurrence and discordance between the two modalities.

Results There was no significant difference between CT and MRI for demonstrating the involvement by NPC in the nasopharyngeal subsites, the choanae, maxillary sinus, infratemporal fossa, parapharyngeal space or orbit (Table 1). A minor difference was noted in defining the caudal extension of the tumour to the oropharynx and the cephalad spread to the sphenoethmoid sinuses. The difference was significant for lesions in the retropharyngeal node, skull base, intracranial area, carotid space, levator palatini muscle and longus colli muscle. Of 18 patients, in whom CT suggested oropharyngeal involvement, seven actually had retro-

CT

MRI

57 (85) 49 (59.7) 31+18a (46.3+26.9a) 18 (14.9) 8+1a (11.9+1.5a) 46 (68.7) 43 (64.2) 15 (22.4) 18 (26.9) 12+8a (17.9+11.9a) 3+3a (4.5+4.5a) 3 (4.5) 9 (13.4) 3 (4.5) 27 (40.3) 24 (35.8) 14 (20.9) 38 (56.7)

57 (85) 49 (59.7) 46 (68.7) 18 (14.9) 8 (11.9) 52 (77.6) 26 (38.8) 34 (50.7) 11 (16.4) 16 (23.9) 3 (4.5) 3 (4.5) 10 (14.9) 3 (4.5) 40 (59.7) 38 (56.7) 39 (58.2) 42 (62.7)

NS NS 0.009 NS NS NS 0.003 0.001 NS NS NS NS NS NS 0.025 0.015 < 0.001 NS

pharyngeal adenopathy disclosed by MRI (Fig. 1). MRI enabled an improved recognition of tumour infiltration of longus colli muscles (34 cases compared with 15 on CT) (Fig. 2). Using CT, we were unable to differentiate sinonasal secretions from tumour invasion in 12 cases with questionable sinonasal opacities: 8 in the sphenoid, 3 in the ethmoid sinus and 1 in the nasal cavity. The distinction appeared evident on MRI (Fig. 3). MRI was significantly more sensitive than CT in revealing skull base involvement (40 cases versus 27) (Fig. 4) and intracranial invasion (38 cases versus 24) (Fig. 5). In all 27 cases, in which skull base invasion was detected on CT, it was demonstrated clearly on MRI. MRI allowed better detection of retropharyngeal nodal metastases than CT (39 cases versus 14). In 25 cases in which retropharyngeal adenopathy was recognised only on MRI, 7 (mentioned above) had been interpreted on CT as oropharyngeal involvement and 18 as primary extension to the carotid space (Fig. 6). Of these cases, 3 were initially interpreted as nodal negative on CT and re-classified as positive after MRI. In detection of cervical adenopathy, CT and MRI were comparable (38 cases versus 42). On the staging system of the American Joint Committee on Cancer (AJCC), upstaging occurred in 15 cases after comparing CT with MRI, including 11 cases changed from T2 and 4 from T3 to T4. Of these 15 cases, we found 14 to have skull base involvement and one to have intracranial invasion on MRI. Downstaging occurred in 3 cases; instead of the oropharyngeal involvement suggested by CT, retropharyngeal lymphadenopathy was recognised on MRI. Using MRI, the overall T-staging determined by CT changed in 18 of 67 patients (26.9 %). In addition, the overall nodal status changed

743

Fig. 1 Nasopharyngeal carcinoma (NPC) with retropharyngeal adenopathy mimicking oropharyngeal extension. Contrast-enhanced CT a shows a soft-tissue mass extending from the left nasopharynx to the oropharynx. Contrast-enhanced fat-suppressed T1-weighted image b shows the enhancing nasopharyngeal carcinoma and the large retropharyngeal nodes (arrows)

1a

b

2a

b

3a

b

Fig. 2 NPC with longus colli muscle invasion. Infiltration of the muscle is clearly depicted on the contrast-enhanced fat suppressed T1-weighted image a but not on contrast-enhanced CT b

Fig. 3 NPC with paranasal sinus involvement and sinusitis. Opacification of the sphenoid and ethmoid sinuses on CT a can be clearly differentiated into an intermediate-signal tumour and high-signal obstructive sinusitis on a T2weighted image b

from negative on CT to positive on MRI in 4 patients (6 %); this was due to detection on MRI of retropharyngeal adenopathy in 3 and cervical adenopathy in one.

Discussion Since NPC is diagnosed by endoscopy, the foremost role of CT or MRI is to determine the extent of primary tumour and the presence of metastatic adenopathy [8]. Accurate assessment of disease extent facilitates appropriate treatment planning and prognosis [9]. In our

744

4a

b

c

d

5a

5b

6a

6b

Fig. 4 NPC with clivus invasion. CT a on bone window setting shows no abnormality in the cortex or trabecular pattern of the clivus b sagittal c axial and d contrast-enhanced fatsuppressed axial T1-weighted images show replacement of the marrow of the clivus by contrast-enhancing tumour Fig. 5 NPC with intracranial invasion. Coronal contrast-enhanced CT a shows no intracranial lesion. MRI b shows dural enhancement in the anterior cranial fossa Fig. 6 NPC with retropharyngeal adenopathy mimicking carotid space invasion. Axial contrast-enhanced CT a shows a soft-tissue mass in the right carotid space, contiguous with the nasopharyngeal tumour. A fat-suppressed T2weighted image b shows a retropharyngeal node (arrow) instead

study, CT and MRI findings were essentially in agreement in patients whose disease was limited to the nasopharyngeal cavity, but not those with tumour spreading beyond the boundaries of the nasopharynx. Although involvement of the nasopharyngeal subsites was shown to a similar degree on both CT and MRI, submucosal extension to palatini muscles was better appreciated on MRI. The pharyngobasilar fascia, the medial border of the parapharyngeal space, is normally seen on MRI and not on CT [2, 3, 10, 11]. In NPC, parapharyngeal space involvement can be assessed directly by MRI, which shows tumour displacement or infiltration of the pharyngobasilar fascia or extension through the sinus of Morgagni [2, 6]. In contrast, involvement of the para-

745

pharyngeal space by CT is inferred indirectly by an abnormal soft tissue deforming the parapharyngeal fibrofatty tissue plane or by outward bulging of an imaginary line between the medial pterygoid plate and the lateral border of the carotid artery [5, 11]. Contrary to the experience of Sham et al. [5], we demonstrated that CT assessment of parapharyngeal involvement could be false-negative, as in 6 of our 67 cases (9 %) parapharyngeal extension, undiagnosed by CT, were demonstrated by MRI. Also, MRI excelled CT in displaying tumour infiltration of longus calli muscles of the head and neck, in our study. As infiltration of longus colli muscles may not be recognised on CT, it has not been included for staging or prognostic prediction of NPC. With increasing application of MRI, the significance of longus colli muscle infiltration can be defined in further studies. In patients with NPC, sinonasal opacity is common on CT; it is occasionally difficult to differentiate whether it represents tumour invasion or sinonasal secretions. On MRI, hydrated secretions within the obstructed paranasal sinuses are of increased signal on T2weighted images. Thus, high-signal secretions can be differentiated from an intermediate signal-intensity tumour [3]. Desiccated or mixed sinonasal secretions may exhibit signal characteristics similar to those of tumour on both T1- and T2-weighted images; contrast-enhanced MRI is then helpful because tumour within the sinuses enhances whereas sinonasal secretions do not enhance and are surrounded by a rim of strongly enhancing sinus mucosa [12]. Accurate differentiation of NPC involvement of the sinonasal system by MRI facilitated the design of radiation ports in 12 of our patients. CT can directly determine the extent of cortical bone destruction and/or remodelling by cancer [3–7]. In 53 patients with NPC studied by Olmi et al. [7], CT showed skull base erosion in 12 patients and MRI in 8. On the other hand, MRI can show tumour involvement of the skull base as a lesion with different signal intensities encroaching on the signal-void bone cortex or replacing the marrow [3, 6]. Contrast-enhanced fat-suppressed MRI provides a better delineation of tumour extension into the clivus and allows discrimination of tumour invasion from oedema of the marrow [13]. Teresi et al. [6] claimed that skull-base destruction was not misinterpreted with MRI although no explicit data were given. In our series, skull base destruction was revealed in 27 of 67 cases (40.3 %) on CT and in 40 cases (59.7 %) on MRI. There was no case in which the skull-base invasion was not visible on MRI, while there were 13 cases (19.4 %) in which it was detected only by MRI. A plausible explanation for the difference between our results and those of Olmi et al. [7] is that contrast-enhanced fatsuppressed MRI was performed in all of our cases, enabling better delineation of skull-base masses, while

only 8 of 53 patients received Gd-DTPA in the series of Olmi et al. Our study highlights another superiority of MRI over CT, revealing intracranial invasion [3, 4, 6, 8, 14], especially in conjunction with the contrast-enhanced fatsuppression technique [4, 13]. In our series, MRI but not CT depicted invasion of the middle cranial fossa in five patients, posterior cranial fossa in eight and anterior cranial fossa in one. Intracranial invasion by NPC frequently appears as a contrast-enhancing lesion juxtaposed to the cavernous sinus [5, 6]. Posterior cranial fossa invasion is less frequent and may result from breakthrough of the posterior cortex of the clivus, or extension via the petro-occipital synchondrosis or jugular foramen [14]. It is seen more readily with MRI which offers superior tissue contrast, pluridirectional scanning and does not show beam-hardening artefact from the dense bone of the skull base. Anterior cranial fossa invasion by NPC via the cribriform plate was seen in one of our patients (Fig. 5). To our knowledge, this has not been reported previously. Recognition of skull base involvement and intracranial invasion may facilitate the diagnosis of supervening bacterial meningitis [15], and planning of aggressive radiotherapy [9] or concomitant chemotherapy [16]. As patients with NPC may be clinically nodal negative at presentation, a contribution of modern imaging is to detect clinically occult lymphadenopathy. Our study validates the unique ability of MRI to identify retropharyngeal adenopathy, which could be misdiagnosed on CT as invasion of the oropharynx or carotid space. In assessing cervical nodal metastases, MRI is currently as good as or better than CT [13, 17]. The fat-suppression technique, in conjunction with unenhanced T2-weighted and contrast-enhanced T1-weighted images, facilitates the detection of lymphadenopathy and the presence of nodal necrosis or extracapsular nodal spread [2, 12, 13, 17]. The nodal status is of prognostic value because nodal-positive patients are prone to have distant metastasis [9, 18]. We showed that MRI provided a more accurate assessment of the extent of NPC than CT, which did not reveal additional significant information. We believe that MRI can consistently yield sufficient diagnostic information, and the additional expense of two imaging methods may not be justified. In our opinion, MRI is good enough to be the single investigation in patients with NPC.

746

References 1. Vogl T (1989) Disease of the aerodigestive tract and of the soft tissues of the neck: comparison of MRI and CT. Rontgenbla¨tter 42: 199–209 2. Vogl T, Dresel S, Bilaniuk LT, Grevers G, Kang K, Lissner J (1990) Tumors of nasopharynx and adjacent areas: MR imaging with Gd-DTPA. AJNR 11: 187–194 3. Kraus DH, Lanzieri CF, Wanamaker JR, Little JR, Lavertu P (1992) Complementary use of computed tomography and magnetic resonance imaging in assessing skull base lesions. Laryngoscope 102: 623–629 4. Brekel MWM, Castelijns JA, Snow GB (1994) The role of modern imaging studies in staging and therapy of head and neck neoplasms. Semin Oncol 21: 340– 348 5. Sham JST, Cheung YK, Choy D, Chan FL, Leong LLY (1991) Nasopharyngeal carcinoma: CT evaluation of patterns of tumor spread. AJNR 12: 265–270 6. Teresi LM, Lufkin RB, Vin˜uela F et al (1987) MR imaging of nasopharynx and floor of the middle cranial fossa. Part II. Malignant tumor. Radiology 164: 817– 821

7. Olmi P, Fallai C, Colagradne S, Giannardi G (1995) Staging and follow up of nasopharyngeal carcinoma: magnetic resonance imaging versus computerized tomography. Int J Radiat Oncol Biol Phys 32: 795–800 8. Foust RJ, Doung RT (1991) Roles of computed tomography and magnetic resonance imaging diagnoses in the treatment of head and neck cancer. Hematol/Oncol Clin North Am 5: 657– 665 9. Teo P, Shiu W, Leung SF, Lee EY (1992) Prognostic factors in nasopharyngeal carcinoma investigated by computed tomography – an analysis of 659 patients. Radiother Oncol 23: 79–93 10. Teresi LM, Lufkin RB, Vin˜uela F et al (1987) MR imaging of nasopharynx and floor of the middle cranial fossa. Part I. Normal anatomy. Radiology 164: 811– 816 11. Leung YK, Sham JST, Chan FL, Leong LLY, Choy D (1992) Computed tomography of paranasopharyngeal spaces: normal variation and criteria for tumor extension. Clin Radiol 45: 109–113 12. Crawford SC, Harnsberger HR, Lufkin RB, Hanafee WN (1989) The role of gadolinium-DTPA in the evaluation of extracranial head and neck mass lesions. Radiol Clin North Am 27: 219– 242

13. Barakos JA, Dillon WP, Chew WM (1991) Orbit, skull base and pharynx: contrast-enhanced fat suppression MR imaging. Radiology 179: 191–198 14. Mineura K, Kowada M, Tomura N (1991) Perineural extension of nasopharyngeal carcinoma into the posterior cranial fossa detected by magnetic resonance imaging. Clin Imaging 15: 172–175 15. Tang LM, Chen ST, Ng SH (1996) Bacterial meningitis in patients with nasopharyngeal carcinoma. Q J Med 89: 71– 76 16. Rotman M, Asis H (1990) Concomitant continuous infusion chemotherapy and radiation. Cancer 65: 823–835 17. Hillsamer PJ, Schuller DE, McGhee RB, Chakeres D, Young DC (1990) Improving diagnostic accuracy of cervical metastasis with computed tomography and magnetic resonance imaging. Arch Otolaryngol Head Neck Surg 116: 1297–1301 18. Lee AW, Poon YF, Foo W et al (1992) Retrospective analysis of 5037 patients with nasopharyngeal carcinoma treated during 1976–1985: overall survival and pattern of failure. Int J Oncol Biol Phys 23: 261–270