European Journal of Radiology 43 (2002) 3741

Application of dynamic MRI to differentiating odontogenicmyxomas from ameloblastomas

Jun-ichi Asaumi *, Hidenobu Matsuzaki, Miki Hisatomi, Hironobu Konouchi,Hiroshi Shigehara, Kanji Kishi

Department of Oral and Maxillofacial Radiology, Field of Tumor Biology, Graduate School of Medicine and Dentistry,Okayama Uniersity Graduate Schools, 2-5-1 Shikata-cho, Okayama 700-8525, Japan

Received 30 July 2001; received in revised form 8 October 2001; accepted 10 October 2001

Abstract

It is often difficult to radiographically distingush odontogenic myxomas from ameloblastomas. In the present study, we triedto differentiate odontogenic myxomas from ameloblastomas using dynamic magnetic resonance imaging (dynamic MRI). Twocases of ameloblastoma with cystic companents and two cases of odontogenic myxoma were compared by dynamic MRI. Thedynamic MRI features of solid areas of ameloblastomas showed a rapid enhancement, reaching maximum contrast at 4560 s,and maintained these enhancement levels or showed a gradual wash-out to 600 s thereafter; in contrast, those of the cystic areasof ameloblastomas showed no enhancement. The dynamic MRI features of the whole area of odontogenic myxomas (weconsidered the whole area to be the tumor substance in the odontogenic myxomas, as based on histopathological examinations)showed a gradual increase in enhancement at 500600 s. The central portions of the odontogenic myxomas, which did not appearto be enhanced on Gd-T1 weighted images also showed a gradual increase in enhancement at 500600 s, though the increase wasminimal. These results indicate that the dynamic MRI features of odontogenic myxomas are different from those of ameloblas-tomas. Therefore, dynamic MRI may be a useful tool for diagnosis of myxoma. 2002 Elsevier Science Ireland Ltd. All rightsreserved.

Keywords: Dynamic MRI; Odontogenic myxoma; Ameloblastoma

www.elsevier.com/locate/ejrad

1. Introduction

Myxoma is a relatively rare tumor of mesenchymalorigin [1,2]. Radiographically, the odontogenic myxomacommonly shows multiple radiolucent areas of varyingsize separated by straight or curved bony septa (soap-bubble appearance) [3,4]. Computed tomography (CT)also shows a multilocular soft tissue mass with bonedestruction and thinning as well as strands of a fine,lacelike density [5]. The above appearance may beindistinguishable from that of an ameloblastoma, espe-cially when the bony septa of odontogenic myxoma areof the curved type rather than the straight type. There-fore, it may often be difficult to radiographycally distin-guish odontogenic myxomas from ameloblastomas,

even if CT was performed. Recently, magnetic reso-nance imaging (MRI) has been used in diagnosing anddefining soft tissue lesions in the oral and maxillofacialregion because it is the most useful modality for analyz-ing the internal structures of lesions with its superiorsoft tissue contrast and multiplanar facility [68]. Wehave previously described the characteristic MR fea-tures of odontogenic myxoma [5], noting that althoughboth the gross and microscopic features are visible byMRI, the signal intensities are not characteristic ofodontogenic myxomas alone. It is only the ability ofMRI to clearly show the erosive extension into adjacentstructures and invasion into the interroots of the teeth[5] that makes it possible to confirm that these lesionsare not consistent mass lesions such as ameloblastomas,as these lesions do not absorb or move the roots of theteeth. However, this type of erosion and invasionsoccurs only in relatively advanced lesions of the maxil-lary sinus. As such, it may be difficult to confirm this

* Corresponding author. Tel.: +81-86-235-6705; fax: +81-86-235-6709.

E-mail address: asaumi@md.okayama-u.ac.jp (J. Asaumi).

0720-048X/02/$ - see front matter 2002 Elsevier Science Ireland Ltd. All rights reserved.PII: S0720 -048X(01 )00453 -3

J. Asaumi et al. / European Journal of Radiology 43 (2002) 374138

characteristic feature in small lesions of odontogenicmyxoma.

There have been many reports that dynamic MRI isuseful for differential diagnosis of some tumors [912],and many investigators have attempted to differentiatebenign from malignant tumors and to assess malignancywith dynamic MRI [9,12]. Dynamic MRI may be usefulin predicting the biological behavior of some tumors andin making a diagnosis. However, to our knowledge therehave been no reports of the use of dynamic MRI withmyxomas. The purpose of the present study was tocompare the dynamic MRI of odontogenic myxoma tothat of ameloblastoma in order to distinguish odonto-genic myxomas from ameloblastomas.

2. Materials and methods

2.1. Cases

Two cases of ameloblastoma with cystic lesions andtwo cases of odontogenic myxoma.

2.2. Dynamic MRI

The MR examinations were carried out with a 1.5 teslaunit (Magnetom Vision; Siemens, Erlangen, Germany)with a CP head coil. Dynamic MRI was performed usingthree-dimensional fast imaging with a steady-state pro-cession sequence (TR/TE=5/2 ms). GdDTPA (Mag-nevist; Nihon Schering, Osaka, Japan) was injected as abolus at approximately 2 ml/s. The beginning of the firstscan was designated as time 0, and the administration ofGdDTPA was started at 6 s before the second scan.Dynamic MRI were acquired from 20 consecutive scansat 1-s intervals (14 s/scan).

2.3. Image analysis

The region of interest (ROI) was drawn in the twopatterns of each lesion using the elective cursor on themonitor (Figs. 14). In the odontogenic myxoma, theROI was drawn to include the whole area, as weconsidered the whole area to be tumor substance in thesetwo cases of odontogenic myxoma based on histopatho-logical examinations. Another ROI was drawn to includeonly the central portion, as only the peripheral portionwas strongly enhanced and the central portion appearedto be no enhancement because the central portion wasweakly enhanced compared with the peripheral portionon Gd-T1 weighted images (WI) (Figs. 1 and 2). In theameloblastomas, the ROI was drawn to include the solidarea or the cystic area, which were proved histopatholog-ically (Figs. 3 and 4). The mean signal intensity on theROI of each lesion was calculated using a workstation(Siemens). The contrast index (CI) was calculated from:

Fig. 1. The region of interest (ROI) in case 1 (odontogenic myxoma)whole area; large area, central portion; small area.

CI= (signal intensity (postcontrast)signal intensity(precontrast))/signal intensity (precontrast). The timecourse of the CI (CI curves) was obtained by plotting theCI on a time course.

3. Results

The CI curves, which were calculated from dy-namic series, are shown in Fig. 5. The whole area ofboth cases of odontogenic myxomas showed a gradualincrease of enhancement at 500600 s (case 1: , case2:). The central portions of both cases of odontogenic

Fig. 2. The ROI in case 2 (odontogenic myxoma) whole area; largearea, central portion; small area.

J. Asaumi et al. / European Journal of Radiology 43 (2002) 3741 39

Fig. 3. The ROI in case 3 (ameloblastoma) solid area; large area,cystic area; small area.

diagnosis. However, there have been few reports ofthe use of dynamic MRI in the head and neck re-gions. The CI curves represent the blood behavior,therefore the enhancement pattern of dynamic MRImay reflect the intratumoral angiogenesis [1316]. Ithas been reported that the CI curves of benign tu-mors increase gradually, while those of malignant tu-mors increase rapidly [17]. The dynamic MRI featuresof some tumors in the salivary glands have been re-ported previously [10,12,18,19]. The CI curve ofpleomorphic adenomas shows a gradual increased en-hancement [10,12], and those of Warthins tumorsshow rapid enhancement, reaching a maximum CI,decreasing rapidly, and then undergoing a gradualwash-out [10,11,19]. These dynamic MRI featureshave characteristic features, and they may useful inmaking a differential diagnosis. In the present study,we examined the dynamic MRI features of ameloblas-toma and odontogenic myxoma, which are difficult todifferentiate using radiographic examinations. Thesolid areas of ameloblastoma show an earlier en-hancement than whole areas of odontogenic myxoma(we considered the whole area to be tumor substancein the odontogenic myxoma). The solid areas ofameloblastoma showed a gradual wash-out or main-tained enhancement levels at 600 s after reaching themaximum enhancement levels, while the whole areasof both cases of odontogenic myxomas showed agradual increase of enhancement at 500600 s. Theseresults indicate that the dynamic MRI features of thetumor substance of ameloblastoma differs from thoseof odontogenic myxomas. Moreover, we compareddynamic MRI features in the cystic area ofameloblastoma and those in the central portions ofodontogenic myxoma, which did not appear to beenhanced on Gd-enhanced MRI. The cystic areas ofboth cases of ameloblastoma showed no enhance-ment, while the central portion of both cases, whichappeared to be the cyst-like area in the odontogenicmyxomas, showed a gradual increase in enhancementat 500600 s, though the increase was minimal. Theseresults indicate that the dynamic MRI features of thecystic area of ameloblastomas differ from the centralcyst-like area of odontogenic myxomas. Thus, dy-namic MRI features can differentiate ameloblastomasfrom odontogenic myxomas in both solid and cysticareas of ameloblastomas or cyst-like areas of odonto-genic myxomas, making the use of dynamic MRI forthis purpose clinically useful. Minami et al. reportedthat Gd-enhanced MRI of ameloblastomas showsstrong enhancement in the solid portions of the tu-mor, including papillary projections, walls, and septa,but it shows no enhancement in the fluids, which arenot hemorrhagic but soft and gelatinous [20]. Thisreport suggested our dynamic MRI findings ofameloblastomas. Kawai et al. reported that Gd-en-

myxoma showed a gradual increase in enhancementto 500600 s, though the increase was minimal (case1: , case 2: ). The solid area in a case ofameloblastoma showed a rapid enhancement, reachingmaximum CI at 45 s, and exhibiting a gradual wash-out thereafter (case 3: ). The solid area of anothercase of ameloblastoma showed rapid enhancement,reaching a plateau at 60 s, and then maintaining theenhancement level at 600 s (case 4: ). The cysticareas of both cases of ameloblastoma showed no en-hancement (case 3: , case 4: ).

4. Discussion

Dynamic MRI may be useful in predicting the bio-logical behavior of some tumors and in making a

Fig. 4. The ROI in case 4 (ameloblastoma) solid area; large area,cystic area; small area.

J. Asaumi et al. / European Journal of Radiology 43 (2002) 374140

Fig. 5. The contrast index curves on the two cases of odontogenic myxoma and two cases of ameloblastoma in the ROIs shown in Figs. 14.Symbols; case 1 (myxoma) whole portion: , central portion: , case 2 (myxoma) whole portion: , central portion: , case 3 (ameloblastoma)solid area: , cystic area: , case 4 (ameloblastoma) solid area: , cystic area: .

hanced MRI of the odontogenic myxoma shows ho-mogeneous high signal intensity although their case ofmyxoma consists of two parts histopathologically,which are a small amount of collagen and fibroblasticproliferation and a scarcely cellular mucoid matrixwhich spindle-shaped and stellate cells are sparselyscattered [21]. On the other hand, we reported previ-ously that in the Gd-enhanced MRI of the odonto-genic myxoma, the peripheral portion of the lesionwith a relatively large quantity of collagen bundleswas slowly enhanced, while the central portion withonly mucoid component was not. It was consideredthat the central portion of the odontogenic myxomasappeared the cyst-like area because the mucoid com-ponent showed slow and weak enhancement. Inodontogenic myxoma, it might take longer time inthe large lesion than in the small lesion because thecontrast agent invaded from the peripheral region ofthe lesion. This may be the reason of the discrepancybetween Kawais case and our previous case of theodontogenic myxoma

In the present study, we tried to differentiate odon-togenic myxomas from ameloblastomas using dynamicMRI. In the ameloblastoma, GdDTPA, the contrastagent, did not enter the cystic part of the tumor, butrapidly entered into the tumor substance. While inthe odontogenic myxoma, the GdDTPA entered thecentral portion, which appeared to be the cyst-likeareas, as well as the peripheral portion, which wasstrongly enhanced on the Gd-enhanced MRI. Thisfinding indicates that dynamic MRI can distinguishameloblastomas from odontogenic myxomas and maybe a useful tool for diagnosis of myxoma.

References

[1] Takahashi H, Fujita S, Okabe H. Immunohistochemical investiga-tion in odontogenic myxoma. J Oral Pathol Med 1991;20:1149.

[2] Green TL, Leighty SM, Waiters R. Immunohistochemical evalu-ation of oral myxoid lesions. Oral Surg Oral Med Oral Pathol1992;73:46971.

[3] Peltola J, Magnusson B, Happonen RP, Borrman H. Odontogenicmyxomaa radiographic study of 21 tumours. Br J Oral Maxillo-fac Surg 1994;32:298302.

[4] Kaffe I, Naor H, Buchner A. Clinical and radiological features ofodontogenic myxoma of the jaws. Dentomaxillofac Radiol1997;26:299303.

[5] Asaumi J, Konouchi H, Hisatomi M, Kishi K. Odontogenicmyxoma of maxillary sinus: CT and MRI-pathologic correlation.Eur J Radiol 2001;37:14.

[6] Lenz M. Computed Tomography and Magnetic Resonance Imag-ing of Head and Neck Tumors. Stuttgart: Thieme, 1993.

[7] Lenz M. Imaging of head and neck tumors. Eur J Radiol2000;33:1512.

[8] Lenz M, Greess H, Baum U, Dobritz M, Kersting-SommerhoffB. Oropharynx, oral cavity, floor of the mouth: CT and MRI. EurJ Radiol 2000;33:20315.

[9] Ikarashi F, Nakano Y, Nonomura N, Kawana M. Radiologicalfindings of adenolymphoma (Warthins tumor). Auris NasusLarynx 1997;24:4059.

[10] Tsushima Y, Matsumoto M, Endo K. Parotid and parapharyngealtumours: tissue characterization with dynamic magnetic resonanceimaging. Br J Radiol 1994;67:3425.

[11] Takashima S, Noguchi Y, Okumura T, Aruga H, Kobayashi T.Dynamic MR imaging in the head and neck. Radiology1993;189:81321.

[12] Park J, Inoue S, Ishizuka Y, et al. Salivary gland masses: dynamicMR imaging and pathologic correlation. Nippon Igaku HoshasenGakkai Zasshi 1997;57:5815 (in Japanese).

[13] Hulka CA, Edmister WB, Smith BL, Tan L, Sgroi DC, CampbellT, Kopans DB, Weisskoff RM. Dynamic echo-planar imaging ofthe breast: experience in diagnosing breast carcinoma and corre-lation with tumor angiogenesis. Radiology 1997;205:83742.

J. Asaumi et al. / European Journal of Radiology 43 (2002) 3741 41

[14] Hulka CA, Smith BL, Sgroi DC, Tan L, Edmister WB, SempleJP, Campbell T, Kopans DB, Brady TJ, Weisskoff RM. Benignand malignant breast lesions: differentiation with echo-planarMR imaging. Radiology 1995;197:338.

[15] Hawighorst H, Knapstein PG, Weikel W, Knopp MV, Zuna I,Knof A, Brix G, Schaeffer U, Wilkens C, Schoenberg SO, EssigM, Vaupel P, van Kaick G. Angiogenesis of uterine cervicalcarcinoma: characterization by pharmacokinetic magnetic reso-nance parameters and histological microvessel density with cor-relation to lymphatic involvement. Cancer Res 1997;57:477786.

[16] Buckley DL, Drew PJ, Mussurakis S, Monson JR, Horsman A.Microvessel density of invasive breast cancer assessed by dy-namic GdDTPA enhanced MRI. J Magn Reson Imaging1997;7:4614.

[17] Eveson JW, Cawson RA. Warthins tumor (cystade-nolymphoma) of salivary glands. A clinicopathologic investiga-

tion of 278 cases. Oral Surg Oral Med Oral Pathol1986;61:25662.

[18] Nishimura H, Kojima K, Abe T. Imaging of neck masses. Jpn JDiagn Imaging 1998;18:53348.

[19] Hisatomi M, Asaumi J, Konouchi H, Yanagi Y, Matsuzaki H,Kishi K. Assessment of dynamic MRI of Warthins tumorsarising as multiple lesions in the parptid glands. Oral Oncol (inpress).

[20] Minami M, Kaneda T, Yamamoto H, Ozawa K, Itai Y, OzawaM, Yoshikawa K, Sasaki Y. Ameloblastoma in the maxillo-mandibular region: MR imaging. Radiology 1992;184:38993.

[21] Kawai T, Murakami S, Nishiyama H, Kishino M, Sakuda M,Fuchihata H. Diagnostic imaging for a case of maxillary myx-oma with a review of the magnetic resonance images of myxoidlesions. Oral Surg Oral Med Oral Pathol Oral Radiol Endod1997;84:44954.