439 MR of Craniopharyngiomas: Tumor Delineation and Charaterization Elizabeth Pusey' Keith E. Kortman 2 onnie D. Flannigan, 3 Jay Tsuruda 2 William G. radley2 MR imaging and CT (with and without ontrast enhanement) were performed in 20 patients with an established or linially suspeted diagnosis of raniopharyngioma. Fifteen had biopsy-proven raniopharyngioma and five had presumed raniopharyngioma based on linial and CT findings. In two ases MR was superior to ontrastenhaned CT in demonstrating the tumor. variable appearane on T1-weighted MR images reflefed-the pathologi appearane of raniopharyngiomas. High intensity on T1-weighted images orresponded to hi h holesterol ontent or presene 0 met e mo 10blO. M was e preferred method in the evaluation of tumor extent, espeially i ~ t e avernous sinus and posterior lival region. CT was superior to MR in deteting the presene of ~ n, whih with the linial history orretly suggested the diagnosis of ranlop aryngloma. oth MR and CT studies are desired Initially to estabhsh the diagnosis and to evaluate tumor extent. MR was the preferred method in deteting the presene of reurrent tumor. Clinial experiene with MR has demonstrated its superiority over CT in the evaluation of many brain tumors, and it has been ited by some authors as the examination of hoie in sreening and delineating tumor extent [1-3] due to its inreased sensitivity and diret multi planar apabilities. The variable histologic appearanes of raniopharyngiomas and the diffiulty in identifying alifiations with MR has ast doubt on its appliability in suspeted raniopharyngioma. Previous reports have inluded isolated ases in larger studies of brain tumors [4-8]. eause the spetrum of MR appearanes of this tumor has not previously been desribed fully, we orrelated the results of MR, CT, and histopathology in 20 patients to haraterize the MR appearane and to determine its role in diagnosis. This artile appears in the May/June 1987 issue of JNR and the ugust 1987 issue of JR. Reeived June 27, 1986; aepted after revision Otober 5, 1986. Presented at the annual meeting of the merian Roentgen Ray Soiety, Washington, DC, pril 1986. 1 Department of Radiologi Sienes, University of California, Los ngeles, Center for Health Sienes, Los ngeles, C 90024. ddress reprint requests to E. Pusey. 2 Department of Radiology, Huntington Memorial Hospital, and NMR Imaging Laboratory, Huntington Medial Researh Institutes, Pasadena, C 91105. 3 Present address: Department of Radiology, Valley Presbyterian Hospital, Van Nuys, C 91405. JNR 8:439-444, May/June 1987 0195-6108/87/0803-0439 merian Soiety of Neuroradiology Subjets and Methods Twenty-seven MR studies were performed in 20 patients 7-66 years old. Fifteen patients with surgial proof and five with linial and CT diagnosis of raniopharyngioma are inluded in the study. MR was performed as part of the preoperative workup in 13 patients. In seven patients it was performed after biopsy or partial surgial resetion. MR and CT studies were performed in all ases. MR studies were performed on one of three imagers. Relative T1-weighted and T2-weighted spin-eho (SE) sequenes were obtained in all patients. Sixteen patients were sanned with a 0.35-T sanner (Oiasonis). Multislie multieho images were obtained with a repetition time (TR) of 1000 or 2000 mse and an eho time (TE) of 28 or 56 mse. Three patients were sanned with a 0.3-T imager (Fonar) Multislie single-eho images were obtained with TR = 500/2000 mse, TE = 28/56 mse. Two patients were sanned on a 0.5-T instrument (Piker) Multislie, dual-eho images were obtained with TR = 1000/2000 mse, TE = 40/80 or 60/120 mse. Images were obtained in at least two planes. Surgial pathology reports were reviewed in all ases. Surgial pathologi speimens were reviewed by the authors in six ases. In one ase analysis of the hemosiderin ontent of the speimen was performed by ultraviolet light spetrosopy.
440 PUSEY ET L. JNR:8, May/June 1987 Results MR studies demonstrated a wide range of appearanes of raniopharyngiomas, espeially on T1-weighted images. In nine patients the tumor showed an inrease in intensity on T1-weighted images onsistent with T1 shortening (Fig. 1). Ten studies demonstrated a tumor of intermediate intensity on T1-weighted images (Fig. 2) with signal intensity greater than CSF and variable homogeneity (Fig. 3). In one ase a ysti tumor showed dereased intensity on T1-weighted images (Fig. 4). The raniopharyngiomas were of high signal intensity on T2-weighted images in all but one ase, suggesting a long T2 relaxation time (Figs. 18 and 28). Pathologi speimens were reviewed in an attempt to orrelate the varying MR appearanes with the wide histologi appearanes of raniopharyngiomas. High-intensity signal on T1 -weighted images was found to orrespond to a high liquid holesterol ontent in the tumor, as demonstrated pathologially (Fig. 1), or to the presene of methemoglobin, as demonstrated by ultraviolet visible spetrosopy (Fig. 5). Moderate-intensity signal on T1-weighted images was found in tumors without high holesterol ontent (Fig. 2). In eah ase the_ tumor demonstrated a high signal intensity on T2- weighted images. In a single ase (Fig. 4), the tumor demonstrated low-intensity signal on both T1- and T2-weighted images. This orresponded to a ysti tumor with a high keratin ontent and deliate bone trabeular network. The presene of extensive ossifiation throughout the aspirated speimen was a surprising finding. pparently, the deliate bone trabeular network ollapsed with sution and ould be aspirated with yst fluid at surgery. In eah ase the typial adamantinomatous pattern was present in some of the pathologi setions, onfirming the diagnosis. Overall, CT was superior to MR in demonstrating alifiations within the tumors. MR failed to demonstrate areas of alifiation in three of 14 ases in whih alifiation was demonstrated by CT (Figs. 1 and 6). In one ase, alifiation would have been inorretly suspeted on the basis of MR. The areas of low signal on MR orresponded to low-density regions on CT believed to represent areas of ysti nerosis (Fig. 3). MR imaging was very sensitive in deteting the presene of pathology. In two ases MR was superior to ontrastinfused CT in demonstrating tumor. In one ase in whih tumor was onfined to the suprasellar region the tumor was _not appreiated on ontrast-infused CT. CT with metrizamide isternography performed )he same day and MR studies o Fig. 1.-8-year-old boy who had multiple surgial resetions for reurrent raniopharyngioma. MR was performed as part of periodi monitoring of tumor., Sagittal SE image (TE = 28, TR = 1000 mse) with relative Tl-weighting. Suprasellar tumor of high signal intensity., Coronal SE image with T2-weighting (TE = 56, TR = 2000 mse). Long T2 of tumor and avernous sinus extension was important in planning radiation therapy. C, xial noninfused CT san shows peripheral alifiation (arrowheads) not apparent on MR sans. D, PathologiC speimen shows large number of holesterol lefts. High holesterol tumor ontent was found to be one ause of high-intensity signal on Tl-weighted images.
JNR:8, May/June 1987 MR OF CRNIOPHRYNGIOMS 441 Fig. 2.-41-year-old man with headahe and visual field defiits., Sagittal image with relative T1-weighting shows moderate signal intensity of suprasellar tumor (TE = 28, TR = 500 mse)., Sagittal image with relative T2-weighting shows high signal intensity of tumor (TE = 56, TR = 2000 mse). C, Coronal MR image shows tumor impinging on right opti nerve (arrow), whih orrelated with symptoms. D, Metrizamide axial CT san shows suprasellar tumor. Contrast-enhaned CT before metrizamide study failed to show pathology. E, Pathologi speimen shows striking lak of holesterol lefts. Representative setion is shown with moderate amounts of keratin (blak arrow) and typial adamantinomatous ells of raniopharyngioma (white arrow). Lak of holesterol or hemorrhage and presene of keratin may aount for longer T1 of this tumor ompared with Figure 1. o E Fig. 3.-19-year-old woman with subsequent surgial resetion demonstrating raniopharyngioma., T2-weighted axial MR study shows areas of foal dereased signal intensity also present on T1-weighted images (TE = 30, TR = 2000 mse). Califiation ould be suspeted on basis of MR study alone., CT san shows foal areas of dereased density orresponding to areas of dereased signal on MR, believed to represent areas of ysti nerosis.
4"42 PUSEY ET L. JNR :8, May/June 1987 Fig. 4.-17-year-old girl with headahes., Unenhaned CT san shows hyperdense mass with suprasellar and posterior extension. and C, Sagittal (TE = 28, TR = 1000 mse) and axial (TE = 56, TR = 2000 mse) SE images, respetively, show low signal-intensity mass. D, xial ontrast-enhaned CT ~an after tumor drainage shows ollapse of yst walls. E, Pathologi speimen of yst aspirate shows deliate bone trabeular network and keratin. :.:.4, 16 ',3 "\-, --~--~-~-'----~--'---~--'-r-----.- ::e 150 nm Fig. 5.-38-year-old woman with bitemporal visual field loss., Coronal relative T1-weighted SE image (TE = 28, TR = 500 mse) shows high-intensity-signal suprasellar tumor., Sagittal T2-weighted image shows high-intensity signal of tumor (TE = 56, TR = 1500 mse). Pathologi speimen showed numerous RCs with few holesterol lefts. C, Ultraviolet light spetrosopy of yst ontents shows photopeak at 630 nm indiating presene of methemoglobin (arrow). Presene of methemoglobin was found to be one ause of T1 shortening.
JNR :8, May/June 1987 MR OF CRNIOPHRYNGIOMS 443 Fig. 6.-19-year-old woman seen for routine follow-up studies after surgial removal of intrasellar raniopharyngioma., Reurrene of tumor shows moderate signal intensity on T1-weighted images (TE = 28, TR = 500 mse)., Puntate alifiation (arrow) identified by CT was not seen on MR study. C, 6 months later. Patient returned with inreasing visual field defiits. Enlargement of tumor mass with suprasellar extension is seen on T1-weighted image (TE = 28, TR = 500 mse). D, Coronal T2-weighted image shows extension of tumor to level of opti hiasm (TE = 56, TR = 2000 mse). D performed 4 days later learly demonstrated the abnormality (Fig. 2). In a seond patient, pathology was suspeted on CT due to the presene of an enlarged sella without demonstration of tumor. MR 2 weeks later showed a 2.S-m tumor with suprasellar extension. In 10 patients the tumor was loated above and within the sella. In nine the tumor was entirely suprasellar. In one the tumor was loated only within the sella (Fig. 6). Posterior extension in five ases was well demonstrated on sagittal MR images, whih was signifiant in planning the surgial approah in these ases. In one of these patients the posterior loation of the tumor was not appreiated on CT, and the tumor was missed during transnasal biopsy. Subsequent MR revealed the posterior loation of the tumor (Fig. 7). Coronal images were espeially useful in demonstrating avernous sinus extension, whih was present in one ase (Fig. 1) and was important in planning postsurgial radiation therapy. Disussion Craniopharyngiomas originate from squamous epithelial rests of Rathke's pouh. They are benign, slow-growing tumors. lthough primarily tumors of hildren and young Fig. 7.-27-year-old man with double vision and headahes. Sagittal SE image (TE = 28, TR = 499 mse) shows high-intensity suprasellar tumor extending posteriorly. iopsy was performed through transnasal approah and normal tissue was obtained. Posterior extension of tumor is important in preoperative surgial planning. Subtemporal exposure may be preferred in ases with signifiant posterior extension.
444 PUSEY ET L. JNR :8, May/June 1987 adults, some may have delayed presentation in middle age or older. They present with headahe, visual symptoms, and symptoms aused by dysfuntion of the hypothalamus and pituitary gland. Growth failure is often the mode of presentation in a hild. The histologi pattern of raniopharyngiomas [9, 10] is variable and inludes a variety of ell types. Diagnosis is made histologially by the identifiation of nests or ords of stratified squamous or olumnar epithelium in loose fibrous stroma reminisent of the enamel organ of the tooth, and they are onsidered to have an adamantinomatous pattern. Craniopharyngiomas may be ysti, often with high holesterol ontent, or solid, and they ontain alium in 75% of ases. The wide range of histologi appearanes of raniopharyngiomas was refleted in their MR appearanes. lthough definitive statements onerning alium and holesterol ontent annot be made without omplete setioning of the entire speimen, qualitative examination of the setions orrelated well with the several MR patterns. High intensity on T1- weighted images was noted in ysti lesions with high holesterol ontent or ontaining methemoglobin. Kjos et al. [4] desribed high-intensity signal on T1-weighted images of hemorrhagi ysts. radley and Shmidt [11] showed that methemoglobin formation with T1 shortening at least partially aounts for the inreasing MR signal intensity of subarahnoid hemorrhage with time. Moderate intensity on T1- weighted images was noted in tumors laking signifiant holesterol or blood. oth these groups demonstrated high signal intensity on T2-weighted images. In a single ase (Fig. 4), low signal intensity was demonstrated on both T1- and T2-weighted images in a ysti lesion high in keratin and having extensive bone trabeulae. In this ase CT demonstrated a hyperdense lesion. raun et al. [12] desribed hyperdense ysti raniopharyngiomas in four of 63 ases in their series. Their analysis of yst ontents demonstrated a high protein ontent in these ases. However, the MR appearane of low T1 and T2 signal intensity in our series was explained by the deliate bone trabeular network found at pathologi examination of the yst aspirate. Despite the presene of ossifiation, the tumor behaved as a yst at surgery, and ollapse of the yst walls was seen on postaspiration CT studies. In the evaluation of raniopharyngioma the presene of alifiation is often of diagnosti signifiane. Holland et al. [8] have shown CT to be superior to MR in the detetion of alifiation, whih was also demonstrated in our study. The failure of MR to demonstrate tumoral alifiation is of less importane in the evaluation of tumor reurrene when the tissue diagnosis is known. MR has proved to be valuable in preoperative and radiation therapy planning due to its multiplanar apabilities. urate preoperative knowledge of the tumor extent may lead to a hange in the surgial approah [13]. transnasal approah may be suffiient for an entirely intrasellar tumor. Suprasellar tumor may require a subfrontal approah to allow optimal exposure of the opti hiasm. Cryoprobes may need to be available in this ase to peel tumor away from the opti hiasm. Supra- and intrasellar tumor extension may additionally require unroofing of the sphenoid sinus. If the tumor extends posteriorly along the livus a subtemporal approah may be preferred. ysti raniopharyngioma may allow a more onservative surgial approah with drainage without removing the walls and with implantation of radioisotopes. MR offers superior information in identifying tumor in the avernous sinus (Fig. 1 ), whih would not usually be explored at surgery and would be treated with radiation therapy. Tumor extension into the sphenoid sinus is not as easily identified by MR due to its resemblane to benign onditions suh as sinus disease and hanges resulting from transnasal biopsy. In ases with supra- and intrasellar tumor, a subfrontal approah with unroofing of the sphenoid sinus would be used, and extension into the sinus would be disovered at surgery. In summary, MR is more sensitive than CT in identifying the presene and extent of tumor. The wide range of appearanes on T1-weighted images reflets the range of histologi appearanes of raniopharyngiomas. T1 shortening may reflet the presene of high holesterol ontent or methemoglobin. CT is superior to MR in deteting alifiations. Currently, CT is more speifi in establishing the diagnosis of raniopharyngioma. MR is preferred in the evaluation of tumor extent for presurgial and radiation planning. MR is valuable in deteting reurrent tumor, partiularly in hildren, to limit radiation exposure. CKNOWLEDGMENTS We thank Harry Vinters, UCL neuropathology department, for photography of pathologi speimens and disussions on pathology of raniopharyngiomas and Ed Helmer for Figures 4 and 4D. REFERENCES 1. rant-zawadzki M, adami JP, Mills CM, Norman D, Newton H. Primary intraranial tumor imaging: a omparison of magneti resonane and CT. Radiology 1984;150 :435-440 2. radley WG, Waluh V, Yadley R, Wyoff RR. Comparison of CT and MR in 400 patients with suspeted disease of the brain and ervial spinal ord. Radiology 1984;152:695-702 3. Hawkes RC, Holland GN, Moore WS, et al. The appliation of NMR imaging to the evaluation of pituitary and juxtasellar tumors. JNR 1983;4:221-222 4. Kjos O, rant-zawadzki M, Kuharzyk W, Kelly W, Norman D, Newton T. Cysti intraranial lesions: magneti resonane imaging. Radiology 1985;155 :363-369 5. Kuharzyk W, rant-zawadzki M, Sobel D, et al. Central nervous system tumors in hildren: detetion by magneti resonane imaging. Radiology 1985;155: 131-136 6. Peterman S, Steiner RE, ydder GM. Magneti resonane imaging of intraranial tumors in hildren and adolesents. JNR 1984;5 :703-709 7. Rink P, Meindl S, Higer HP, ieler EU, pfannenstiel P. rain tumors: detetion and typing by use of CPMG sequenes and in vivo T2 measurements. Radiology 1985;157 :103-106 8. Holland, Kuharzyk W, rant-zawadzki M, Norman D, Haas OK, Harper PS. MR imaging of alified intraranial lesions. Radiology 1985;157:353-356 9. Petito CK, DeGirolami U, Earle KM. Craniopharyngiomas: a linial and pathologial review. Caner 1976;37: 1944-1952 10. urger PC, Vogel FS. Surgial pathology of the nervous system and its overings. New York: Wiley, 1982 : 511-528 11. radley WG, Shmidt PG. Effet of methemoglobin formation on the MR appearane of subarahnoid hemorrhages. Radiology 1985; 156: 99-1 03 12. raun IF, Pinto RS, Epstein F. Dense ysti raniopharyngiomas. JNR 1982;3: 139-141 13. Rand RW, Konovalov N. Craniopharyngiomas. In: Rand RW. Mironeurosurgery. St. Louis: Mosby, 1985 :187-195