185

Dynamic CT Scanning for Visualization of the Parasellar Carotid Arteries

Wendy A. Cohen 1 Richard S. Pinto Irvin I. Kricheff

This article appears in the March / April issue of AJNR and the May issue of AJR. Received June 22. 1981; accepted after revision October 20, 1981. Presented at the annua l meeting of the American Society of Neuroradiology, Chicago, April

198 1 . This work was suppo rt ed in part by E. R. Squibb & Sons, Medical Developmental Department, Princeton , NJ 08540 and by General Electric, Milwaukee, WI 53201. , All authors: Department of Radiology , New York University M edica l Center , 550 First Ave., New York , NY 10016. Add ress reprint requ ests to W. A. Coh en. AJNR 3 :185-189, March / April 1982 0195-6108 / 82 / 0 3 02-0185 $00 .00

© Ameri ca n Roentg en Ray Society

Evaluation of patients before transsphenoidal hypophysectomy for large intrasellar mass lesions has required bilateral internal carotid artery angiography . Using intravenous injection of contrast medium , a method has been developed to visualize the parasellar carotid arteries with rapid sequence sequential computed tomographic scanning . In 18 patients, the cavernous segments of the internal carotid arteries were well seen in 27 of 28 instances with technically complete examinations. The vascularity of the mass lesions and vascular encasement was also demonstrated .

The radiologic evaluation of abnormalities of the sella turcica before the development of computed tomography (CT) consisted of sku ll films, comple x motion tomography , angiography, and pneumoencephalography . Enl arg ement of the sella turcica on plain films is most often caused by pituitary adenoma and is less commonly due to other lesion s, such as c ran ioph aryngioma, meningioma , dysgerminoma, teratoma , aneurysm , or an abe rrant co urse of th e internal carotid arteries [1]. A parasellar aneurysm may present with symptoms secondary to endocrinologic dysfunction or to local compression of the optic nerves, optic chiasm, or cavernous sinus without history of prior subarac hnoid hemo rrh age [2-4]. Widespread experience with CT has confirmed its value in both diagnosing and demonstrating the extension of an intrasellar or parase ll ar process [5]. However, identification of vascular abnorm aliti es within the pituitary fos sa and suprasellar cistern has remained difficult , and the differential di ag nosis of a parasellar aneurysm versus tumor rem ai ns. Many large intrasellar tumors are res ected by a transsphenoidal app roach . Before operation , bilateral internal carotid angiography has been perfo rm ed to demonstrate the vascular anatomy, confirm tumor extension, show possible vascular encasement by tumor, and el iminate the possibility of a vasc ular lesion [6]. High-resolution CT after intravenous in fu sion of contrast mate ri al will adequately demonstrate the supraclinoid carotid arterie s and allow differenti ation of cavernous sinus from bone [7], but is unable to resolve the course of the carotid artery through the cavernous sinus. We report a tec hnique using rapid sequence scanning (dynamic CT) with variable intersca n loca tion and timing to vis ualize th e parase ll ar vascular stru ctures after an intravenou s bolus injection of contrast medium. Thi s has el im inated conventional angiography as a prerequisite for transsphenoidal surg ery.

Materials and Methods Ei ghteen patient s were suspected of having abnormalities in the reg ion o f the se ll a turc ica because of hi story, endoc rin e c hang es, visual c hang es, abnormalit ies on sku ll films , or demonstrated parasellar lesions on ro utine noncontrast and co ntrast en hanced CT . Dynam ic CT was performed with a GE CT / T 8800 scan ner usin g graphical data analysis capab ilities avail able in ex perimental software and im ag in g prog rams supplied by General

186

COHEN ET AL.

Fig. 1.-Lateral sku ll " scout view " show ing scan position ing for coronal dynamic CT . Lin es mark levels of scans. Line 2 is first scan as described in part 1 of Materi als and Met hods.

El ectri c (GEDIS 6 .1 6-6.21). The examination consisted of two parts with an optional third part. For each part, baseline scans thro ugh the structures of interest were perform ed. Initial levels were c hosen fro m a digitized image of th e lateral hea d (" scout view " ). Th e images were obtained with or without co ntrast enhancement. For th e dynamic study an intravenous bolu s injec ti on of Renografin-76 was given through an 18 gauge angiocath eter in a larg e antecubital vein. All injection s were wit h a mechan ica l injector , deli verin g 40 ml of warmed co ntrast agent (11 g iodine) over 4-5 sec . Arm to brain transit tim e in ad ults, determined empiricall y, is about 8 sec . Th e scann er required 2 sec for rotor preparation, thus the injec ti on was started 4 sec before th e co mmence ment of th e initial scan. Max imum co ntrast density appeared in vascular structures in th e second and third scans of a rapid sequence series. Using th e rapid seq uence scanning method , a single scan was obtained over 4 .8 sec. Using th e tabl e inc rement mode, 6.5 sec. were required between scans, during w hic h tim e th e table c hanged locati ons to a preprog rammed position . If th e scan loca tion was not c hanged , interscan tim e co uld be red uced to 1.5 sec.

c

AJNR :3 , Marc hI April 1982

o

Fig. 2.-Norm al coronal dyna mi c CT scans. A and B , Preliminary scans o f sella turcica and dorsum. C and D, After bolu s con trast injec tion . Parasellar carotid art eri es (black arrows), suprac linoid vessels, and basilar artery (white arrowhead).

posteriorly. Baseline and dynamic CT scans were compared to differentiate vascu lar structures from bon e, cavernous sinus, or tumor (fig . 2).

Part 2 Th e pati ent was repositioned supine, and adjace nt 10-mm-thick baseline scans were obtained starting 10 mm inferior to the floor of th e sella and ending at th e level of the suprasellar cistern. Dynamic CT scans were again obtained, combined with a bolus intravenous injec tion of contrast material at the same levels as the baseline scans. The second scan (which has maximum change in contrast density) included the anterior clinoids. Scanning started at the more ca udal level and moved craniad.

Part 1 (Coronal Scanning) The patient was placed in a position for direct coro nal scanning , ly in g prone with his head ex tended and th e scanner gantry tilted so th at th e scan plane was approximately perpendicular to the pl anum sphenoid ale. Gantry ang ulati on and scan position were determined fro m the " lateral" scan view. Baseline 1 O-mm-thick adjacent nonco ntrast scans were obtained , starting 10 mm in front of the tips of th e an terior c linoids to enco mpass a part of th e planum sphenoidale and tuberculum , and ending behind th e dorsum sellae (fi g . 1). The ro utine dynamic CT seri es co nsisted of fo ur to five scans taken at the same locati ons as th e baseline scans using th e tabl e incre men t mode and a bolus injection of co ntrast agent. Scanning started at th e most anterior point of the baseline scans and moved

Part 3 (Optional) If a question remain ed concerning the vascu larity of a visualized structure, rapid seq uence imaging , combined with an intravenous bolu s injection of contrast medium , was performed through th e midpart of th e structures to be identified in a single location. For th ese patients, graph ic analysis of change in contrast density over tim e within speci fi c stru c tures was perform ed . After a repre sentative area within a structure was chosen, co mputer programs were availabl e th at averaged th e CT number within th e region for each scan of th e dynamic CT series. The data were presented with either absolute CT number or th e c hange in CT number (t. Hounsfi eld units [H]) plotted as a function of scan time .

AJNR :3, M arch i April 1982

CT OF PARASELLAR CAROTID ARTERIES

Fig . 3 .- 0ynamic CT scan , axial plane. Vessels of circ le o f Willis in pati en t wi th pituitary adenoma. Regions of interest ce nt ered on basilar art ery ( 1) and pitu itary ad enoma (2) . Graph s show change of CT number with tim e (to Hounsf ield units). Graph 1 , norm al vasc ul ar pattern . Graph 2, slower ri se to plateau found with many ad enomas.

187

Fig. 4 .- 0ynamic CT scan of c raniopharyngioma . Regions o f interest centered over sella turc ica ( 1) and intracavern ous ca rotid arte ry (2). Graphs of inc reased CT number (IN C CT# ). Graph 1, pattern of craniopharyngioma. li tt le apprecia ble c hange in contrast density . Graph 2, pattern o f vasc ular struct ures .

Results Pathologic entities in the 18 patients consisted of pituitary adenoma (11 patients) , pituitary microadenoma (2) , craniopharyngioma (2), microadenoma after radiotherapy (1), and normal pituitary (2) . Another patient with a sphenoid wing meningioma is included in this report . This patient had a dynamic CT scan of the lesion (as in part 3); however, no attempt was made to visualize the other parasellar vessels . Six patients had incomplete studies : three initial patients had only one part of our protocol performed and three had early termination of the study due to extravasation of contrast agent into the soft tissues of the upper arm. Eight patients were also studied angiographically. Fourteen patients were studied as per protocol and form the basis of this analysis. The cavernous segment of the right internal carotid artery was well visualized in 13 (93 % ) of 14 patients but only partially seen in 1 (7 % ) of 14. The cavernous segment of the left internal carotid artery was well seen in all patients. Both supraclinoid carotid arteries were demonstrated in 10 (71 % ) of 14 and one of the two arteries was seen in 1 (7 % ) of 14. The A 1 segment of the anterior cerebral artery was seen in 10 (71 % ) of 14. The maximal attenuation values of both arterial structures and of various lesions were assessed . The parasellar carotid arteries on axial projections ranged from 94 to 258 H (mean , 170 H) and on coronal projections from 1 23 to 448 H (mean, 237 H). Adenomas on either axial or coronal scans ranged from 41 to 105 H (mean , 65 H). If a change in CT number (LlH) alone is considered, carotid arteries on axial scans ranged from 89 to 170 H (mean , 120 H); on coron al scans, 44 to 185 H (mean , 94 H); and pituitary adenoma, 2 to 27 H (mean , 17 H). The single meningioma studied had

a maximum attenuation value of 75 H and a LlH of 41 H. Craniopharyngiomas did not significantly chang e in density . Graphica l Analysis

Contrast density ch ang e (t.H) was plotted again st tim e for the 14 patients who had dyn ami c scans perform ed in a single plane (fig . 3 ). Plots obtain ed from arteri al stru ctures demonstrated a rapid , sh arp ri se and fall in attenuati on values over the first 15-20 sec. A second , small recirc ul ation curve occurred during the next 15-20 sec, whi ch was 26-40 sec from the initi al sc an . For these vasc ul ar structures the baseline absolute CT density was in th e rang e of 60-80 H after th e first inj ecti on of co ntrast materi al. LlH varied from 70 to 180 H, and after th e early large pea k , a plateau ranging between 14 and 41 H (mean, 25 H) occurred . Similar analysis of the appea ran ce of contrast material in pituitary adenomas demonstrated a gradu al increase in attenuation value during th e first 30 sec after th e intravenous bolus injec tion of c ontrast material. Subsequ ently, the c urve formed a pl ateau ranging betwee n 6 and 2 5 H (mean , 12 H) that extend ed into th e recircul ation phase . The slope of th e initial wash-in ph ase for pituitary adenomas was flatter than th at obtain ed from vasc ul ar struc tures . Eve n tumors that had ea rly pea ks in co ntrast density in a tim e simil ar to that of arteri al stru c tures demon strated a slow rise after the initial " arteri al" phase and arteri al peaks, whi ch were lower in maxi mum CT number th an th ose see n in vasc ular structures. The least co ntrast passage occurred in craniopharyngiom as, whi ch demonstrated insignifi cant early chang es in CT attenuation values and little enh ance ment at any tim e (fig. 4) . In all cases, it was possibl e to differenti ate a prim ary vas c ul ar stru cture from a soft-ti ssue stru c ture.

188

COHEN ET AL.

AJNR :3, M arch ! April 1982

Discussion An giog raphy of both intern al carotid arteries in pati ents with large sell ar and parasell ar lesions co ntinu es to be used to locali ze vasc ular stru c tures, to demonstrate d ispl acements, and to ru le o ut a vasc ul ar eti ology of th e intrase ll ar path olog y. Convention al CT clea rly shows th e suprac linoid carotid arteri es ; however, th e rapid , earl y enh ance ment of pituitary ade nomas may prec lud e separation of tumor from th e enh ance d cavern ous sinu s and parase ll ar carotid arteries. Arte ri ography of th e carotid arteri es c learly demonstrates suprase ll ar tum or extension by elevation of th e A 1 seg ments of th e anteri or cerebral arteri es, but th e lateral extent of th e tumor ca n only be describ ed if th ere is spread outside th e se ll a turcica th at either disp laces or visibly encases th e ca ve rn ous segment of th e intern al carotid artery [8, 9]. In so me cases, encase ment can on ly be inferred from a co mbin ati on of CT and angiographic finding s. In one of our cases, dynami c CT was abl e to demonstrate th e c aroti d artery embedded within a tumor th at extend ed laterally beyond th e vessel , a finding not apprec iated angiographically (fig. 5). A previ ous report of dyn ami c CT [10] c haracterized th e change in c ontrast density in relation to tim e of variou s lesions after bolus injection of c ontrast materi al. Vascul ar lesions had a large chang e in contrast density with steep , early ri sing and down slopes, whil e pituitary adenom as demonstrated a slow ris e, broad peaks, and a gradu al dec rease in c ontrast density with a max imum peak at 2 min (fig. 3 ). Even during th e second and third parts of our protoco l, wh en patients had already received contrast mater ial, th e higher peak and mark ed chang e of density of vasc ul ar stru ctures allowed the differentiation of vessels from intrase ll ar mass lesions with prior c ontrast enh ancement. Nonvasc ul ar stru ctures were also c lea rly distingui shed . Du e to th e intensity of th e co ntrast enh ance ment in vasc ular stru c tures, in most cases it was unn ecessary to co mpare dyn ami c CT with baseline CT scans to distinguish vasc ul ar stru c tures from bone in th e region of th e sell a. In a pati ent with an enl arg ed pituitary fossa, accurate id entifi ca ti on of th e locati on of th e parase ll ar vessel s on a sing le noncontrast scan is diffic ult. To assure th at no vasc ul ar abnorm ality is prese nt, th e ca rotid artery within th e ca vern ous sinus as well as in th e suprac linoid region must be vi sualized , whi c h requires a rapid c hang e of sc an positi on. For most pati e nts with enl arg ement of th e sell a, its entire vo lume ca n be imaged in two 10 mm sca ns. Using a 5 sec inj ecti on rate and th e tabl e in crement mod e, parasell ar vascul ar stru ctures could be visu alized on the two scans th at enco mpassed th e se ll a (fig . 6) . If an abnorm al vasc ul ar stru cture in a parasell ar or intrasell ar location co ntinu ed to be suspected , a dynami c CT sca n centered over th e suspected les ion was obtain ed. Thi s scan sequ ence constituted th e opti onal third part of our protocol. An other use of th e opti onal third part was to provid e furth er inform ation about th e vascul ar anatomy aro und th e circ le of Willi s. In two cases, thi s provi ded a bett er definiti on of th e suprac linoid vesse ls, partic ul arl y th e A 1 segm ents of th e ante ri or cerebral arteri es . However, th e primary use of the third part was to obtai n transit tim e analyses.

Fi g. 5. -Dynami c CT scan of recurrent pituitary ad enoma. Encase ment of ri ght ca rotid art ery by lateral tumor ex tension . Right ( black arro w) and left (open arrow) intracavern ous carotid art eri es.

For most patients it was necessary to obtain both axial and coronal scans, although if parasellar vascular structures cou ld be c learly identified after a single injection , one series would be adequate. For example , in two cases, one cavernous c arotid artery segment was not we ll seen on coronal scans but was visualized in its intracavernous position on axial dynamic CT . During the initial phases of our study, we performed two injections to confirm the validity of the method . A difficu lty in imaging parasell ar vascular structures on dynamic CT occurred if the intrasellar mass lesion was large and markedly displaced the carotid arteries . The internal carotid arteries in a patient with a large craniopharyngioma were poorly seen, although the absence of vascular structures within the sella was clearly demonstrated . A second patient, as yet not operated upon , has marked lateral exten sion of tumor. The laterally displaced vessel was vi sualized but its course through the cavernous region was poorly resolved . Cerebral angiography by the femoral route has been reported to have a risk of major c omplications of 0.28 % and of minor compli cations of 6 .65% [11]. Dynamic CT has a risk of a major c ompli c ation the same as that of any intravenous injection of contrast material. The minor compl ications-e xtravasation of contrast material into the upper arm-resulted in local discomfort for 15 to 30 min , easi ly reli eved by warm soaks . No permanent sequ ela occurred. Th ese compli c ations occurred early in our investigation wh en injection rates of greater than 10 ml / sec were used . At our c urrent injection rate of 8 ml / sec , no contrast medium extravasation has occurred . The multiple injections of contrast materi al delivered a max imum total iodine dose of 44.4 g if all three parts of the protocol were used . Fluids w ere enco uraged aft er th e examination. All patients had normal renal fun ction and no ren al complications .

CT OF PARASELLAR CAROTID ARTERIES

AJNR :3, M arch i April 1982

A

B

c

1 89

D

Fig. 6 .-Cran io pharyng ioma. A and B, Intern al caro tid art ery injecti ons showin g upward di spl acement of A 1 segm ent of anteri o r ce reb ral art eri es . C

and D , Coro nal dynamic CT sca n of same patient, sim ilar vasc ul ar co nfiguration . C, Elevated A 1 seg ments. D , Cavernous carotid art eries (arrowheads ).

Our study proved that rapid sequence CT imaging in our protocol will visualize well the parasellar vascular tree, locate major vessels encased by tumor, and eliminate the possibility of a vascular lesion in the differential diagnosis of a parasellar tumor. Since it requires only intravenous injection of contrast material , it may be performed on an outpatient basis. Dynamic CT is, in our opinion, a viable alternative to angiography as a prerequisite to transsphenoidal surgery of parasellar tumors.

3. Bull JWD . Contribution of rad iology to the stu dy of in trac rani al aneurys ms. Br M ed J 1962 ;2: 1 70 1- 1 708 4 . Heisa k anen 0 , Nikki P. Large intrac rani al aneu rys ms . Acta Neurol Sca nd 1962;38 : 1 9 5 - 208 5. Naidich TP , Pinto RS , Kushn er MJ, Lin JP , Kri c heff II , Leeds NE , Ch ase NE . Eva lu ati on of se ll a and parasell ar masses by computer tom ograph y. Radiology 1976; 120 : 9 1 - 99 6 . Kern EB, Pearso n BW, Mc Donald TJ , Laws ER Jr. Th e transseptal approac h to lesions of th e p ituitary and parase ll ar reg ions. Lar yngoscope 1979; 89: 1 - 34 7 . Hayman LA , Eva ns RA , Hinc k VC . Rapid hi g h d ose (RHO) contrast co mputed tom ograph y o f pari se ll ar vesse ls. Radiology 1979;13 1 : 12 1-1 23 8. Taveras JM , W ood EH . Diagnostic ne uro rad io logy. Baltimore: Willi ams & Wilkin s, 1976 : 737 -738 9. Epstein BS, Ep stein JA . Th e angiog raphi c demonstrati on and surgica l implicati ons of im bedd ing of the ca roti d syphons by a large pituitary adenoma. AJR 1968; 104: 1 62- 1 67 10. Wing SO , Anderson RE , Osborn AG . Dynami c c rani al co mputed tomograph y: prelimin ary results. AJNR 1980; 1 : 135 -1 39 11 . Huc km an MS, Sh enk GL, Neems RL , Tin or T. Transfemoral cerebral arteri ograph y versus direct perc utaneous caroti d and brachi al arteriograph y: a co mp ari son of co mp li cati on rates . Radiology 1979; 132 : 93 - 98

ACKNOWLEDGMENTS We thank Patri c ia Lewi s for te c hnica l assistance and Lydi a Logozo, Linda Michaels and Jill Tepper for assistance in manusc ript preparation .

REFERENCES 1. Boyce OW , Huckman MS . Contigu ous internal ca rotid arteri es in empty se lla syndrome . Radiolog y 1976;120 : 1 20 2. White JC . Aneury sms mistaken for hypo ph yseal tumors. Clin Neurosurg 1964;10 :2 24-250