865

0361-803X/93/1604-0865© American Roentgen Ray Society

Misperfusion of the Liver DuringHepatic Artery InfusionChemotherapy: Value of PreoperativeAngiography and Postoperative PumpScintigraphy

A. Cahid Civelek1James V. Sitzmann2

Bennett B. Chin1

Anthony Venbrux3

Henry N. Wagner, Jr.1

Louise B. Grochow4

Received August 20, 1992; accepted after nevi-sion November 5, 1992.

I The Russell H. Morgan Department of Radi-

ology and Radiological Science, Division of NuclearMedicine, The Johns Hopkins Medical Institutions,615 N. Wolfe St., Baltimore, MD 21287-9217. Ad-dress correspondence to A. C. Civelek.

2Department of Surgery, Division of SurgicalOncology, The Johns Hopkins Medical Institutions,Baltimore, MD 21287-9217.

3The Russell H. Morgan Department of Radi-ology and Radiological Science, Division of Inter-ventional Radiology, The Johns Hopkins Medicalinstitutions, Baltimore, MD 21287-921 7.

4Department of Oncology, The Johns HopkinsMedical Institutions, Baltimore, MD 21287-9217.

OBJECTIVE. One purpose of this study was to determine if patients who have ana-tomic variations in their hepatic arteries are at increased risk for complications asso-

ciated with the use of intrahepatic arterial infusion pumps. We also tried to determinethe value of perfusion studies obtained with 9SmTo-microspheres or 99mTc_macroag�gregated albumin in detecting postoperative hepatic or visceral misperfusion and inpredicting complications in patients with anatomic variants despite pre- or intraoper-ative attempts to correct the arterial abnormality.

SUBJECTS AND METHODS. We prospectively compared findings on scintigramsobtained after delivering the radionuclide through intrahepatic arterial infusion pumpswith anatomic variations in hepatic arteries seen on celiac and superior mesenterichepatic arteriograms obtained before placement of the pump in 49 consecutive

patients with colon carcinoma metastatic to the liver.RESULTS. Despite pre- or intraoperative attempts to correct arterial abnormalities

to ensure optimal perfusion of the liver in 24 patients with hepatic arterial anomaliesseen on preoperative arteriograms, only two patients had normal findings on postop-erative perfusion studies performed with �mTc�microspheres and/or 99mTc_macroaggregated albumin. Abnormalities included perfusion of extrahepatic organs, in-

cluding the spleen in 12 patients, stomach in seven, bowel in four, and pancreas inthree. Eight patients had no perfusion of the left lobe of the liver, and three had noperfusion of the right lobe. Two patients had minimal or no perfusion of both lobes. In23 of 25 patients with no demonstrable variations in vascular anatomy on preopera-tive celiac and superior mesenteric arteriograms, findings on hepatic pump scinti-grams were normal. Of the 24 patients with abnormal scintigraphic findings, 20 hadsubsequent clinical complications. However, only two of the 25 patients with normalscintigraphic findings had clinical complications.

CONCLUSION. Our results indicate that patients with anatomic variations in thehepatic arterial system are at high risk for misperfusion during chemotherapy despitepre- or intraoperative efforts to alter the perfusion for chemotherapeutic agents deliv-ered by intrahepatic arterial infusion pumps. Misperfusion can be detected by usingpump scintigraphy, and therefore patients should be closely monitored with �Tc-macroaggregated albumin perfusion studies to ensure successful delivery of the

chemotherapeutic agents and to avoid serious clinical complications caused by mad-vertent perfusion of other organs.

AJR 1993;160:865-870

Patients with coborectal cancer metastatic to the liven generally have a poorprognosis. Their response rate to systemic chemotherapy is approximately 20-40% [1]. Liven tumors derive their blood supply primarily from the hepatic arteryrather than the portal venous system [2]. Although a significant increase in sun-vival has not been proved, the response rate for chemotherapeutic agents admin-istered through the hepatic artery has been reported to be 30-75% [3-6].

Technical complications associated with use of the hepatic artery infusion pumpinclude infusion of high concentrations of chemotherapeutic agents to extrahepaticsites; dislodgment, migration, and thrombosis of the catheter; and infection [7, 8].

866 CIVELEK ET AL. AJR:160, April 1993

After surgical placement of the hepatic arterial catheter, thestandard method used to predict distribution of chemothena-

peutic agents and to determine unwanted extrahepatic penfu-sion is postoperative pump scintignaphy, in which the radio-nuclide is delivered via the hepatic artery [9, 10].

When the catheter of an intraarterial hepatic infusionpump is placed in the gastnoduodenal artery to deliver a che-mothenapeutic agent to the entire liver, the presence of ahepatic arterial anomaly may cause infusion of other organson incomplete hepatic infusion. This may cause systemicand/on local toxic effects due to the chemothenapeuticagents on result in inadequate treatment responses. Theanatomy of hepatic circulation is carefully evaluated withpreoperative splanchnic arteniography before placement of

the catheter, and if necessary, the surgical approach is mod-ified to achieve satisfactory liven perfusion.

The purpose of this study was to determine if variations inhepatic arterial anatomy place patients at higher risk forcomplications associated with the use of a hepatic arteryinfusion pump. We also wanted to determine the value ofperfusion studies performed with 99mTcmicnOspheres or99m Tc-macnoaggregated albumin in detecting postoperativehepatic or visceral misperfusion that may occur in suchpatients despite pre- or intnaoperative attempts to correct thearterial anomalies.

Subjects and Methods

During a 3.5-year interval, 49 consecutive patients, 18 women

and 31 men 40-83 years old (mean, 63 years), with colon cancermetastatic to the liver had an infusion pump implanted in the hepatic

artery for intraarterial delivery of chemotherapeutic agents. Confir-mation of hepatic metastases was based on histologic reports,

results of liver function tests, and findings on cross-sectional imag-ing (e.g., CT, MR imaging, sonography). All patients had selective

celiac, superior mesenteric, and occasionally common hepatic onleft gastric arteriography before placement of the infusion pump.Celiac axis injection was preferred over selective injection of thecommon hepatic artery because the left gastric artery is not opaci-fied on arteniograms obtained with selective common hepatic injec-tion. Thus, the high prevalence of hepatic arterial variants would not

be detected with routine selective common hepatic arteniography.Selective catheterizations were performed, via the gastroduodenalartery or left gastric artery, if appropriate or if therapeutic interven-

tion was contemplated. Digital subtraction arteniography occasion-

ally was used to supplement the cut-film arteriographic studies orwhen abnormal renal function dictated the use of smaller quantitiesof contrast material. Generally, nonionic contrast medium was usedfor the cut-film and digital subtraction studies. If anomalies of arterial

distribution were found on arteniography, alternative surgical tech-

niques (e.g., placement of dual catheters, ligation of appropriate

arteries, on modified placement of a catheter) were used for implan-

tation of the arterial catheter to ensure uniform and reproducibleliver perfusion during chemotherapy (intraopenative injection of fluo-

rescein dye into the perfusion catheter was performed to confirmuniform hepatic perfusion).

Within 3-5 days after placement of the infusion pump and beforeintraarterial chemotherapy was started, all patients had perfusion

scintigraphy, in which the radionuclide was delivered via the infusion

pump. This study determines the hepatic arterial perfusion pattern

via the infusion pump catheter(s) and corresponds to the distributionof chemotherapeutic agent.

The overlying skin of the abdomen was cleaned, and a special

atraumatic needle (Huber needle, Shirley Infusaid Inc. , Norwood,

MA) to prevent backfbow was used to inject 1-3 mCi (37-111 MBq) of99mTcmicrospheres (Nuclear Products 3M Center, St. Paul, MN; 20patients) or 99mTc_macroaggregated albumin (Du Pont Menc Phar-

maceutical Co., Bilbenica, MA; 29 patients) in a volume of 0.5-1 .5 ml

of saline into the side port of the pump at a rate of 0.5-1 .0 mI/mm.The patient was placed supine on an imaging table, and a

gamma camera detector with a lange field of view was positionedoven the patient’s abdomen from the xiphoid process to the pelvis.For determination of flow distribution, dynamic images of the liven inthe anterior projection were obtained at 3 sec per frame for 60 sec

with a low-energy high-resolution collimator. The distribution of themicrospheres was determined by obtaining anterior and posteriorimages of the liver-spleen region for 500,000 counts per view. Ante-

nor views of the pelvis and chest also were obtained for 500,000

counts per view to determine perfusion of abdominal soft tissue out-

side the liver and of the lung. In order to facilitate detection of small

regions of extrahepatic activity, images were recorded on film and

on an optical disk so that image brightness and contrast could beadjusted.

A 99mTc_sulfur colboid liver-spleen scan was obtained the sameday or the day after perfusion scmntigraphy in order to determine thesize and shape of the liven and spleen. These images were helpful

in distinguishing distorted liver tissue from neighboring nonhepatic

soft-tissue activity that may have been seen in the perfusion stud-

es. The nadionuclide studies were evaluated by two experiencednuclear medicine physicians who did not know the results of the

arteniographic studies or of transcatheten or surgical occlusions per-

formed to remedy abnormal perfusion patterns.

Results of the nadionuclide study were defined as abnormal if any

visceral organ other than the liver (nonhepatic soft tissue) was visu-

alized or if one or more lobes of the liver showed no perfusion. Lack

of perfusion was attributed to occlusion of the catheter or the artery

or to misplacement of the catheter tip during surgery.

A variant arteniognaphic finding was defined as any anatomic van-ation from the “classic” distribution [11 , 12]. Several variations inarterial supply from the celiac axis to the stomach, duodenum, liver,

and spleen also have been described and categorized by Michels[13]. Postoperative perfusion scintigraphic findings in the 49patients were evaluated and compared with findings on arterio-grams obtained preoperatively.

Results

In the 49 patients, findings on preoperative arteniognamswere normal in 25 and variant in 24. The 24 variant findingswere anomalous origins of the left hepatic artery in ninepatients, of the right hepatic artery in five, and of the gas-troduodenal artery in two; tnifurcation of the common hepaticartery in four; and other combinations of findings in four(e.g., one patient with an accessory right hepatic artery andthree patients with arterial communications between lefthepatic and gastroduodenal on left gastric arteries). In fivepatients, the origin of the right hepatic artery was variant: itarose from the aorta in one, the superior mesentenic artery intwo, the gastroduodenal artery in one, and the celiac axis inone. Nine patients had variant origins of the left hepaticartery: arising from the celiac axis in two, the left gastricartery in four, the superior mesentenic artery in one, and thegastroduodenal artery in two.

The preoperative arteniographic and postoperative scinti-graphic findings are compared in Table 1 . Of the 49 patients,

TABLE 1 : Number of Patients with Normal and VariantFindings on Arteriograms Obtained Before Placement of anInfusion Pump in the Hepatic Artery and on ScintigramsObtained After Placement

Fig. 1.-A, Anterior scintigram obtained by in-jecting 99mTc..�acreagg,.�ated albumin intoinfusion pump in hepatic artery. Left lobe of liveris not perfused because left hepatic artery is cc-cluded.

B, Anterior 99mlcsulfur collold scintigramshows multiple focal defects in liver caused bymetastatic colon cancer. Most lesions in rightlobe were well perfused In macroaggregated al-bumin study (A).

Fig. 2.-A and B, Anterior (A) and posterior(B) scintigrams obtained by injecting 9�JmTo-microspheres into infusion pump in hepatic arteryshow perfusion of both lobes of liver, spleen(Sp), stomach (St), and, probably, pancreas (P).This represents marked extrahepatic perfusion,which can cause systemic and local toxic ef-fects when chemotherapeutic agents are dcliv-ered via pump. In this patient, a large bleedinggastric ulcer developed. Subsequently, cathe-ter was repositioned, and an aberrant arterywas embolized nonsurgically before continua-tion of hepatic intraarterial chemotherapy.

A �

Sp

AJR:160, April 1993 MISPERFUSION DURING HEPATIC ARTERY CHEMOINFUSION 867

A B

ScintigraphicArteniograp hic Findings

Findings Variant (n = 24) Normal (n = 25)

Normal(n=25) 2a 23b

Abnormal (n = 24) 22c 2d

aVaniant hepatic arterial anatomy was corrected intnaoperativety by ligation

of an aberrant artery in one and with modified placement of catheter and liga-

tion of an aberrant artery in the other.bBecause of mechanical failure, pumps were removed in two patients.

CNineteen patients had clinical complications.

dOne patient had an unexplained clinical complication.

25 had normal and 24 had abnormal findings on perfusionscintignams (Figs. 1-3). Of the 24 patients with abnormalfindings on perfusion scintignams, 22 had variant findings onarteniognams (e.g., hepatic arterial blood supply variants) andtwo had normal findings on arteniograms. Abnormal scinti-graphic findings included extrahepatic perfusion in 16

patients (Figs. 2 and 3) and abnormal liven perfusion in 13.Some patients had more than one abnormality. Of those with

extrahepatic perfusion, the distribution included the spleen in12 patients, stomach in seven, bowel in four, and pancreas inthree. Of those with abnormal liver perfusion, the abnormali-ties included no perfusion ofthe left lobe in eight (Fig. 1), noperfusion of the night lobe in three, and minimal or no perfu-sion of both lobes in two (Fig. 3). Of the 24 patients withabnormal findings on scintignarns, 20 had subsequent clinicalcomplications. In 19 of the 20 patients, malperfusion of theliver and/or catheter dysfunction was shown on repeatedarteniograms. In the last patient, the spleen was visualized,and a subsequent gastrointestinal complication necessitateddiscontinuation of chemotherapy. However, of the 25 patientswith normal scintigraphic findings, only two patients had theirpumps surgically removed, as a clinical complication, be-cause of mechanical failure. The remaining 23 had no majorclinical complications other than mild to moderate clinicalsymptoms attributable to the effects of chemotherapy.

Of the 24 patients with variant findings on preoperativearteniograms, 22 had abnormal scintigraphic findings afterplacement of the pump. Only two with variant findings onpreoperative arteniognams had normal findings on perfusionscintigrams. Despite efforts (in all patients) to surgically onnonsurgically alter arterial perfusion so that only liven wouldhave homogeneous perfusion, flow patterns were normal on

perfusion scintigrams of only two patients with variant find-

A

Fig. 4.-A, Preoperative selective celiac arteriogram shows splenic artery (5), left gastric artery (LG), and common hepatic artery (CH). Main left he-patic artery (LH) arises from common hepatic artery. An accessory left hepatic artery (ALH) arises from left gastric artery (black arrows). These two yes-sels supply left lobe of liver. Note “hairpin” turn (straightsolld white arrow)from which arises right hepatic artery (RH). Right gastroepiploic artery (openarrows) is seen along greater curvature of stomach. Gastroduodenal artery is short (curved arrow). Small curved vessel (arrowhead) is right gastricartery.

B, Postoperative digital subtraction arteriogram obtained by injecting one of two ports in dual-lumen hepatic infusion pump shows contrast materialflowing in right hepatic arterial branches (RH) as well as in vessels feeding gastrointestinal tract (arrowheads). Small vessels are branches of right gas-tric artery. Tip of one of pump’s two catheters was surgically placed in gastroduodenal artery (arrow).

C, Arteriogram obtained after transcatheter embolotherapy shows tip of 5-French catheter (straight arrow) opacifying right hepatic arteries. Stumpof embolized right gastric artery trunk (curved arrow) is seen faintly. Small coils (arrowhead) were used to occlude branches of right gastric artery. Notesurgically placed infusion pump arterial catheters (P). Transcatheter embolotherapy of this vessel allowed successful direct hepatic infusion of chemo-therapeutic agents and relieved gastrointestinal signs and symptoms (e.g., gastritis, ulcer, bleeding).

868 CIVELEK ET AL. AJR:160, April 1993

Fig. 3.-A, Anterior scintigram (enlarged)ob-tamed by injecting 99mTc...macroaggregated al-bumin into infusion pump in hepatic arteryshows no perfusion of liver but marked perfu-sion of pancreas (P), spleen (Sp), and part ofstomach (St). This was caused by thrombosisof proper hepatic artery, creating backflow andcausing perfusion of extrahepatic structures.Infusion of drugs will cause systemic and gas-trointestinal toxic effects. Scintigraphic find-ings were confirmed with a hepatic arteniogram.

B, Anterior 9Smlcsulfur colloid liver scinti-gram is included for comparison.

ings on preoperative arteniograms. In both of these patients,hepatic arterial perfusion was altered intnaoperatively: byligation of an aberrant on collateral artery in one and by mod-ified placement of the pump catheter and ligation of an aben-rant artery in the other. Attempts to alter hepatic arterialperfusion included surgical placement of pumps with dualcatheters in eight, modified placement of catheters intraop-eratively in nine, and nonsurgical occlusion of unwantedvessels by using transcatheter embobothenapy techniquespostoperatively in six patients. Attempts to alter hepatic pen-fusion through either specific placement of the catheter intra-operatively, surgical ligation of a vessel, or nonsungicaltranscatheter embolization of vessels were often unsuccess-ful (Fig. 4). Nineteen of the 22 patients with variant findingson hepatic arteniograms, on 22 of the 49 patients, nonethe-

less had clinical complications after placement of the pumpthat required additional therapeutic interventions (Table 2).

Of the 25 patients with normal findings on preoperativearteniograms, 23 had normal perfusion patterns on scinti-grams and two had abnormal patterns. Minimal uptake ofnadionuclide in the spleen was detected in one patient anduptake in the bowel in another. The patient with splenic per-fusion later had a clinical complication that necessitatedalteration and eventually discontinuation of the chemothen-apy infusion. The patient with uptake of radionuclide by thebowel did not have any significant clinical complication, andreevaluation of the perfusion study indicated an initial error ininterpretation. What was initially interpreted as slight uptakeof radionuclide by bowel was thought in retrospect to repre-sent radionuclide contamination on the patient’s abdomen.

TABLE 2: Number of Patients with Clinical Complications andTypes of Therapeutic Interventions Required to Correct theComplications

Type of Complicationand Intervention

Findings on Hepatic Arteniograms

Variant Normal Total

(n = 24) (n = 25) (n = 49)

perfusion. In 22 (92%) of 24 patients with variant findings onthe preoperative arteniogram, an abnormal flow pattern wasseen on the postoperative scintigram. Additionally, 19 of the24 patients with variant findings on preoperative artenio-grams had one on more clinical complications requiring addi-tional therapeutic intervention (Table 2). On the other hand,

� � �-�-�- only three of the 25 patients who had normal findings on pre-operative arteniognams (i.e., no arterial anomalies) had clini-

8 1 9 cal problems. One patient with a scintignam on which thespleen was faintly visualized had a gastrointestinal compli-

9 0 9 cation that necessitated alteration and eventual discontinua-4 0 4 tion of the chemotherapy. Pumps were removed because of

mechanical failure in two other patients with normal findings6 0 6 on Scintigrams. In 23 (92%) of 25 patients, normal findings

on preoperative arteniograrns were associated with normal

1 2 3 findings on perfusion scintignams, and no major clinical com-8 0 8 plications other than mild to moderate clinical symptoms9 0 9 attributable to the effects of chemotherapy were seen.

In our study, the findings on perfusion scintignams couldbe used to predict the risk and/on type of future clinical corn-plications in 22 (45%) of the 49 patients. The scintignams

showed perfusion of specific visceral organs on occlusion ofan artery. Of these, 19 patients had variant findings on pne-

AJR:160, April 1993 MISPERFUSION DURING HEPATIC ARTERY CHEMOINFUSION 869

Visceral complications (gastnitis,

ulcer, bleeding) requiring alter-ation of chemotherapy infusion

Occluded artery, requiring

streptokinase treatmentCatheter occlusionEmbolization and/or ligation of an

aberrant or collateral arteryRemoval of hepatic artery infusion

pump because of mechanical

failure

Placement of dual cathetersModified placement of catheters _______________________

Note-Some patients with variant findings on hepatic arteriograms hadmore than one clinical complication.

Discussion

High concentrations of chemotherapeutic agents infusedthrough the hepatic artery can result in a significant tumorresponse and possibly prolong survival in patients with meta-static colon cancer [1 , 3].However, success depends on direct

delivery of high concentrations of chemothenapeutic agents tothe area of involvement. These high concentrations cannot begiven systemically because of toxic side effects. Perfusionscintigraphy in which radionuclide is delivered via the hepaticartery accurately reflects the true distribution of hepatic andextrahepatic chemotherapy perfusion patterns [1 , 6].

As the goal of therapy is to deliver high concentrations of

chemothenapeutic agents to the liven only, catheters must beplaced into arteries that perfuse the entire liver. Most corn-monly, the catheter is surgically placed in the gastroduode-nal artery at the point where the artery joins the properhepatic artery. When the chemothenapeutic agent is injectedinto the reservoir of the hepatic pump, the agent will reachthe liver via the catheter. The agent is pumped into theproper hepatic artery and then the night and left hepaticarteries after being mixed with the blood entering from thecommon hepatic artery. If the preoperative hepatic artenio-gram shows any vascular anomaly or variation (e.g., a nighthepatic artery arising from the superior mesentenic artery ora left hepatic artery arising from the left gastric artery), thisapproach is modified to achieve perfusion of the entire liver.Two separate catheters may be inserted into both hepaticarteries or into an accessory hepatic artery. An unusual cob-lateral artery may be surgically ligated.

In our study, we found a high correlation between evi-dence of variant arterial anatomy on preoperative artenio-grams and abnormal findings on subsequent perfusion

scintignams. This correlation was present despite preopera-tive and/or intraopenative modifications in the catheters on inhepatic vessels and despite the intnaopenative injection offluonescein dye into the perfusion catheter to test hepatic

operative hepatic arteniognams.Factors that make perfusion scintignaphy more precise

than intnaopenative arteniognaphy for predicting specific clini-cal complications include the following: (1) Injection tech-nique.-The slow injection of small volumes of radionuclideresembles the patient’s blood flow pattern more closely thandoes the injection of larger volumes of contrast material [14].(2) Incorrect identification of the vessels-When an unusual

and complex vascular anomaly is present, even an expeni-enced radiologist or surgeon may not be able to identify allthe vessels (Fig. 4). This could lead to improper placementof the catheter and result in unwanted perfusion of visceralorgans (Fig. 2). Perfusion scintigraphy may later show theerror [1 5]. (3) A nonfunctional collateral circulation-Small,nonfunctional collateral vessels may not be visible on the mi-tial preoperative arteniogram. These collateral vessels maybecome functional when the pressure gradients changewithin the liven, possibly because of progression of the dis-ease on alterations in hemodynamics. Also, visualization ofextremely small collatenals with angiography may be impos-sible before changes in intravascular pressure induced bysurgery or subsequent chemotherapy. Arteniognams are notrepeated frequently, so follow-up perfusion scintigrams willshow these changes. (4) Nonvisualization of small communi-cating vessels-Small communicating vessels betweenlarger arteries may not be visible on standard celiac/superior mesentenic artery arteniognams because of techni-cal factors such as sensitivity of the film, dilution of contrastmedium, X-ray exposure time, small volume of contrastmedium in these vessels, manipulations of the catheter onguidewire resulting in vasoconstniction, and failure to selec-tively catheterize small branch vessels. During perfusionscmntigraphy, data are collected for relatively long periods (5-10 mm per view) and may indicate the presence of these

small vessels by showing even faint uptake of nadionuclidein the areas these vessels perfuse [11 , 15].

870 CIVELEK ET AL. AJR:160, April1993

In the presence of an anatomic variation involving the dis-tnibution of the celiac trunk and/on hepatic artery, placementof a hepatic infusion pump is technically more complicated.For example, in the case of an aberrant night hepatic artery,the smaller caliber of the artery, the tight fit of the catheter inthe lumen, and the technical problems of closing the arteniot-

omy increase the risk of arterial occlusion. Also, the anatomiclocation makes dissection of this vessel more difficult thanthat of its normal anatomic counterpart [1 5]. In the case of anaberrant left hepatic artery, dissection of the left gastric andleft hepatic arteries to ligate all collateral vessels to the lessencurvature of the stomach (so that the problems of gastnitiswill be avoided) is very difficult. Kemeny et al. [15] reportedthat in seven patients with tnifuncation of the common hepaticartery, thrombosis of the common hepatic artery developedin four and was shown by hepatic pump perfusion scintigna-phy. Visualization of extremely small collaterals by angiogna-phy may be possible only after changes in intravascularpressure induced by the surgery on possibly by progressionof the disease. Fortunately, current surgical technique withbeaded ligation makes the migration of the catheter ex-tremely unlikely [1 , 15].

Failure to place the catheter in the proper position or thepresence of an aberrant artery not only can result in failureof the therapy but also may cause serious complications.When the arteniognaphic findings were variant, 92% of thepatients in our study had abnormal postoperative flow shownby scintigraphy; normal arteniognaphic findings were associ-ated with normal flow in 23 (92%) of 25 patients. In ourstudy, without any a priori information given to the physician,scintigraphic findings in patients with abnormal artenio-graphic findings (n = 24) were highly accurate for predictingclinical complications (19 patients, or 79%).

In conclusion, the function of the hepatic artery infusionpump and the position and patency of its catheters can beassessed easily by using scmntignaphy with 99mTc�micro�spheres or 99mTc_macroaggregated albumin, and the anat-

omy of the hepatic artery can be used to predict postsungicalperfusion failures. Pre- and/or intnaoperative efforts to alterhepatic arterial perfusion before delivery of chemotherapeutic

agents via hepatic artery infusion pumps are often unsuc-cessful, as shown by pump scintignaphy. Thus, although mis-perfusion appears to be less likely when the hepatic arterialanatomy is normal, all patients who have a subcutaneoushepatic artery infusion pump implanted for delivery of chemo-

therapeutic agents should have pump scintignaphy beforechemotherapy. In particular, patients in whom preoperativeangiognams show variations of hepatic vascular anatomy areat high risk for unsuspected perfusion abnormalities andshould be closely monitored with nadionuclide studies toavoid complications from unsuspected perfusion of organsother than the liven. Should unsuspected perfusion abnormal-ities be detected with pump scintigraphy, tnanscatheten embo-lotherapy may be useful in specific patients to alter blood flow

nonsungically and correct the perfusion abnormality.

REFERENCES

1 . Daly JM, Kemeny N, Odenman P, Botet J. Long-term hepatic arterial infu-

sion chemotherapy. Arch Surg 1 984; 119:936-941

2. Breedis C, Young G. Blood supply of neoplasms in the liver. Am J Pathol

1954;30:969-9853. Austin A, Vogt D, Bukowski A, Shepard K. Hepatic artery infusion chemo-

therapy for cotorectal liver metastases. C/eve C/in J Med 1990;57:692-697

4. Thrall J, Gyves J, Ziessman H, Ensminger WD. Hepatic arterial chemo-

therapy: phanmacokinetic rationale and nadionuclide perfusion imaging.In: Nuclear medicine annuaL New York: Raven, 1984:211-235

5. Oberfield A, McCaffney J, Polio J. Prolonged and continuous intra-artenialhepatic infusion chemotherapy in advanced metastatic liver adeno-

carcinoma from colorectal primary. Cancer 1 979;44:414-4206. Kaplan WD, D’Orsi CJ, Ensmingen WD, Smith EH, Levin DC. intraarterial

radionuctide infusion: a new technique to assess chemotherapy perfusion

pattern. Cancer Treat Rep 1978;62:699-7037. Kaplan WD, Ensmingen WD, Come S. Radionuclide angiography to pre-

dict patient response to hepatic chemotherapy. Cancer Treat Rep 1980;

64:1217-12228. Ziessman HA, Thrall JH, Yang PJ, et at. Hepatic arterial perfusion scintig-

raphy with Tc-99m-MAA: use of a totally implanted drug delivery system.

Radiology 1984;1 52:167-1 72

9. Ziessman HA, Juni JE, Gyves JW, Brady TM, Ensmingen WD. Throm-bosis and streptokinase lysis during hepatic intraarterial chemotherapy:the value of perfusion scintignaphy. AJR 1985;144:1067-1068

10. Miller DL, Cannasquillo JA, Lutz RJ, Chang AE. Hepatic perfusion during

hepatic artery infusion chemotherapy: evaluation with perfusion CT and

perfusion scintigraphy. J ComputAssist Tomogr 1989;13:958-96411. Cho K, Andrews J, Williams D, Doenz F, Guy G. Hepatic arterial chemo-

therapy. Radiology I 989; 173:783-791

12. Lehnen K, Reiser M, Gebhardt U, Heuck A, Schaff J. DSA-control ofimplanted devices for arterial hepatic perfusion. Cardiovasc InterventRadiol 1987; 10:71-74

13. Michels NA. Newer anatomy of the liven and its variant blood supply andcollateral circulation. Am J Surg 1 966;112:337-347

14. Than T, Sitzmann JV, Grochow LB. A method for hepatic arterial perfusionin rats. J Surg Res 1989;47:489-496

15. Kemeny MM, Hogan JM, Goldberg DA, et at. Continuous hepatic arteryinfusion with an imptantable pump: problems with hepatic artery anoma-

lies. Surgery 1986;99:501-504