Combined hepatocyte-selective and blood-pool contrast agents for the CT detection of experimental liver tumors in rabbits.

PURPOSE: To determine the imaging characteristics of a new computed tomographic (CT) contrast material with both hepatocyte-selective and blood-pool components (iodinated triglyceride (ITG)-dual) versus standard iohexol. MATERIALS AND METHODS: VX2 carcinoma was inoculated in seven rabbits. Animals underwent nonenhanced, iohexol-enhanced (600 mg of iodine per kilogram of body weight), and ITG-dual-enhanced (blood-pool moiety, 100 mg of iodine per kilogram; hepatocyte-selective moiety, 100 or 200 mg of iodine per kilogram, injected 90 minutes apart) helical CT. Livers were removed, preserved in formalin, suspended in agar, and sectioned transversely at 3-mm intervals. Attenuation values for normal liver and tumors were obtained, and blinded readers evaluated images for lesions by using a modified free-response receiver operating characteristic (ROC) method. RESULTS: A total of 47 separate tumor sites were detected at pathologic examination. ITG-dual-enhanced scans obtained with 300 mg of iodine per kilogram demonstrated similar liver opacification to iohexol-enhanced scans obtained with 600 mg of iodine per kilogram, but with less lesion enhancement, which resulted in better liver-to-lesion contrast. Blinded readers had a higher sensitivity, accuracy, and area under the ROC curve for ITG-dual-enhanced scans as compared with iohexol-enhanced scans (P: <.01). CONCLUSION: ITG-dual-enhanced CT quantitatively and qualitatively improved liver lesion detection versus iohexol-enhanced CT. Future clinical trials with various human tumor types after potential approval for human use are needed to determine the ultimate role of this or other dual-mechanism contrast materials.

CT-monitored percutaneous cryoablation in a pig liver model: pilot study.

PURPOSE: To determine the safety and feasibility of percutaneous cryoablation with computed tomographic (CT) guidance in a pig liver model. MATERIALS AND METHODS: Nine angiographic balloons (mean diameter, 9 mm) were placed in the livers of seven domestic pigs (mean weight, 30.0 kg +/- 14.0 [SD]) as tumor-mimicking lesions. By using ultrasonographic and CT guidance, two 2.4- or 3.0-mm cryoprobes were placed flanking the balloon, and a 15-20-minute freezing process was performed. Hemostasis was achieved by placing absorbable cellulose fabric down the probe tract. After 24-96 hours, animals were sacrificed, and their livers were removed and were sectioned axially at 5-mm intervals for comparison with CT images. RESULTS: All animals survived the procedure without complication. No serious hemorrhage was found in any case. Ice balls were readily visualized at CT because they appeared as areas of decreased attenuation (1.0 HU +/- 20.7) when compared with areas of normal liver (48.2 HU +/- 6.3, P < .05). The mean ablative margin was 1.7 cm, and only one of nine cases, the one with probe failure, had a positive margin. Beam-hardening artifact from the metal probes was present but did not interfere with the procedure. Ice-ball size and shape corresponded closely to the area of necrosis determined at histopathologic analysis. CONCLUSION: CT-monitored percutaneous cryoablation is feasible and safe in this pig liver model.

Hepatic cryosurgery via minilaparotomy in a porcine model: an alternative to open cryosurgery.

BACKGROUND: Cryosurgery of liver lesions is becoming increasingly accepted for the ablation of liver tumors. Attempts at laparoscopic cryosurgery have been very limited and often need to be converted to open laparotomy due to the complexity of the procedure. METHODS: Seven domestic pigs were anesthetized, and 17 small (0.7 cm mean diameter) tumor mimicking agar "lesions" were percutaneously placed in the liver. Two small subcostal incisions ( approximately 2.0 cm) were placed, and an endocavitary ultrasound transducer (with a 2. 4-mm cryoprobe mounted on it) was placed on the liver surface. Lesions were localized and directly punctured with one or two cryoprobes under ultrasound guidance, and a single 15-min freeze was undertaken. The animals were then killed, and their livers were removed and serially sectioned. RESULTS: Total time for probe placement was approximately 10 min after incisions had been made. Animals tolerated the procedure well and all survived until they were killed. No intraabdominal complications were detected at exploration. Mean cryolesion dimensions were 3.0 cm (single probe) and 3.3 cm (dual probe) (p > 0.05). Positive margins were detected in one lesion treated with a single probe, and in none of the lesions treated with dual probes. Mean margins were 0.9 cm: 1.2 cm for the single probe and dual probe techniques, respectively. Liver surrounding control agar lesions demonstrated a thin rim of necrosis, approximately 0.5 mm wide. CONCLUSIONS: We conclude that minilaparotomy is an effective, safe, and simple method for performing hepatic cryosurgery in this animal model. This minimally invasive technique may benefit a subset of patients with lesions in accessible locations. Lesions in posterior locations may not be as amenable to this technique due to deterioration of ultrasound image quality in the far field.

Mucinous vacuolar change of porcine urothelium induced by regional embolism or cryoablation.

Porcine urinary tract epithelium responds to systemic infections with an increase in mucin secretion that has been called "mucinous degeneration." Here we describe similar changes in calyceal, pelvic, and ureteric urothelium in pig renal tracts 14 days after local embolism and cryoablation (four kidneys) or cryoablation alone (six kidneys). Large mucin-filled vacuoles surrounded by smaller cytoplasmic vesicles were present in the affected urothelium but were rare in calyceal urothelium of normal kidneys. These mucinous vacuolar changes were proportional to the extent of renal necrosis and could be important in the pathologic responses of xenotransplanted pig kidneys.

Evaluation of an experimental low-attenuation gastrointestinal contrast agent for CT imaging of intestinal ischemia in an animal model.

RATIONALE AND OBJECTIVES: The authors evaluated the use of MRX-320, a low-attenuation, expansile oral contrast agent, for the demonstration of intestinal ischemia in an animal model. MATERIALS AND METHODS: Nine dogs were given either MRX-320, water, or diatrizoate as an oral contrast agent through a jejunostomy tube. Two dogs received no oral contrast agent. Helical computed tomography (CT) was performed before the intravenous injection of 2.5 mL/kg iohexol at 4 mL/sec, during the arterial phase, and during the portal venous phase. Mesenteric ischemia was surgically induced, and the imaging protocol was repeated. Three readers rated the randomly assigned images for quality and demonstration of ischemia. Attenuation values for the intestinal lumen and wall were recorded. RESULTS: Examinations performed with MRX-320 provided the best discrimination between ischemic and nonischemic conditions (P < .05), followed by examinations with no oral contrast medium, examinations with water, and examinations with diatrizoate. Images obtained with MRX-320 also scored significantly higher on measures of image quality than those obtained with water or no oral contrast medium (P < .05). On images obtained with MRX-320, the bowel lumen measured -836.5 HU (P < .05 compared with other techniques). Water provided the least uniformity of distention, and diatrizoate provided the least mucosal detail. CONCLUSION: The use of MRX-320 as an oral contrast agent with an intravenous bolus of iohexol at CT increased reader confidence for the diagnosis of intestinal ischemia and improved subjective measures of image quality.

Laparoscopic cryosurgery of the kidney in the porcine model: an acute histological study.

OBJECTIVES: To verify histologically whether cryosurgery of the kidney can be accomplished reproducibly without injuring adjacent structures, using a combination of ultrasound and laparoscopic guidance. MATERIALS AND METHODS: Six kidneys from three domestic female farm pigs were utilized in the study. Under general anesthesia and after obtaining pneumoperitoneum, the lower pole of the kidney was mobilized laparoscopically and the ureter and adjacent bowel were protected with saline-soaked gauze. Two 3.8 mm-cryoprobes were placed percutaneously into the lower pole and cryoablation was carried out under laparoscopic and ultrasound guidance using a double-freeze technique (10-minute freeze and 5-minute thaw cycles to a probe temperature of -185 degrees C to -196 degrees C) in five kidneys (one control). The kidneys, adjacent ureter and bowel were harvested acutely, and macroscopic, histologic, and electron microscopic evaluation of all specimens was performed. RESULTS: Macroscopically, clear margins of cryodestruction corresponded with the ultrasound images of the iceball. In all five treated kidneys, reproducible cell death corresponding to visible margins of cryodestruction were verified histologically. Cell death was further corroborated by electron microscopy. Adjacent structures (ureter and bowel) were sectioned and no significant damage was noted. Blood pressure remained constant throughout the procedure. A crack in the renal parenchyma of one kidney was noted during the thaw phase; at harvest that animal was found to have an intraperitoneal hemorrhage. CONCLUSION: Combined laparoscopic and ultrasound-guided cryoablation of the lower pole of the kidney can be accomplished reproducibly in the porcine model without injury to adjacent structures.

Hepatic cryoablation: US monitoring of extent of necrosis in normal pig liver.

PURPOSE: To determine the accuracy of ultrasonography (US) for prediction of hepatic tissue necrosis after cryoablation in normal pig liver. MATERIALS AND METHODS: Five normal pig livers were treated with cryoablation monitored with US. After a single freeze cycle at 50% flow capacity, the widest diameter of the cryolesion was identified and marked with wire placement (22 wires in five lesions). Livers were removed 24 hours later, and wire tracks were marked with India ink. Livers were sectioned, and the distance was measured between wire tracks and tissue necrosis caused by freezing. RESULTS: The mean volume of areas of tissue necrosis was 11.6 cm3 +/- 4.0, the mean diameter was 2.9 cm +/- 1.0, and the mean maximum diameter was 2.9 cm +/- 0.7. The mean distance between the edge of necrosis and the wire track was 1.1 mm +/- 1.4. By excluding one outlier (6.5 mm), the mean distance from the ice ball to the necrotic area was 0.8 mm +/- 0.8. Uniform necrosis of hepatic parenchyma within the cryolesion was confirmed. CONCLUSION: US can be used to predict reliably the size of the necrotic area after hepatic cryoablation in normal pig liver. Knowledge of a small but consistent underestimation of tissue necrosis is important when planning cryoablation.

Lipid-based blood-pool CT imaging of the liver.

RATIONALE AND OBJECTIVES: We have recently developed an iodinated lipid-based contrast agent capable of residing in the blood pool for extended periods of time relative to conventional water-soluble contrast agents. The purpose of this study was to examine the effects of combining this new blood-pool agent (ITG-PEG) with a hepatocyte-selective agent (ITG-LE; Molecular Biosystems) for accurate CT detection of small (< 10 mm) VX2 tumors in rabbit liver. MATERIALS AND METHODS: Preliminary pharmacokinetic analyses were conducted in SD rats (12) by injection of either I-125-labeled ITG-PEG or I-125-labeled ITG-LE followed by subsequent blood collection and quantification of radioactivity. Preliminary CT studies were conducted in both normal (3) and tumor-bearing NZW rabbits (2). Tumor-bearing rabbits were laparotomized and VX2 cells injected directly into the hepatic parenchyma to produce a total of eight focal lesions (2-10 mm diameter). Animals underwent CT scanning 10 days later with multiple techniques including noncontrast and helical i.v. enhanced (600 mg I/kg iohexol), and then 24 hours later using both ITG-PEG and ITG-LE (200 mg I/kg). Tissue density measurements (HU) of liver, tumor, and blood (descending aorta) were acquired in each case for comparison. Tumor morphology was verified by gross pathologic inspection. RESULTS: Pharmacokinetic analysis in rats as well as CT studies in normal rabbits revealed that ITG-PEG remains in the blood-pool phase for more than 2 hours following i.v. administration. In fact, blood density in normal rabbit obtained with ITG-PEG was 95.1 HU +/- 5.8 at 120 minutes compared to 90.7 HU +/- 6.1 immediately after injection. Although liver enhancement was greater with iohexol (67 HU within 1 minute of injection), than for ITG-PEG/ITG-LE (32 HU, 60 minutes postinjection), liver to lesion ratios favored ITG-PEG/ITG-LE due to significant enhancement of tumor itself with iohexol (+40 HU). Tumor enhancement was minimal with ITG-PEG/ITG-LE. Lesions were subjectively much better defined with ITG-PEG/ITG-LE with sharper edge definition. CONCLUSION: In these animal models, a new iodinated lipid-based contrast agent composed of both blood pool and hepatocyte-selective components afforded favorable CT imaging results compared to a conventional urographic agent, albeit at one-third the total iodine dose.

Monitoring renal cryosurgery: predictors of tissue necrosis in swine.

PURPOSE: To examine parameters for monitoring renal cryosurgery and correlate with histopathologic necrosis in a porcine model. Parameters include: 1) tissue temperature, 2) distance of tissue from the cryosurgical probe insertion site, and 3) inclusion of tissue by the visible iceball. MATERIALS AND METHODS: Following midline incision, 6 healthy kidneys from 3 domestic pigs were treated using a nitrogen-based cryosurgical system with 3.4 mm outer diameter cryoprobe. Temperatures were monitored at 6 sites in each kidney using prototype thermocouples. Gross and histologic analysis was performed on tissue harvested from each thermosensor site 24 hours postoperatively. RESULTS: All animals survived to complete the full protocol. Post-procedure bleeding was managed surgically. Histopathology revealed that complete ablation was uniformly produced at temperatures of -19.4C or lower in 13/13 tissue samples. Within 16 mm. of the probe insertion site, cells were uniformly ablated in 17/17 samples while degree of tissue death varied widely beyond this margin. Cell death was more likely found at points encompassed by the visible iceball (16/18, 89%) than those beyond it (2/18, 11%) [p <0.0001, Chi-square] while 2/18 (11%) of samples harvested from within the iceball margin were partially viable. CONCLUSIONS: Temperature monitoring using thermocouples during porcine renal cryosurgery demonstrated complete homogeneous necrosis of tissues reaching -19.4C or lower. Distance beyond 16 mm. from the cryoprobe and direct visualization of the iceball proved to be less reliable predictors of tissue necrosis. Management of bleeding post-thaw was necessary in every case.

Laparoscopic renal cryotherapy in swine: comparison of puncture cryotherapy preceded by arterial embolization and contact cryotherapy.

Variable results of open, percutaneous, and laparoscopic cryosurgery of the kidney have been reported. Minimally invasive approaches to lower-pole renal ablation were prospectively compared in swine: laparoscopic puncture cryoablation preceded by arterial embolization, laparoscopic contact cryosurgery, and arterial embolization alone. Eighteen kidneys in nine domestic pigs were randomized to one of three treatment groups of six organs each: (1) puncture cryotherapy preceded by selective lower-pole arterial embolization; (2) contact cryotherapy alone; and (3) arterial embolization alone. Under general anesthesia, appropriate animals underwent selective arterial embolization of the lower pole via femoral cutdown using Gelfoam pledgets. Animals randomized to cryotherapy underwent dissection of the lower pole of the kidney using a single 12-mm umbilical and two midclavicular-line ports in the lateral position. Under external ultrasound guidance, two 3-mm cryoprobes (Cryomedical Sciences, Rockville, MD) were positioned in the lower pole of the kidney, and double-freeze technique to -190 degrees C was performed using puncture or contact application. Ultrasonography was used to guide probe insertion during puncture cryotherapy only. Kidneys were harvested 11 to 14 days after the procedure. Both puncture and contact kidneys demonstrated cell death and subsequent necrosis by light microscopy and electron micrography. On average, puncture lesions were heavier than contact lesions, 19.3 g v 10.1 g (P = 0.02; unpaired t-test), whereas the kidneys were equivalent in weight: 74.1 g v 74.1 g. Puncture lesions represented 26.4% of total kidney weight v 13.5% in the contact group (P = 0.002; unpaired t-test). Only focal infarcts were identified in the embolization group. No evidence of adjacent visceral injury was identified in any of the groups, and no animal required conversion to open laparotomy. One kidney bled after contact cryotherapy, and hemostasis was achieved laparoscopically. Laparoscopic cryotherapy is reproducible and effective using either the puncture technique preceded by arterial embolization or the contact technique, without evidence of damage to adjacent structures. Contact lesions were less voluminous and more peripheral than puncture lesions. Embolization alone produces focal infarcts and focal zones of cell death.

Minimally invasive alternatives to traditional operative hepatic cryoablation: works in progress.

Primary and secondary malignant disease of the liver remains a major health problem in the United States and abroad. It is estimated that over 130,000 new cases and 55,000 deaths will result from colorectal carcinoma in 1997. In this same year, 13,600 new cases and 12,400 deaths are also expected to result from primary hepatic tumors. Worldwide, hepatocellular carcinoma results in approximately 250,000 deaths yearly.

Perivascular and intralesional tissue necrosis after hepatic cryoablation: results in a porcine model.

BACKGROUND: Cryosurgical ablation of malignant hepatic tumors is being increasingly used for definitive treatment of metastatic colorectal and primary hepatic tumors. The lack of tumor necrosis near vessels that results from inadequate freezing may contribute to local recurrence and thus limit the applications of this therapy. This study was designed to determine whether single-freeze cryoablation could cause necrosis of both the pervascular and intralesional hepatic parenchyma. METHODS: Ten pigs were treated with one 15-minute cycle of cryoablation. Five additional animals were treated with overlapping cryolesions to simulate a double freeze. After 24 hours, animals underwent reoperation with portal vein cannulation and infusion of formalin. Serial sectioning and hematoxylin and eosin staining of cryolesions were performed. RESULTS: Complete cell death was visualized within all cryolesions. There was no difference between once or twice-frozen tissue. Vessels within or adjacent to cryolesions showed necrosis of hepatic tissue up to the vessel wall. No sections revealed incomplete necrosis of perivascular hepatic parenchyma. CONCLUSIONS: Single-freeze cryoablation results in necrosis of intralesional hepatic parenchyma without added benefit from repeat freezing. Complete necrosis of the perivascular tissue suggests that cryosurgical ablation can effectively cause necrosis immediately adjacent to vessels without concerns of incomplete ablation resulting from the heat sink effect.

CT depiction of experimental liver tumors: contrast enhancement with hepatocyte-selective iodinated triglyceride versus conventional techniques.

PURPOSE: To compare findings at computed tomography (CT) enhanced with a hepatocyte-selective contrast agent (iodinated triglyceride) and/or iohexol and at CT during arterial portography (CTAP). MATERIALS AND METHODS: Rabbit livers were directly inoculated with VX2 carcinoma. Results were compared for five helical CT examinations: unenhanced CT, iohexol-enhanced CT (600 mg iodine per kilogram of body weight [mg I/kg]), CTAP (with iohexol [600 mg I/kg]), triglyceride-enhanced CT (126 mg I/kg), and dual-contrast-enhanced CT (triglyceride plus iohexol [425 mg I/kg]). Attenuation of normal liver and tumor were compared with analysis of variance techniques and blinded reader evaluations. RESULTS: Normal liver attenuation was greatest at CTAP (127.3 HU +/- 5.3 [mean +/- standard error of the mean]), followed by dual-contrast-enhanced CT (112.4 HU +/- 1.2), iohexol-enhanced CT (97.9 HU +/- 2.2), triglyceride-enhanced CT (82.3 HU +/- 1.1), and unenhanced CT (54.9 HU +/- 1.8). Liver-to-lesion attenuation difference at triglyceride-enhanced CT was significantly greater than at iohexol-enhanced CT (P < .05), and attenuation differences at dual-contrast-enhanced CT were comparable to those at CTAP. Tumors did not enhance at triglyceride-enhanced CT, which increased conspicuity. Sensitivity values for lesion detection at dual-contrast-enhanced CT were greater than those at iohexol-enhanced CT or at CTAP (P < .05). CONCLUSION: At CT enhanced with triglyceride (especially when combined with iohexol), sensitivity values and liver-to-lesion attenuation differences were greater with lower iodine doses than with iohexol or at CTAP.

Preclinical evaluation of an iodinated particulate contrast agent for use during angiography: work in progress.

PURPOSE: To study the feasibility of using an iodinated particulate contrast agent, iodipamide ethyl ester (IDE), for angiography. MATERIALS AND METHODS: IDE at doses of 40-100 mg of iodine per kilogram was diluted to a total volume of 5-20 mL and used for digital subtraction angiography in nine dogs under general anesthesia. Equivalent images were obtained by using water-soluble contrast medium (WSCM) for comparison (iohexol) in seven animals. All images were reviewed by blinded reviewers and graded subjectively on a five-point scale. RESULTS: Angiographic studies of multiple vascular territories performed with IDE yielded images of slightly lower overall quality compared with images obtained with WSCM (P = .14, Mann-Whitney U test). Arterial phase images were subjectively superior with WSCM when compared with IDE (P < .0001, chi 2.) Depiction of the corresponding veins during the venous phase on the IDE angiograms was superior to that on WSCM angiograms in 12 of 21 cases, although this did not reach statistical significance (P > .05 chi 2). Images of the renal vein and portal vein achieved with IDE were graded as superior to those achieved with WSCM in eight of 10 reviews. CONCLUSION: Angiography is feasible with IDE. Compared with WSCM, IDE produced images of lesser quality during the arterial phase, but of equal or superior quality in the venous phase depending on the vessel studied. Because it is excreted slowly in bile and is isotonic, it may prove useful in patients with renal insufficiency, diabetes, multiple myeloma, or severe coronary disease.

Conventional color Doppler velocity sonography versus color Doppler energy sonography for the diagnosis of acute experimental torsion of the spermatic cord.

OBJECTIVE: We compared color Doppler velocity sonography and color Doppler energy sonography for the diagnosis of spermatic cord torsion in a canine model and determined the degree of torsion necessary to acutely halt testicular blood flow. MATERIALS AND METHODS: Spermatic cord torsion was created in five dogs by exposing and rotating the ipsilateral testis 0 degree, 180 degrees, 270 degrees, 360 degrees, 450 degrees, and 540 degrees. Detorsion followed. The testicles were scanned at each torsion stop using both color Doppler velocity sonography and color Doppler energy sonography. Doppler parameters were optimized (by phantom and test scans) and maintained at a tolerable noise level throughout the experiment. Readers who were unaware of the degree of torsion compared flow in the rotated and contralateral control testes. RESULTS: Flow became undetectable by color Doppler velocity sonography and color Doppler energy sonography at 450 degrees in four of five cases and at 540 degrees in one of five cases. We found no significant difference between the velocity and the energy techniques for detecting this absence of flow (p > .05, Wilcoxon test). We found a significant difference in degree of flow for both techniques when comparing controls and all degrees of torsion combined (p < .006, Mann-Whitney test), but significance was achieved at lesser degrees of torsion with the velocity technique than with the energy technique (180 degrees and 360 degrees, respectively, Wilcoxon test). CONCLUSION: Color Doppler energy sonography was not significantly more sensitive than color Doppler velocity sonography for the diagnosis of spermatic cord torsion in this model. Complete occlusion of arterial inflow occurred at 450-540 degrees of torsion.

MR hysterosalpingography in a rabbit model.

Our objective was to evaluate the efficacy of MR imaging as compared with conventional hysterosalpingography (HSG) for the detection of fallopian tube patency after uterine injection of contrast material. Rabbit uterine horns (n= 18) were catheterized transvaginally. Five fallopian tubes were ligated and 11 were left unaltered. T1-weighted gradient-echo MR images were obtained before, during, and after injection of 1.0-3.0 mL of a dilute gadolinium-containing contrast agent. Corresponding conventional studies were performed with an equivalent volume of iohexol. Images were evaluated by two blinded readers. Observers agreed in all cases on the presence (n = 11) or absence (n = 5) of peritoneal spill with conventional HSG. Interpretation of MR HSG concurred with conventional HSG in 14 of 16 cases for each observer (P > .05). Reasons for misdiagnosis included small amounts of spill (n = 2), artifact (n = 1), and subtle spill between bowel loops (n = 1). Sensitivity and specificity for MR HSG were 95.5% and 70%, respectively, for the diagnosis of tubal patency/occlusions. No statistical difference was found between MR HSG and conventional HSG for the diagnosis of fallopian tube patency/obstruction. Potential advantages of MR HSG include no ionizing radiation, potentially diminished local contrast toxicity, superior visualization of uterine fibroids and endometriosis, and visualization of ovaries. We conclude that this technique warrants further study, including the use of a primate model to better simulate human anatomy.

Lesion visualization by targeted computed tomography liver enhancement in dogs.

RATIONALE AND OBJECTIVES: We conducted a pilot study to determine the potential advantages of using liver-specific targeted computed tomography (CT) contrast agents for lesion detection. METHODS: Eight dogs had liver infarcts created by percutaneous injections of ethanol. Each dog underwent CT scans with four imaging techniques: unenhanced, intravenous contrast enhanced (IVCE), CT arterial portography (CTAP), and targeted liver enhancement with iodipamide ethylester (IDE) particles. Lesions were assessed quantitatively to determine liver-to-lesion density differences and the drop in density across liver edge as a quantitative measure of edge sharpness. Expert readers subjectively analyzed data to determine lesion visibility and edge sharpness. RESULTS: Liver-to-lesion density differences were greatest with CTAP (56.4 +/- 35.5 Hounsfield units [H]) followed by IDE (41.1 +/- 7.0 H), i.v. (22.7 +/- 6.0 H), and unenhanced scans (13.6 +/- 4.1 H; ps < .05 for CTAP versus unenhanced and IDE versus unenhanced). Edges were best defined both subjectively and quantitatively on IDE-enhanced scans. CONCLUSION: Targeted liver-specific contrast agents have potential to increase lesion visibility when compared with standard i.v. contrast enhancement of the liver by increasing lesion edge definition and liver-to-lesion attenuation differences. Further work in animal tumor models, and clinical trials as agents become available, appears justified.