A 17-month-old neutered female Labrador Retriever was referred to the University of Florida Small Animal Hospital for treatment of a suspected intrahepatic portosystemic shunt. The dog had a 3-to 4-month history of abdominal distention that was evident on initial physical examination. Results of physical examination were otherwise unremarkable. The referring veterinarian had performed a peritineocentesis 3 weeks prior to referral, and cytologic analysis of the fluid sample revealed a pure transudate. At that time, results of a CBC were within reference limits; results of serum biochemical analysis (including measurement of bile acids concentration) included a BUN concentration of 9 mg/dL (reference range, 9.0 to 29.0 mg/dL), albumin concentration of 2.7 g/dL (reference range, 2.5 to 4.0 g/dL), glucose concentration of 125 mg/dL (reference range, 75 to 125 mg/dL), alanine transaminase activity of 192 U/L (reference range, 0 to 120 U/L), preprandial bile acids concentration of 107.4 μmol/L (reference range, 5 to 25 μmol/L), and postprandial bile acids concentration of 238.3 μmol/L (reference range, 5 to 25 μmol/L). Therefore, treatment with S-adenosylmethionine (14 mg/kg [6.4 mg/lb], PO, q 24 h) combined with silibinin-phosphatidylcholine (4 mg/kg [1.8 mg/lb], providing 35 mg of silybin A and B, PO, q 24 h)^a and clindamycin (300 mg, PO, q 12 h) was initiated, and the dog was referred for further evaluation.

The patient was positioned in dorsal recumbency without sedation and underwent transabdominal ultrasonography.^b This revealed microhepatica, dilated intrahepatic and mesenteric vasculature, an estimated large volume of peritoneal effusion, and aberrant vessels (most evident caudal to the left kidney). Splenomegaly and a scant amount of echogenic material within the gallbladder lumen were also noted. The patient was anesthetized for multiphase abdominal CT angiography.^c After endotracheal intubation, temporary apnea was induced with manual hyperventilation (duration, approx 2 min) to facilitate acquisition of detailed images. Iohexol^d (640 mg/kg [290 mg/lb]) was administered via an IV catheter placed in the left cephalic vein as a single bolus with a power injector (flow rate, 3 mL/s). Pre- and postcontrast volume-rendered CT angiographic images of the caudal portion of the thorax, abdomen, and pelvis were acquired during the arterial, venous, and delayed phases. Peak contrast enhancement in regions of interest was evaluated with bolus tracking software.^e Subsequently, 2-D coronal and sagittal multiplanar reformations, and a 3-D reconstruction of the portal system were generated.^f Review of images showed contrast enhancement of the portal vasculature during the arterial phase, attenuation of the right branch of the hepatic artery, advanced ectasia of the left branch of the hepatic artery supplying a high-flow left HAVM with both a large fistula and complex intraparenchymal nidus, and large splenorenal and portosytemic varices. Poor hepatic venous opacification was evident, including on delayed-phase images, suggesting peripheral attenuation of intrahepatic portal vasculature. Additional findings included a diminutive right branch of the hepatic artery (2.4-mm diameter), moderate splenomegaly, and a large volume peritoneal effusion. The diagnosis of a left intrahepatic HAVM with secondary portal hypertension, peritoneal effusion, multiple acquired portosystemic shunts, and microhepatica was confirmed. The dog recovered from anesthesia without apparent complications. Treatment options discussed with the owners included left partial liver lobectomy and endovascular embolization of the HAVM. The owners elected embolization and returned the dog for treatment 2 weeks later.

On readmission, the dog was prepared for the procedure. Hydromorphone was administered for sedation, and general anesthesia was induced with propofol (1.5 mg/kg [0.7 mg/lb] IV). An endotracheal tube was then placed, and anesthesia was maintained with delivery of isoflurane in oxygen. Heart rate, direct blood pressure, arterial oxygen saturation (by means of pulse oximetry), respiratory rate, and temperature were monitored throughout. The caudal portion of the abdomen and both inguinal regions were prepared for surgery. Prior to transportation of the patient to the endovascular suite, peritoneocentesis was performed with removal of 2 L of peritoneal fluid. This was performed for patient comfort and to improve compliance of the abdomen prior to the procedure. The patient was then positioned in dorsal recumbency, and the right inguinal region was prepped and draped for sterile femoral arterial access. Ultrasound-guided^b access to the right common femoral artery was achieved by use of the Seldinger technique with a 4F microintroducer.^1,g Accurate guidewire location in the right external iliac artery was confirmed with fluoroscopy.^h Femoral arterial access was then obtained with removal of the 4F dilator (sheath) over an 0.035-inch guidewire, followed by insertion of a 5F × 11-cm-long introducer sheath.ⁱ External iliac and caudal abdominal femoral arteriography was then performed with manual injection of 3 mL of a 1:1 solution of iohexol^d and sterile saline (0.9% NaCl) solution to confirm appropriate access. A 5F × 65-cm-long reverse curve angiographic catheter^j was advanced over a 0.038-inch hydrophilic guidewire, then positioned in the aortic arch; the catheter was then retracted caudally to select the celiac artery. The reverse curve of the angiographic catheter was oriented ventrally when viewing the patient from above. Celiac arteriography was performed and demonstrated predominant opacification of the left hepatic arterial inflow to the HAVM and trace flow to a diminutive right branch of the hepatic artery. The catheter was positioned so that it extended approximately 1 cm into the common hepatic artery and remained in this position for the duration of the procedure.

A 3F × 150-cm-long microcatheter^k and 0.010-inch × 205-cm-long platinum guidewire^l were advanced within the 5F angiographic catheter and used to select the left branch of the hepatic artery for diagnostic angiography. This revealed a hybrid HAVM consisting of a large, direct, apparently extrahepatic arteriovenous communication (Cho-Do type 1)² abutting the caudal left lateral liver lobe and an arteriovenular left hepatic intraparenchymal nidus (Cho-Do type IIIb; Figure 1).² Delayed portal venous angiography demonstrated hepatofugal flow via the caudal mesenteric vein to a middle hemorrhoidal (caudal rectal vein) anastomosis. The large diameter and high-velocity blood flow through the HAVM were thought to exceed safe parameters for treatment with glue embolization alone. Therefore, we elected to treat the patient with a commercially available sterile compressed gelatin sponge^m mixed with sterile saline solution manually injected via the microcatheter in an attempt to attenuate flow through the lesion to facilitate subsequent glue embolization. After 2 injections of 3 mL of the embolic solution with no observable angiographic reduction in flow, the decision was made to deploy multiple 0.018-inch fibered platinum coilsⁿ into the arteriovenous fistula moiety to further attempt to attenuate blood flow and enable glue embolization of the residual nidus. A total of 21 coils were subsequently deployed within the fistula and successfully reduced blood flow, as evaluated with fluoroscopy, to an extent that was subjectively assessed as sufficient to safely proceed with glue embolization. Repeated angiography at this time revealed the dominant hepatic artery that supplied the largest portion of the intraparenchymal nidus of the HAVM. This vessel was selected for injection and anatomically confirmed with manual angiography (Figure 1).

View Full Size

Digital subtraction angiographic images acquired during treatment by means of endovascular staged coil and glue embolization of a 17-month-old neutered female Labrador Retriever with a high-flow HAVM. A—The HAVM prior to treatment. B—Multiple 0.018-inch fibered platinum coilsⁿ have been deployed into the arteriovenous fistula moiety to attenuate blood flow. C—Appearance of the HAVM after glue embolization (3 mL; 3:1 mixture of ethiodized oil^o and n-butyl cyanoacrylate^p). Notice the lack of arteriovenous shunting at the completion of the procedure.

Citation: Journal of the American Veterinary Medical Association 251, 7; 10.2460/javma.251.7.824

• Download Figure
• Download figure as PowerPoint slide

The microcatheter was flushed with 5 mL of sterile 5% dextrose solution in water. Three mL of a 3:1 mixture of ethiodized oil^o and n-butyl cyanoacrylate^p was administered in a primed (5% dextrose solution in water) polyacrylamide syringe through the microcatheter, forming an approximately 4-cm-long cast of the HAVM nidus that finely and uniformly penetrated the arteriolar and venular segments. The microcatheter was then slowly retracted and placed in a location approximately 3 cm upstream, such that it further centrally occluded the ectatic main left hepatic arterial inflow and would prevent future recanalization. The result was complete embolization of the HAVM and contributing subsegmental hepatic arteries; the absence of blood flow to the lesion with right hepatic arterial redistribution was confirmed on postprocedural hepatic arteriography (Figure 1). Celiac arteriography was then performed, which confirmed the absence of any additional arterial communications with the nidus. The procedure time was 130 minutes.

All catheters were then removed, and the femoral arterial defect was closed routinely by means of arterial ligation after an arterial cut-down procedure. During recovery from anesthesia, a small amount of regurgitation was noted and was removed from the oropharynx with a suction unit. The dog recovered in the critical care unit overnight without further apparent complications. It was bright and alert and eating and drinking well by the next morning. The severe abdominal distension evident on initial physical examination appeared to have resolved within 24 hours of the procedure, although repeated abdominal ultrasonography at that time demonstrated a small volume of residual peritoneal effusion. The dog was discharged 2 days after the procedure with instructions for activity restriction and planned repeated CT angiography at a follow-up examination 6 to 8 weeks later. Treatment with S-adenosylmethionine (14 mg/kg, PO, q 24 h) combined with silibinin-phosphatidylcholine (4 mg/kg [providing 35 mg of silybin A and B], PO, q 24 h])^a and cephalexin (20 mg/kg [9.1 mg/lb], PO, q 12 h) was prescribed.

The owners reported that the dog appeared well; however, it developed nasal discharge 5 weeks after the procedure. On examination 6 weeks after treatment, it was found to have mucopurulent nasal discharge that was ultimately attributed to nasopharyngeal stenosis. The stenosis was thought to have been the result of nasopharyngeal reflux and regurgitation subsequent to a congenitally short palate. Computed tomography of the head and an endoscopic examination of the upper respiratory tract confirmed the presence of severe (luminal diameter, 4 mm) nasopharyngeal stenosis. The stenosis was treated with endoscopic balloon dilation to a diameter of 10 mm, followed by copious lavage of the nasopharynx with saline solution. A CBC performed at this time revealed a mild stress leukogram, with all other values within reference limits. Serum biochemical analysis, including evaluation of bile acids concentration, revealed a BUN concentration of 8 mg/dL, albumin concentration of 2.9 g/dL, glucose concentration of 87 mg/dL, alanine transaminase activity of 550 U/L, preprandial bile acids concentration of 55.1 μmol/L, and postprandial bile acids concentration of 165.7 μmol/L.

Repeated 4-phase abdominal CT angiography was performed 7 weeks after embolization treatment and demonstrated no perfusion of the HAVM, although, as expected, multiple portosystemic varices remained. Compared with results of initial CT angiography, the right hepatic arterial branch diameter had increased from 2.4 mm to 4.2 mm at its bifurcation from the common hepatic artery (Figure 2). Additionally, a large thrombus was present in the main portal vein that extended beyond the porta hepatis into the left, right, and central portal branches, causing complete obstruction of the portal vein. There was subjective enlargement of the right hepatic division and notable atrophy of the left division, compared with the initial CT angiographic images. The peritoneal effusion and splenomegaly had completely resolved. The dog continued to receive S-adenosylmethionine and commenced treatment with amoxicillin-clavulanic acid^p (14 mg/kg, PO, q 12 h) for 6 weeks.

View Full Size

CT angiographic images of the patient in Figure 1 acquired before (A) and 7 weeks after (B) endovascular treatment. A—Notice the dilated portal vein (caret) and diminutive right hepatic artery (asterisk; diameter, 2.4 mm). B—A large portal vein thrombus (caret) is now evident. Right hepatic artery diameter is 4.2 mm.

Citation: Journal of the American Veterinary Medical Association 251, 7; 10.2460/javma.251.7.824

• Download Figure
• Download figure as PowerPoint slide

Subsequent follow-up occurred via several telephone calls with the owner and included discussions regarding further treatment of the upper airway disease with serial balloon dilations and the placement of a covered nasopharyngeal stent. The owners declined these options and the dog was reported to be doing well; however, nasal discharge and stertor persisted 4 months after the balloon dilation procedure. Unfortunately, because of the persistent nasal discharge and the guarded prognosis associated with nasopharyngeal stenosis, the owners elected euthanasia 5 months after treatment of the HAVM. At the time of euthanasia, there were no apparent clinical signs related to the HAVM. A necropsy was not performed.