Restoration of hepatopetal portal flow in a cat with left hepatic arteriovenous malformation following left hepatic division resection and staged attenuation of acquired portosystemic shunts

Abstract

A 5-month-old Munchkin cat was referred for hematuria. Computed tomographic angiography revealed hepatic arteriovenous malformation (HAVM) within the left lateral liver lobe and multiple acquired portosystemic shunts (APSSs), in addition to a bladder calculus. Following medical management, the cat underwent attenuation of arterioportal communications. Because ligation of the left hepatic artery did not normalize portal pressure, the left hepatic division was resected. Surgical ligation of a subset of the APSSs was performed 1 month later. Postoperatively, clinical signs improved, hepatopetal portal flow was restored, and biochemical liver function parameters normalized. The remaining APSSs regressed spontaneously. To our knowledge, based on the English-language literature, this is the first reported case of feline or canine HAVM in which liver function parameters normalized.

Hepatic arteriovenous malformation (HAVM) is an uncommon congenital anomaly in which a nidus—a dense network of dysplastic arteries—drains directly into branches of the portal vein. This leads to arterial blood entering the portal system, causing portal hypertension (PH) and the emergence of acquired portosystemic shunts (APSSs) [1, 16]. In previous reports, nearly 40 cases of canine HAVM have been treated with feeding artery ligation, embolization, or hepatic lobectomy [2, 3, 6, 14]. In contrast, management of feline HAVM has rarely been reported; ligation and coil embolization of feeding arteries have been documented in only a single case each, with clinical signs resolving but liver function tests remaining abnormal [12, 18]. We herein report a feline case of HAVM managed via hepatic lobectomy and staged surgical ligation of APSSs, resulting in the restoration of hepatopetal portal perfusion and liver function.

A 5-month-old intact male Munchkin, weighing 1.6 kg (BCS=2/5), was referred to the Hokkaido University Veterinary Teaching Hospital for evaluation of hematuria (Day 1). The cat had persistent hematuria for 6 weeks. Urine culture yielded Staphylococcus pseudintermedius; antibiotics were administered based on susceptibility testing, but clinical signs persisted. Additionally, the cat was noted to be developmentally stunted compared with his littermates.

Upon presentation, physical examination revealed mild emaciation and abdominal distension. Abdominal ultrasonography showed a 3-mm calculus and suspended material resembling a blood clot within the urinary bladder. It also revealed tortuous, pulsatile abnormal vessels within the liver, hepatofugal flow of the splenic vein, and ascites. Based on these findings, HAVM with associated ammonium urate urolithiasis was suspected, and further workups to confirm the diagnosis were performed on Day 16. Serum biochemistry showed slight increases in liver enzyme activities, and elevated ammonia (NH₃) and total bile acids (TBA) concentrations in both fasting and postprandial samples (Table 1). Coagulation testing showed a prolonged activated partial thromboplastin time. Additional testing revealed markedly decreased factor XII activity, leading to a diagnosis of congenital factor XII deficiency.

Table 1. Selected hematologic, biochemical, and coagulation parameters at pre- and post- procedure timepoints.

Parameter		Day 16	Day 58	Day 154	Day 842	Feline reference interval
Hematocrit		29.3	30.9	37.5	40.1	30.3–52.3 (%)
White blood cell (WBC)		11,110	14,520	12,730	5,090	2,870–17,020 (/μL)
Platelets		14.9	24.2	22.8	15.1	15.1–60.0 (×104/μL)
Total protein (TP)		5.7	5.9	6.8	7	5.7–7.8 (g/dL)
Albumin		2.3	2.5	2.9	3.3	2.3–3.5 (g/dL)
Glucose		96	91	98	105	71–148 (mg/dL)
Blood urea nitrogen (BUN)		16.6	14.1	19.5	23	17.6–32.8 (mg/dL)
Creatinine		0.6	0.5	0.9	1.1	0.9–2.1 (mg/dL)
Alanine aminotransferase (ALT)		99	51	63	50	22–84 (U/L)
Alkaline phospatase (ALP)		128	155	92	19	27–125 (U/L)
Ammonia		195	65	10	N/A	28–78 (μg/dL)
Total bile acids (TBA)
	Preprandial	84.3	2.8	2.1	N/A	≤4.7 (μM)*
	Postprandial	47.1	20.4	4	3.2	≤9.1 (μM)*
Prothrombin time (PT)		9.9				9.1–10.7 (sec)
Activated partial thromboplastin time (APTT)		83.1				21.6–58.8 (sec)
Fibrinogen		151				67–203 (mg/dL)

Values in bold indicate results outside the normal feline reference interval. *TBA reference intervals were derived from a previous study using the same assay platform [4].

Computed tomographic angiography (CTA) was performed using an 80-slice helical scanner (Aquilion PRIME, Canon Medical Systems, Otawara, Japan). The cat was anesthetized with propofol (to effect; PropoFlo28, DS Pharma Animal Health, Osaka, Japan) and maintained with inhalational isoflurane (IsoFlo, Zoetis Japan, Tokyo, Japan). Rocuronium (0.1 mg/kg IV; Eslax, MSD, Tokyo, Japan) was administered to eliminate spontaneous respiration. Following the pre-contrast scan, iopamidol (Iopamidol 150, Fuji Pharma, Tokyo, Japan) was administered at 4 mL/kg (600 mgI/kg) via the cephalic vein using a power injector over 20 sec. Postcontrast scans, including triple arterial phases, a portal phase, and a delayed phase, were obtained using a bolus-tracking technique. The trigger threshold was set at 200 Hounsfield units in the abdominal aorta. The early arterial phase scan was automatically initiated 3 sec after the trigger; the middle and late arterial phase scans were then obtained at 3-sec intervals. The portal and delayed phases began 40 sec and 3 min after the trigger, respectively. Each scan extended from the thoracic inlet to the pelvis. The cat was administered sugammadex (4 mg/kg IV; Bridion, MSD, Tokyo, Japan) to reverse the rocuronium-induced neuromuscular blockade and was allowed to recover. Imaging showed that the entire liver was underdeveloped. On postcontrast images, an aneurysmally dilated portal vein and numerous small tortuous vessels surrounding it (i.e., a nidus) were observed within the left lateral lobe (Figs. 1 and 3A). These vessels, along with the cranial and caudal mesenteric veins, were enhanced during the early arterial phase, whereas no contrast opacification was seen in the jejunal vein, consistent with arterioportal communications. The celiac artery and the left hepatic artery were dilated, and the aortic diameter was reduced caudal to the origin of the celiac artery, implying diversion of a considerable amount of blood through the celiac artery. It was difficult to identify the branching points of the numerous small tortuous vessels from the left hepatic artery. Although most were considered to branch within the left lateral lobe, some were suspected to originate at the hepatic hilum. In addition, multiple tortuous shunt vessels were observed between the left portal vein and the internal thoracic vein; between the splenic vein and the left gonadal vein; between the left gastric vein and the azygos vein; and between the cranial mesenteric vein and caudal vena cava (CVC), and gallbladder varices were also present. Based on these findings, the cat was diagnosed with HAVM and multiple APSSs, and medical therapy was initiated with amoxicillin (13.7 mg/kg, BID; Amoxiclear, Kyoritsu Seiyaku, Tokyo, Japan) and lactulose (0.5 mL/head, TID; Kowa, Nagoya, Japan). For the hematuria, considering the possibility of idiopathic renal hemorrhage, enalapril (0.3 mg/kg, SID; Sandoz, Tokyo, Japan) was prescribed, and the condition improved within a few days.

Fig. 1.

Three-dimensional computed tomographic image reconstructed from the middle arterial-phase acquisition. The terminal portion of the left portal vein is aneurysmally dilated (asterisk) and is surrounded by small, tortuous vessels forming a nidus. The left hepatic artery (red arrow) and left portal vein (purple arrow) are dilated. Multiple acquired portosystemic shunts are also present (green arrows).

Fig. 3.

Maximum-intensity projection images from portal-phase acquisition (window level: 160 Hounsfield units [HU]; window width: 400 HU; slab thickness: 7.5 mm). (A–C) Images obtained before the first surgery, before the second surgery, and 3 months after the second surgery, respectively. The dilated left portal vein is indicated by an asterisk (A). The splenic vein–left gonadal vein shunt (arrow) and the cranial mesenteric vein–caudal vena cava shunt (arrowhead) decreased after the first surgery and disappeared by the final examination. Note that the right hepatic division was markedly enlarged after the first surgery.

To attenuate the arterioportal communications, surgery was planned to first occlude the dilated left hepatic artery under real-time portal pressure monitoring. If portal pressure did not normalize, the plan was to proceed with resection of the left hepatic division, based on the assumption that some feeding arteries branching from the left hepatic artery at the hilum were likely to remain but would be difficult to visualize and ligate individually. The APSSs were planned to be ligated or attenuated either during the same operation or in staged procedures, depending on the portal pressure after attenuation of the arterioportal communications. Levetiracetam (22 mg/kg, TID) was started 3 days before surgery. The blood type was A, and preoperative cross-matching was performed. A laparotomy was undertaken on Day 27.

The cat was premedicated with fentanyl (2.5 μg/kg IV; Terumo, Tokyo, Japan), and anesthesia was induced with propofol (to effect) and maintained with 1.0–2.0% isoflurane in 50% oxygen mixed with medical air. Intraoperative analgesia was provided via a continuous rate intravenous infusion (CRI) of fentanyl (10–15 μg/kg/hr), combined with a local anesthesia using rectus sheath block with levobupivacaine (2 mg/kg; Popscaine, Maruishi, Osaka, Japan). During the procedure, lactated Ringer’s solution was infused intravenously (3 mL/kg/hr), and ampicillin (20 mg/kg IV; Viccillin, Meiji Seika Pharma, Tokyo, Japan) was administered at induction and repeated every 2 hr during the operation. A midline celiotomy was performed. During dissection of the falciform ligament, the shunt vessel between the left portal vein and the internal thoracic vein was ligated. Highly vascularized APSSs were observed in the mesentery and around the left renal vein, and the pancreas was edematous, substantiating portal hypertension. An indwelling catheter was placed in the jejunal vein to start portal pressure monitoring; pulsations were noted in the pressure waveform, and the mean portal pressure was markedly elevated at 38 mmHg. The left lateral and medial hepatic lobes were dark red, and an aneurysmally dilated portal vein was observed within the left lateral lobe (Fig. 2A and 2D). To facilitate access to the hepatic hilum, the left triangular ligament was incised from the lateral aspect toward the coronary ligament. While dissecting the most medial portion of the triangular ligament at its junction with the coronary ligament, the CVC was inadvertently perforated, resulting in hemorrhage. The defect was managed by primary suturing of the CVC and administration of an emergency whole-blood allotransfusion, after which the procedure was resumed. At the hepatic hilum, the Glisson’s sheath was incised, and the dilated left hepatic artery was isolated. When the left hepatic artery was temporarily occluded with vascular forceps, the mean portal pressure decreased to 23 mmHg and the pulsation of the pressure waveform decreased (Fig. 2B and 2E); therefore, the vessel was ligated and divided with a 3–0 braided nylon suture. Because portal pressure remained elevated and the portal waveform was still pulsatile, incomplete attenuation of the arterioportal communications was suspected; this was confirmed by a further decrease in portal pressure to 15 mmHg after occlusion at the base of the left liver division (Fig. 2C and 2F). Therefore, resection of the left hepatic division was performed with en bloc ligation at the hilus. No intraoperative complications occurred other than hemorrhage from the CVC, and the cat recovered uneventfully from anesthesia and resumed voluntary eating the day after surgery. Because no postoperative complications were observed, the cat was discharged on the 8th postoperative day (Day 35).

Fig. 2.

Intraoperative photographs (A–C) and corresponding physiologic monitor recordings (D–F). P1 and P2 represent aortic and portal blood pressure, respectively. Following dissection of the left triangular ligament (A), the left lateral lobe (LLL) and left medial lobe (LML) appear darker than the quadrate lobe (QL), and a varicose projection of the portal vein is visible on the LLL (arrow). Portal pressure is markedly elevated, with pulsations evident on the portal pressure waveform (D). Clamping the left hepatic artery with a Satinsky clamp reduces portal pressure and diminishes pulsatility, without altering aortic blood pressure (B, E). Occlusion of blood flow at the base of the left hepatic division further decreases portal pressure (C, F).

Histopathological analysis of the resected liver lobes revealed severe cystic dilatation of the portal vein, proliferation of tortuous bile ducts and arteries, and fibrosis of the liver parenchyma, consistent with a clinical diagnosis of HAVM. At the periphery of the lobe, the lobular structure was small and irregular, and the diameter of the portal veins was markedly decreased, suggesting portal hypoplasia.

One month after surgery (on Day 58), the cat showed a marked improvement in activity level. Ultrasonography confirmed the conversion of splenic vein blood flow to a hepatopetal direction. Repeated CTA revealed an increase in liver volume, reduction of the APSSs, and a decrease in ascites (Fig. 3B). Although fasting NH₃ and TBA concentrations normalized, postprandial TBA remained marginally elevated (Table 1). Consequently, the patient underwent bladder stone removal and castration, during which additional ligation of APSSs was also performed on Day 62. The anesthesia protocol was the same as in the first surgery, except for the additional use of a CRI of medetomidine (3 μg/kg/hr; Domitor, Nippon Zenyaku Kogyo, Koriyama, Japan). As in the initial surgery, a laparotomy was performed under general anesthesia, and a catheter was placed in the jejunal vein; portal pressure measured 15 mmHg. When the splenic vein–left gonadal vein shunt was ligated, no significant change in portal pressure was observed. Subsequently, portal angiography was conducted; iohexol (Omnipaque 300, GE Healthcare, Oslo, Norway) was administered at 2 mL/kg (600 mgI/kg) via the jejunal vein under fluoroscopic monitoring. Some contrast flowed into a cranial mesenteric vein–CVC shunt (Fig. 4), but its proximity to the root of the mesentery made ligation impractical, so no procedure was performed to attenuate it. Additionally, the left gastric vein–azygos vein shunt was left untreated due to its small blood flow volume. Cystotomy and castration were performed without complication. The removed calculus consisted of ammonium urate. The cat was discharged on the 3rd postoperative day (Day 65), and medical therapy was discontinued.

Fig. 4.

Digital subtraction angiography image following administration of contrast medium into the jejunal vein (asterisk). The portal vein and right portal branches (purple arrows) are enhanced. Additionally, contrast flows into the CVC (blue arrow) through the cranial mesenteric vein–CVC shunt (green arrow).

Three months after the second surgery (Day 154), the cat achieved an optimal body condition with a weight of 2.7 kg (BCS=3/5), and blood tests showed normalization of NH₃ and TBA concentrations (Table 1). On repeated CTA, the untreated APSSs were no longer visualized, suggesting spontaneous regression; ascites had also resolved (Fig. 3C). At the time of the last follow-up, 26 months after the second surgery (Day 842), postprandial TBA remained within the reference interval, and the cat was clinically doing well.

HAVM, also called a hepatic arteriovenous or arterioportal fistula, has been sporadically reported in small animals. By definition, arteriovenous malformation refers to a congenital dense network (nidus) between arteries and veins, whereas an arteriovenous fistula is a single communication and is more often an acquired lesion secondary to trauma, surgical intervention, or neoplasia [1, 16]. These terms have been used interchangeably to describe intrahepatic arteriovenous communications. However, in small animals, almost all reports of intrahepatic arterioportal communications describe a nidus, and most affected animals are young [16, 18, 19]. Therefore, the term HAVM is more appropriate in these cases.

In the present case, left divisional liver lobectomy followed by ligation of accessible APSSs achieved restoration of hepatopetal portal perfusion and normalization of liver function parameters, which persisted throughout a follow-up period of >2 years. Clinically, the cat gained weight and showed increased activity levels, accompanied by resolution of ascites and increased hepatic volume postoperatively. Stunted growth, lethargy, and abdominal effusion are frequently observed in dogs and cats with HAVM, as well as anorexia, vomiting, and neurologic signs [3, 8, 9, 12, 18]. On the other hand, hematuria has never been described in these patients, and in the present case, hematuria improved with preoperative medical therapy, so the causal relationship between hematuria and HAVM was unclear. Chanoit et al. proposed a classification of surgical/interventional outcomes in dogs with HAVM as follows [3]; excellent (documented normal liver function and resolution of PSS), good (no clinical signs and no need for medical therapy), fair (no clinical signs but requiring some medication), poor (clinical signs refractory to treatment or death/euthanasia attributable to HAVM). According to this classification, the current case would be categorized as “excellent.” A case report published in Japanese described a dog with HAVM that underwent hepatic lobectomy combined with staged ligation of APSSs, similar to the present case, and reported normalization of liver function tests [6]. In the English-language literature, this approach has also been attempted in a dog; however, the animal died immediately postoperatively from hemorrhage [13]. To our knowledge, no reports in the English-language literature have documented normalization of liver function in dogs or cats with HAVM. It has been speculated that incomplete attenuation or recanalization of arterioportal communications, persistent portal hypertension, continued shunt flow through APSSs, and hepatic insufficiency are reasons that prevent restoration of portal perfusion and normalization of hepatic function [3].

Complete attenuation of arterioportal communications is of paramount importance in treating HAVM. Indeed, incomplete attenuation of arterioportal communications may result in persistent clinical signs and necessitate reoperation, as reported in dogs with HAVM [3]. In the present case, complete attenuation was achieved by hepatic lobectomy with the aid of preoperative CTA and intraoperative portal pressure and waveform monitoring. HAVM is usually suspected based on ultrasonographic findings such as ascites and abnormal pulsating vessels within the liver; however, ultrasonography is insufficient for detailed evaluation of abnormal vessels. In previous reports, 2 cats underwent laparotomy for suspected HAVM based on ultrasonography or nonselective angiography. However, arterioportal communications could not be assessed in detail nor attenuated, and both cats died or were euthanized perioperatively [8, 9]. CTA allows a more comprehensive assessment of vascular anatomy; a retrospective study of 33 dogs and 3 cats with arterioportal communications reported that outflow portal vein branches were identifiable in all cases [16]. On the other hand, clear identification of the inflow artery was possible in only 13 of 32 cases (40.6%) because the presence of a nidus interfered with arterial evaluation [16], as in the present case. Therefore, in the present case, intraoperative portal pressure and waveform monitoring were used to assess the completeness of shunt attenuation. After ligation of the left hepatic artery, portal pressure remained elevated; additional reduction was achieved through hepatic lobectomy, resulting in near-normal portal pressure and an excellent postoperative outcome. A previous report described portal pressure normalization after ligation of the afferent artery in a cat with HAVM [12]. In these cases, portal pressure was a useful indicator of attenuation of arterioportal communications. Conversely, it has been reported that portal pressure did not decrease in a cat following coil embolization of the afferent artery, nor in a dog that underwent extensive hepatic lobectomy, despite selective angiography confirming attenuation of arterioportal communications in both cases [6, 18]. Factors such as fibrosis of the liver parenchyma, portal hypoplasia, or decreased liver volume resulting from lobectomy may have impeded the reduction in portal pressure [18]; this may explain why the present case showed marginal elevation of portal pressure even after resection of the left hepatic division. Thus, portal pressure is not an absolute indicator of attenuation of arterioportal communications, and intraoperative selective arteriography should be considered to evaluate residual shunts when portal pressure does not decrease.

In the present case, the CVC was injured intraoperatively, and blood transfusion was performed. In a previous report, hemorrhage was observed in 7 of 13 dogs undergoing liver lobectomy for HAVM, and 2 of them died in the perioperative period [3]. As with other liver surgeries, intraoperative hemorrhage represents a significant complication during the surgical attenuation of hepatic arterioportal communications. In cases of HAVM, hemorrhage can be exaggerated by portal congestion and increased bleeding through the hepatic veins and CVC. Therefore, comprehensive preoperative evaluations, such as coagulation testing, blood typing, and cross-matching, along with the availability of adequate transfusion supplies, are essential preparations. Interventional radiographic embolization of the feeding artery offers an alternative approach for HAVM, characterized by its minimally invasive nature and its benefit of conserving liver volume. This technique has recently been adopted in several dogs and a cat with HAVM [2, 3, 14, 18]. However, specific complications such as embolization of vessels outside the malformation and postoperative opening of collateral arterioportal communications have been reported [3, 14]. In addition, postoperative mesenteric vein or portal vein thrombosis has been reported in both surgical and interventional procedures [2, 3]. Therefore, meticulous monitoring during the postoperative period is essential in both approaches.

The cranial mesenteric vein–CVC shunt and the left gastric vein–azygos vein shunt were left untreated in the present case; however, they were considered to have regressed spontaneously because liver function parameters had normalized and the shunts were difficult to discern on CTA 3 months after the second surgery. Although definitive confirmation of shunt closure would require direct portal venography, this procedure is not routinely feasible in clinical practice. Moreover, the conversion of splenic vein blood flow 1 month after the initial surgery substantiated the reduction in shunting volume through the splenic vein–left gonadal vein shunt, which could have led to spontaneous closure had it remained untreated. In the present case, mean portal pressure decreased to 15 mmHg after resection of the left liver division. Although this value remained mildly elevated, intrahepatic vascular resistance likely decreased markedly compared with the preoperative state, thereby allowing some portal blood flow to perfuse the liver. As a result, hepatotropic factors such as insulin, insulin-like growth factors, glucagon, and hepatocyte growth factor could reach the remnant liver [5], potentially promoting hepatic regeneration and hypertrophy. This hypothesis is supported by the marked enlargement of the right hepatic division observed on postoperative CTA. Experimental studies of liver regeneration following partial hepatectomy have demonstrated sinusoidal remodeling and expansion of the intrahepatic vascular bed [10]. These regenerative changes may have further reduced intrahepatic vascular resistance and facilitated restoration of hepatopetal portal perfusion, thereby decreasing blood flow through the APSSs and contributing to their spontaneous regression.

However, since the CVC typically maintains lower pressure than the portal vein [12], blood flow through APSSs may persist after HAVM attenuation. The pathophysiology of persistent shunting resembles that of congenital PSS; neurotoxic substances originating within the gastrointestinal tract enter the systemic circulation, and hepatotropic factors bypass the liver, thereby inhibiting its growth [5]. In previous reports, 2 dogs were euthanized 15 and 17 months after HAVM attenuation because of hepatic encephalopathy or gastrointestinal signs [3, 15]. In those cases, large-volume PSSs were identified immediately prior to euthanasia, suggesting that APSSs may have contributed to the clinical deterioration. Hepatocellular insufficiency is another possible contributing factor. In the present case and in dogs with HAVM reported in the literature, histopathological analysis of hepatic regions adjacent to the arterioportal communications showed absent or reduced portal vein profiles, proliferation of small arterioles and bile ductules, hepatocellular atrophy, and fibrosis—changes comparable to those seen with congenital PSS [11, 15]. Based on reports of congenital PSS and our case experience, improvement in liver function is possible following attenuation of APSSs despite these pathological changes [7, 17]. Therefore, ligation of residual APSSs should be considered in patients with persistent shunting and abnormal liver function tests after HAVM attenuation.

In conclusion, attenuation of the arterioportal communications by arterial branch ligation or hepatic lobectomy, guided by preoperative CTA and intraoperative portal pressure monitoring, was an effective and feasible approach, although meticulous preparation for potential intraoperative hemorrhage is essential. In cases in which liver function remains abnormal, surgical ligation of APSSs may confer further improvement.

CONFLICT OF INTEREST

The authors declare that there were no conflicts of interest.