Materials and Methods

Case selection—All dogs with IHPSS evaluated by the University of Pennsylvania Interventional Radiology service from July 1, 2001, to August 31, 2009, or the Animal Medical Center Interventional Radiology service from September 1, 2009, to January 31, 2011, were considered for inclusion in the study (these dates corresponded to the primary author's [CW] time and location of employment). Dogs were excluded if the owners declined consent for this procedure or patient follow-up was < 90 days after surgery for surviving dogs (animals dying within 90 days were included). Endovascular treatment was determined on a case-by-case basis; however, treatment delay was strongly considered if intraoperative resting portal pressures were excessive (≥ 16 mm Hg [21 cm H₂O]) or the portal venous pressure–CVP gradient was excessive (≥ 6 mm Hg [8 cm H₂O]).

Medical records review—Preoperative data collected from medical records of those dogs meeting the inclusion criteria included signalment, typical and atypical clinical signs of PSS (including a history of melena or other signs of gastrointestinal ulceration), previous surgeries, medical treatment, CBC and serum biochemical analysis results, and liver function test results. Preoperative imaging findings (ultrasonography, CT angiography, and MRA) were reviewed and recorded. Peri- and intraoperative data collected included access site, shunt description (angiographic location, number, and complexity), pre- and postembolization caudal vena cava and portal pressures (including balloon occlusion shunt pressures when available), stent type and size, number and size of coils placed, ancillary interventional devices placed, procedure time (from jugular access initiation to vascular sheath removal), peri- and intraoperative complications, hospitalization time, immediate postoperative complications (within 1 week), late complications (> 1 week after surgery), number of procedures performed, current medications at the time of last follow-up consultation, and patient outcome including survival time (date of procedure until the time of death).

Angiography and PTE—All dogs were treated medically (lactulose, antimicrobials, low-protein diet, and, for patients seen later in the study, proton-pump inhibition) initially following diagnosis of the IHPSS for a period of weeks to months. Typical recommendations for medical management duration were at least 4 weeks prior to the procedure and when, if possible, the dogs were at least 5 months of age. This medical treatment duration and age were chosen arbitrarily to provide sufficient time for the animal to grow and to reduce the chance of the patient outgrowing the implanted devices. Antiseizure medications were not routinely prescribed unless the patient had persistent neurologic signs following appropriate treatment with the previous medications listed above (lactulose, antimicrobials, low-protein diet, and, for patients seen later in the study, proton-pump inhibition). When possible, CT angiography or MRA was performed to delineate the shunt anatomy and obtain caval and shunt measurements under a separate anesthetic episode.^15,16 All PTE procedures were performed with patients under general anesthesia with a standard liver dysfunction protocol and often a neuromuscular blockade to minimize respiratory artifact during digital subtraction angiography. Cefoxitin^a (30 mg/kg [13.6 mg/lb] was administered IV once, followed by 20 mg/kg [9.1 mg/lb]) IV every 2 hours during the procedure. Some variation in the procedure occurred over the approximate 10-year period of treating these patients, but the basic procedure as described was performed with slight variation of that previously described (Figure 1).^17,18,b

Serial ventrodorsal images acquired during a left IHPSS PTE procedure in a 6-month-old sexually intact female Labrador Retriever. The patient's head is to the left in each image. A—Digital subtraction portogram and cavagram demonstrating PSS entrance (asterisk) at the level of the caudal vena cava (CVC). B—Radiograph of stent delivery system with constrained stent prior to deployment. C—Radiograph immediately following stent deployment. Note stent bulge into the shunt entrance at the level of the left hepatic vein. D—Digital subtraction venogram of the shunt demonstrating placement of the stent across the entire shunt entrance. E—Digital subtraction venogram following placement of thrombogenic coils at the shunt and caudal vena cava interface. F—Radiograph obtained at completion of the PTE procedure.

Citation: Journal of the American Veterinary Medical Association 244, 1; 10.2460/javma.244.1.78

• Download Figure
• Download figure as PowerPoint slide

View Full Size

Serial ventrodorsal images acquired during a left IHPSS PTE procedure in a 6-month-old sexually intact female Labrador Retriever. The patient's head is to the left in each image. A—Digital subtraction portogram and cavagram demonstrating PSS entrance (asterisk) at the level of the caudal vena cava (CVC). B—Radiograph of stent delivery system with constrained stent prior to deployment. C—Radiograph immediately following stent deployment. Note stent bulge into the shunt entrance at the level of the left hepatic vein. D—Digital subtraction venogram of the shunt demonstrating placement of the stent across the entire shunt entrance. E—Digital subtraction venogram following placement of thrombogenic coils at the shunt and caudal vena cava interface. F—Radiograph obtained at completion of the PTE procedure.

Citation: Journal of the American Veterinary Medical Association 244, 1; 10.2460/javma.244.1.78

• Download Figure
• Download figure as PowerPoint slide

Serial ventrodorsal images acquired during a left IHPSS PTE procedure in a 6-month-old sexually intact female Labrador Retriever. The patient's head is to the left in each image. A—Digital subtraction portogram and cavagram demonstrating PSS entrance (asterisk) at the level of the caudal vena cava (CVC). B—Radiograph of stent delivery system with constrained stent prior to deployment. C—Radiograph immediately following stent deployment. Note stent bulge into the shunt entrance at the level of the left hepatic vein. D—Digital subtraction venogram of the shunt demonstrating placement of the stent across the entire shunt entrance. E—Digital subtraction venogram following placement of thrombogenic coils at the shunt and caudal vena cava interface. F—Radiograph obtained at completion of the PTE procedure.

Citation: Journal of the American Veterinary Medical Association 244, 1; 10.2460/javma.244.1.78

• Download Figure
• Download figure as PowerPoint slide

Patients were placed in dorsal recumbency, with the neck extended and a radiopaque stent guide^c placed between the patient and the tabletop. The ventral cervical region was clipped of hair, scrubbed, and draped, exposing the right jugular vein (preferred instead of the left jugular vein because of the relatively straight anatomic access to the vena cava). All guidewire, catheter, stent, and coil manipulations were performed by means of standard sterile technique under fluoroscopic guidance. A 3- to 5-mm skin incision facilitated percutaneous placement of either a 10F or 12F vascular introducer sheath,^d depending on the anticipated caval stent delivery system size.

For left or right divisional shunts, the patient remained in dorsal recumbency. For central divisional shunts, the animal was repositioned in lateral recumbency (or the c-arm was rotated for lateral imaging) to facilitate angiographic identification of the shunt entrance into the caudal vena cava. A 4F catheter^e and 45° angle hydrophilic guidewire^f combination were used under fluoroscopic guidance to selectively access the shunt and portal vein.

During angiography, the contrast agent^g was typically diluted 1:1 with sterile saline (0.9% NaCl) solution to reduce total contrast load, maintained at a standard dose of < 3 mL of contrast/kg of body weight for the entire procedure. Digital subtraction portal venography was then performed to confirm appropriate catheter placement, delineate shunt anatomy, and identify the presence or absence of portal perfusion (arborization of a branch of the portal vein). If considerable portal perfusion was present (second-generation portal vessels or more), balloon occlusion was often performed to assess whether complete shunt attenuation would be tolerated. If so, alternate complete shunt embolization techniques were considered that would provide complete acute occlusion, such as vascular plug placement, caval stent graft placement, or complete coil occlusion. Otherwise, a 5F marker catheter^h and 45° angle hydrophilic guidewire^f combination were advanced into the sheath beside the 4F catheter^e and into the caudal vena cava. A digital subtraction caudal vena cavagram was performed under 20 cm H₂O PPV to compress the thoracic caudal vena cava and therefore maximally distend the abdominal vena cava. The maximal abdominal vena cava diameter was then determined by calibrating the measurement software to the known distance between radiopaque marks on the marker catheter, thus minimizing error that occurs secondary to magnification. These measurements were considered in relation to the cross-sectional measurements of the vena cava made on any prior CT or MRA scan (not performed under PPV). Next, a combination digital subtraction portogram and caudal vena cavagram was performed without PPV to identify the precise location and length of the confluence of the shunt and the caudal vena cava as well as to determine the maximal intrathoracic caudal vena cava diameter (non-PPV). Caval diameters and shunt opening lengths were used to choose an appropriately sized SEMS: typically, at least 10% to 25% greater in diameter than the maximal caval diameters measured and at least 20 mm longer cranially and caudally to the shunt confluence with the vena cava when possible. Portal pressures were recorded through the end-hole 4F catheter,^f and CVP measurements were recorded through the marker catheter before stent placement. If resting portal pressures were > 16 mm Hg (21 cm H₂O) or the pressure gradient between the portal vein and CVP were > 6 mm Hg (8 cm H₂O), the embolization procedure was reconsidered and repeated angiography was recommended in approximately 3 months. In certain patients, the owners declined additional procedures and the embolization was performed in light of the potential risks (eg, acute portal hypertension with possible death, chronic portal hypertension, and possible acquired shunt development).

To proceed, the 4F catheter^e was removed from the portal vein, an exchange-length floppy-tipped polytetrafluoroethylene guidewireⁱ was advanced down the marker catheter, and the marker catheter was removed over the wire. The stent delivery system^j,k (containing the stent) was advanced over the polytetrafluoroethylene wire and deployed under fluoroscopic guidance. The stent was positioned in relation to the underlying stent guide^c such that the junction of the PSS and caudal vena cava was spanned by the stent and such that the stent did not extend into the right atrium. A 4F catheter^e and 45° angle hydrophilic guidewire^f combination were used again to select the portal vein or shunt through the stent interstices, and a repeated digital subtraction venogram was performed to confirm stent placement across the entire shunt orifice. Repeated pressure measurements were obtained to determine whether the stent placement alone had raised the pressure within the portal vein or the portal venous pressure–CVP gradient. Another 4F catheterⁱ and 45° angle hydrophilic guidewire^f combination was used to select the shunt at the level of its communication with the caudal vena cava but on the portal side of the stent. At this location, thrombogenic stainless steel coils^l (usually 0.035 inches × 5 cm length × 8 mm in diameter) were advanced through the second catheterⁱ and deployed into the shunt; the presence of the caval stent prevented migration of the coils into the caudal vena cava. Coils were subsequently added during portal pressure measurements to avoid creating portal hypertension. Although the guidelines changed throughout the period of the study, coils were typically added until the shunt mouth was covered with coils (initial cases), the shunt pressures had increased approximately 6 to 7 mm Hg, or maximal pressures approached 16 mm Hg (most cases). Repeated angiography was performed at the completion of the procedure, ultimate shunt and caval pressures were recorded, the jugular sheath was exchanged for a 7F multilumen catheter,^m and the patient was allowed to recover from anesthesia.

The same guidelines for portal pressure and portal venous pressure–CVP gradient were used when performing subsequent procedures in those dogs requiring additional shunt attenuation after the initial procedure. In some cases, dramatic improvement in portal perfusion permitted complete (or near complete) shunt attenuation without reaching the limits of the pressure guidelines.

Patients recovered in the intensive care unit on IV fluid therapy, lactulose, antimicrobials, low-protein diet, and (during or after approx 2005) proton-pump inhibitors. Analgesics were not routinely necessary and used at the discretion of the attending intensive care unit clinicians. Patients were typically discharged 2 days after surgery, and owners were instructed to continue all previous medications for 1 month. Patients were also discharged with 10 to 14 days of an additional broad-spectrum antimicrobial. Medications were slowly tapered (except the proton-pump inhibitor), and the dog transitioned to a regular diet over the next 2 to 4 weeks. If clinical signs did not return following discontinuation of medications and the special diet, no further treatment was recommended. Repeated serum biochemical analysis and CBC were again recommended at 3 months after the procedure and then encouraged every 6 to 12 months for the first year and yearly thereafter. If clinical signs returned, medications were reinstituted and an additional angiography procedure was recommended.

Follow-up evaluation—Follow-up information was collected by a follow-up examination or telephone contact with the referring veterinarian or owner. Complications were defined as minor (non–life-threatening) or major (life-threatening) and intraoperative (during procedure), early (within 1 week after surgery), or late (> 1 week after surgery). On the basis of the presence of clinical signs, current medications (if present), and serum biochemical analysis results, the post-PTE outcome was defined as excellent (absence of clinical signs without low-protein diet or medication), fair (absence of clinical signs with low-protein diet or medications), or poor (continued or worsening clinical signs with special diet and medications, no response to surgery, or surgical-related death). Serum biochemistry results were used to classify subtle, nonspecific clinical signs as either likely PSS associated (typical PSS biochemical changes) or likely PSS unassociated (absence of standard PSS biochemical parameters). For outcome determination, proton-pump inhibitors were not considered as PSS medication. Cause of death categories were defined as death confirmed due to IHPSS or PTE by means of postmortem examination, death suspected due to IHPSS or PTE, death unlikely due to IHPSS or PTE, or death confirmed not due to IHPSS or PTE by postmortem examination or medical records review.

Statistical analysis—Descriptive statistics of the entire study population were calculated. Categorical variables were summarized by frequencies and percentages. Continuous variables (eg, pre- and post-PTE biochemical and CBC variables) were summarized as mean ± SD for normally distributed data and median and interquartile range (25th to 75th percentile) for data that did not meet the normality of residuals assumption on the basis of the Shapiro-Wilk test. The 2-sample t test or Mann-Whitney test (the nonparametric counterpart to the 2-sample t test), as deemed appropriate, was used to compare 2 groups (single vs multiple shunt complexity or the first half of cases vs the second half of cases) for continuous data. The χ² test or Fisher exact test, as deemed appropriate, was used to compare groups for categorical data.

Patient outcome from the procedure was defined as excellent, fair, poor, or unknown. The comparison of dogs with an excellent outcome versus all other outcomes was analyzed for numerous pre- and post-PTE biochemical and CBC variables. For simplicity, all continuous data were reported as median (interquartile range [25th to 75th percentile]) and analyzed with the Mann-Whitney test. Categorical data were analyzed appropriately as described.

Those factors that appeared to be associated with excellent outcome in the univariate analysis (P < 0.10) were included in a logistic regression model. Backward selection was used to remove variables that did not significantly contribute information to the model, given other factors included in the model. Given the limited number of other outcome events, only models with ≤ 3 variables were considered in the final analysis. The Hosmer-Lemeshow test was used to examine the models' goodness of fit. Finally, a receiver operating characteristic curve was constructed to evaluate the ability to predict an excellent outcome. The area under the receiver operating characteristic curve was used as a measure of the model's discriminatory ability. The following guidelines were used to describe the quantitative-qualitative relationship between the area under curve and the accuracy in predicting excellent outcome: 0.9 to 1.0 = excellent, 0.8 to 0.9 = very good, 0.7 to 0.8 = good, 0.6 to 0.7 = fair, and 0.5 to 0.6 = poor. The analysis of time to death was accomplished by applying standard methods of survival analysis (ie, computing the Kaplan-Meier product limit curves overall and stratified by outcome [excellent outcome vs all other outcomes] and other groups [ie, sex, access site, shunt angiographic anatomy, shunt complexity, preoperative gastrointestinal bleeding, and postoperative gastrointestinal bleeding]). In cases where the endpoint event, death, had not yet occurred, the number of days from the procedure until last follow-up was used and considered censored. These 2 groups were compared by use of the log rank test. The median survival time was obtained from the Kaplan-Meier and product-limit estimates, and their corresponding 95% CIs were computed with the Greenwood formula to calculate the SE. Unless otherwise specified, values of P < 0.05 were considered significant.

Results

Study population—Examination of the medical records identified 104 canine patients that were examined at the Matthew J. Ryan Veterinary Hospital of the University of Pennsylvania or the Animal Medical Center interventional radiology services with confirmed IHPSSs during the study period. Three owners declined treatment at the time of examination, and 1 declined the investigational PTE procedure in favor of traditional surgery. The other 100 dogs satisfied inclusion criteria and received endovascular procedures.

Thirty-five dog breeds were represented in the study, with 53 male dogs and 47 female dogs (Table 1). The age of dogs ranged from 2.5 to 51 months (mean ± SD, 10.6 ± 8.6 months). Body weights ranged from 2.6 to 51.6 kg (5.7 to 113.5 lb; mean ± SD, 18.5 ± 8.9 kg [40.7 ± 19.6 lb]).

Preoperative findings—All 100 patients were being treated with standard PSS medications prior to the procedure. Prior to medical management, 1 dog had repeated ammonium biurate urolithiasis as the only clinical sign and the remaining 99 dogs had typical PSS clinical signs that differed in severity, including hepatoencephalopathy (wall-walking, stargazing, ataxia, seizures, or lethargy) and general signs of failure to thrive prior to medical management. Five dogs were subjectively determined to have abnormal neurologic signs at the time of the procedure, despite medical management. Historically, clinical signs that would not typically be associated with PSSs in dogs in our clinical experience or as reported previously ^2,4,7,8,12 (ie, melena [n = 13], ascites [5], hematochezia [1], blood in emesis [1]) were identified preoperatively in 17 dogs. In 1 dog, the presence of ascites, hematochezia, or blood in vomit was unclear from the record. Previous surgical attenuation or attempted attenuation had been performed in 9 dogs previously. In 1 dog, previous surgical information was unclear. Preoperative serum biochemical and CBC abnormalities were summarized (Table 2).

Angiography and PTE procedure—Percutaneous vascular access site was recorded in 96 dogs; access was performed via the right jugular vein alone (n = 92), the right jugular vein and percutaneous portal vein access (2), the left jugular vein (1), or the right jugular vein and right femoral vein (1). Angiography revealed single IHPSSs in 90 dogs distributed as primarily right divisional shunts (n = 38), left divisional shunts (33), or central divisional shunts (19). The remaining 10 dogs had either 2 IHPSSs (n = 2), an IHPSS and EHPSS (1), or > 2 IHPSSs (7). The 2 dogs with 2 separate IHPSS included a right and central divisional shunt and a left and central divisional shunt. One dog had a left divisional intrahepatic shunt and a left gastric to phrenic vein extrahepatic shunt. The remaining 7 dogs with > 2 IHPSSs had primarily right divisional (n = 4), left divisional (2), or central divisional (1) shunts.

Resting CVPs ranged from 0 to 16 mm Hg (mean 7.0 ± 3.4 mm Hg; Table 3). Resting portal pressures ranged from 0 to 20 mm Hg (mean ± SD, 8.7 ± 3.8 mm Hg) and were > 16 mm Hg in 2 (2%) dogs, both with multiple IHPSSs. The resting portal venous pressure–CVP gradient recorded in 87 dogs ranged from −2 to 10 mm Hg (mean ± SD, 4.9 ± 3.9 mm Hg) and was ≥ 6 mm Hg in 7 (8%) dogs, 4 of which had multiple IHPSSs. One dog had both excessive resting portal pressures and resting portal venous pressure–CVP gradient (this dog had multiple IHPSSs). Of the 8 dogs with excessive resting pressures outside of the PTE guidelines, 5 dogs ultimately did not receive PTE, and these dogs are addressed separately. Resting portal pressure and CVP measurements were available in 8 of the 10 dogs with multiple IHPSSs. The mean ± SD resting portal venous pressure–CVP gradient in these 8 dogs (4.9 ± 3.9 mm Hg) was significantly (P < 0.001) higher than the mean resting portal venous pressure–CVP gradient in the remaining 79 dogs with a single IHPSS (1.5 ± 2.0 mm Hg).

Aside from the 5 dogs not receiving PTE, the remaining 95 dogs ultimately had endovascular shunt attenuation performed. Nine of these dogs had angiography performed alone initially, with the PTE procedure performed at a later date for various reasons. Five of these 9 dogs required specialized stents or equipment not routinely available at the time (stent-graft, atrial septal defect occluder, custom or alternate size stent, or alternative procedure planning). One dog had an excessive portal venous pressure–CVP gradient with multiple IHPSSs; the procedure was repeated 6 months later, and PTE was performed. One dog had an excessive portal venous pressure–CVP gradient; the procedure was repeated 1 month later and PTE performed. One had a small portal venous pressure–CVP gradient of 4 mm Hg and narrow shunt opening; concern over complete occlusion with a single coil prompted a repeated procedure 4 months later, at which time a marginally lower portal venous pressure–CVP gradient of 3 mm Hg prompted the owner to permit PTE. One dog had confusing shunt anatomy, compared with the preoperative MRA resulting in increased contrast use necessitating a follow-up procedure to complete the PTE.

A total of 111 PTE procedures were performed in 95 dogs over the study period. Thirteen (14%) dogs had 2 procedures performed, and 2 (2%) dogs had 3 procedures performed. Three dogs were confirmed angiographically or with shunt occlusion pressure measurements to tolerate complete shunt occlusion; these patients received either a covered caval stent-graft,ⁿ an atrial septal defect occluder,^19,o or a vascular plug^p to achieve successful acute, complete shunt occlusion by means of similar interventional radiology techniques. One of these dogs needed a repeated procedure because the dog grew and the atrial septal defect occluder^o device became too small to maintain complete occlusion.

The remaining 92 dogs had partial endovascular shunt attenuation performed initially with the standard technique described. Stent type and sizes were recorded in 84 dogs; 78 (93%) dogs received a single stent, and 6 (7%) dogs received 2 stents. Prior to 2004, laser-cut SEMSs were not routinely available in the necessary sizes, and the first 16 dogs were treated in this period. Of these initial 16 dogs treated with the mesh stent^j and coil technique, 8 had the stent size and type recorded; 5 (63%) received a single stent, and 3 (38%) received 2 mesh SEMSs. Subsequently, 78 dogs had procedures when laser-cut nitinol SEMS^k were available; stent size and type were recorded in 76 dogs, of which 72 (95%) received a single stent and 4 (5%) received 2 stents. Single laser-cut stent diameters ranged from 10 to 26 mm (mean, 20.0 ± 3.7 mm), and lengths ranged from 35 to 80 mm (mean, 71.9 ± 13.7 mm). The 3 most common sizes used were 22 × 80 mm (n = 18), 20 × 80 mm (11), and 24 × 80 mm (9). One dog with an extremely large vena cava received a 33 × 88-mm custom mesh stent.

Two of 92 dogs had pressures or gradients rise sufficiently following placement of the caval stent alone to preclude the placement of any coils: 6.5 mm Hg (8.7 cm H₂O) portal venous pressure–CVP gradient in one dog and 15 mm Hg (20 cm H₂O) portal pressure in the other dog.

The remaining 90 dogs had thrombogenic coils placed in addition to the caval stent, of which 75 dogs had a single PTE procedure. The most commonly used size of thrombogenic coil was 8 mm in diameter × 5 cm long; between 1 and 30 of these specific size coils were placed for a single PTE in 53 dogs, and the remaining 22 dogs received a variety of coil sizes. The most common number of coils placed was 4 coils (n = 13), 5 coils (10), or 6 coils (8). There was a significant (P < 0.001) difference in the median number of coils placed in the first half of initial PTE procedures (median, 3.5 coils; range, 0 to 6), compared with the second half of initial PTE procedures (median, 8.0 coils; range, 0 to 30).

Following the initial PTE procedures, post-PTE portal pressures ranged from 1 to 22 mm Hg (mean, 11.3 ± 4.6 mm Hg), post-PTE change in portal pressures ranged from −2 to 12 mm Hg (median, 2.0 mm Hg), and post-PTE portal venous pressure–CVP gradient ranged from −2 to 17 mm Hg (median, 4.0 mm Hg; Table 3). Interestingly, there was a significant difference between the first half and the second half of cases for median post-PTE portal pressure increase (1.0 mm Hg [range, −2 to 7.5 mm Hg] vs 4.5 mm Hg [range, −2 to 12 mm Hg], respectively; P < 0.001) and median post-PTE portal venous pressure–CVP gradient increase (2.0 mm Hg [range, −2 to 16 mm Hg] vs 5.0 mm Hg [range, 0.5 to 17 mm Hg], respectively; P = 0.006), with the second half of cases increasing for both. For second and third PTE procedures, no pressure gradients precluded additional treatment.

Initial PTE procedure length recorded for 75 cases ranged from 30 to 340 minutes (median, 90 minutes). There was no significant (P = 0.454) difference in median length of procedure between the first and second half of the initial PTE procedures (92.5 minutes [range, 38 to 340 minutes] vs 90 minutes [range, 45 to 135 minutes], respectively). A median second PTE procedure duration of 70 minutes was recorded in 8 cases, which ranged from 20 to 240 minutes. A third PTE procedure duration was recorded in 1 case and was 195 minutes.

Intraoperative complications—Major or minor intraoperative complications occurred in 15 of the 111 (14%) procedures. Major intraoperative complications occurred in 3 of 100 (3%) dogs, including substantial (temporary) portal hypertension > 30 mm Hg that resolved following balloon- or catheter-assisted coil displacement (2/100 [2%]) and severe acute gastrointestinal hemorrhage with subsequent owner-elected euthanasia (1/100 [1%]). The latter patient represented the only intraoperative death (< 1% [1/111 procedures]), and this euthanasia event occurred prior to performing the PTE procedure. Minor non–life-threatening intraoperative complications occurred during 14 of 111 (13%) procedures in 12 of 100 (12%) dogs, including stent misplacement requiring placement of another stent (5/111 [5%] procedures), with subsequent coil launch (3/111 [3%]), chemosis and subjective hyperemic mucus membranes due to presumptive contrast reaction (5/111 [5%]), and temporary severe hypotension (1/111 [0.09%]).

Early postoperative results and complications—A median hospitalization time of 2 days following initial PTE procedure was recorded in 82 cases and ranged from 1 to 13 days. Early postoperative (within 1 week after PTE) complications occurred 21 times in 16 dogs following 111 PTE treatments (21/111 [19%] treatments). Fourteen major life-threatening early postoperative complications in 12 dogs (14/111 [13%] treatments) included seizures and hepatoencephalopathy (7/111 [6%]), cardiac arrest (2/111 [2%]), hemorrhage from jugular access site requiring transfusion (2/111 [2%]), pneumonia (1/111 [1%]), suspected portal hypertension (1/111 [1%]), and acute death of unknown cause (1/111 [1%]). Four of these dogs (4/111 [4%]) died or were euthanized within the first week following the procedure because of cardiac arrest, intractable seizures, acute death of unknown cause, and seizures and pneumonia. Seven minor non–life-threatening early postoperative complications in 5 dogs (7/111 [6%] treatments) included cardiac arrhythmias (3/111 [3%]), hyperthermia (2/111 [2%]), noticeable mild abdominal distension and ascites (1/111 [1%]), and minor muscle twitching (1/111 [1%]).

Late (> 1 week) postoperative and long-term (> 90 day) results and complications—Two dogs were lost to follow-up at 118 and 127 days after PTE. For the total remaining 98 dogs (39 dead and 59 alive), follow-up times ranged from 0 to 3,411 days (median, 962 days; 95% CI, 695 and 1,216 days). Post-PTE serum biochemical and CBC values, changes in preoperative and postoperative values, and percentage changes in preoperative and postoperative values were summarized (Table 2).

Those factors significantly (P < 0.05) associated with an excellent outcome included lower preoperative globulin concentration, TS concentration, resting bile acids concentration, and postprandial bile acids concentration; higher postoperative (last follow-up) Hct, TS concentration, albumin concentration, BUN concentration, cholesterol concentration, MCV, and postprandial bile acids concentration; greater change in Hct, TS concentration, albumin concentration, globulin concentration, BUN concentration, MCV, and cholesterol concentration; and greater percentage change in Hct, TS concentration, albumin concentration, globulin concentration, BUN concentration, MCV, and cholesterol concentration (Table 4). Only the change in Hct, change in TS concentration, change in albumin concentration, change in globulin concentration, change in BUN concentration, change in MCV, change in cholesterol concentration, number of days between repeated procedures, and shunt anatomy were included in the selection process of the logistic regression model because these were clinically meaningful potential predictors associated with clinical outcome suggested by a value of P < 0.10 in the univariate analyses. Change in globulins concentration (increased value from before to after surgery; P = 0.002) and change in TS concentration (increased value from before to after surgery; P < 0.001) were the only variables that were significantly associated with excellent outcome in the final multivariate model, with P < 0.05 as the significance level. The Hosmer-Lemeshow goodness-of-fit test demonstrated that a good fit was achieved in this model (χ² = 12.08; degrees of freedom, 8; P = 0.148). The area under the curve of 0.89 indicated very good accuracy in predicting excellent outcome in these dogs. The odds of having an excellent outcome increased by a factor of 3.3 (OR, 3.34; 95% CI, 1.82 and 6.11) as a patient's change in TS concentration increased by 0.3 g/dL (Table 5).

Eighty-eight treated dogs were reevaluated for recurrence of any clinical signs such as seizure activity, twitching, or possible vague hepatoencephalopathy-related episodes. Fifty-nine of these 88 (67%) dogs had no clinical signs, and 29 of 88 (33%) ultimately developed ≥ 1 of these signs during follow-up evaluation. Five of these dogs had additional PTE procedures that resolved the clinical signs. For the remaining 24 of 88 (27%) dogs, the owners declined additional procedures and the patients either improved with or without medical management or ultimately deteriorated. Four of the 5 dogs that developed neurologic clinical signs during medical treatment prior to PTE had recurrence of these clinical signs after PTE. It was not possible to retrospectively confirm how many of these clinical signs were truly secondary to hepatoencephalopathy.

Evaluation for the development of any clinical signs associated with possible gastrointestinal ulceration such as melena, hematochezia, consistent blood work changes, or confirmed gastrointestinal ulcer or perforation was available for 90 treated dogs. Nineteen (19/90 [21%]) dogs had signs consistent with gastrointestinal ulceration during the follow-up period. Three (3/19 [16%]) of these dogs had similar clinical signs prior to the PTE procedure.

Evaluation for the development of other possibly related or unrelated conditions included cancer, unknown illness, renal disease, aggression, trauma (hit by car), tick-borne disease, pneumonia, cardiac disease, shifting leg lameness of unknown origin, and possible diskospondylitis.

Nutritional and medical management information was available for 85 dogs at least 1 month following PTE and was summarized (Table 6), along with long-term IHPSS-related clinical sign manifestation and recurrence information, which was available for 81 dogs. Ultimate long-term IHPSS PTE outcome was determined for 86 dogs in which sufficient follow-up information was available. On the basis of the presence of clinical signs, current medications (if present), and serum biochemical results, PTE outcome was perceived by the authors to be excellent (66%), fair (15%), or poor (19%).

Survival times for the 95 dogs receiving PTE ranged from 0 to 3,411 days (median, 2,204 days; 95% CI, 1,295 to 2,461 days; Figure 2). On the basis of the log rank test, there was a significant (P < 0.001) difference between dogs with excellent outcome versus dogs with other outcomes with respect to time to death. The Kaplan-Meier estimate of the median time to death in the dogs with excellent outcome was higher than that of dogs with other outcomes (2,435 vs 649 days, respectively). There were no significant differences in survival time with respect to sex (P = 0.869), access site (P = 0.387), shunt anatomy (P = 0.073), shunt complexity (P = 0.811), or preoperative gastrointestinal bleeding (P = 0.392). However, dogs with postoperative gastrointestinal bleeding had a significantly (P = 0.003) shorter survival time, compared with dogs without postoperative gastrointestinal bleeding (929 vs 2,435 days, respectively). Cause of death was evaluated in the 36 dogs that had died and included death confirmed due to IHPSS or PTE (17%), death suspected due to IHPSS or PTE (25%), death unlikely due to IHPSS or PTE (28%), or death confirmed not due to IHPSS or PTE (31%). Non-IHPSS or PTE causes of death included gastrointestinal ulceration or perforation, other unknown disease processes not associated with liver disease, renal disease, cancer, aggression with owner-elected euthanasia, trauma (hit by car), tick-borne disease, pneumonia, and cardiac disease.

Kaplan-Meier survival curves of overall survival time in 95 dogs receiving PTE for confirmed IHPSS (median survival time, 2,204 days; A) and for dogs with an excellent outcome (median survival time, 2,435 days) versus all other outcomes (median survival time, 649 days; B).

Citation: Journal of the American Veterinary Medical Association 244, 1; 10.2460/javma.244.1.78

• Download Figure
• Download figure as PowerPoint slide

View Full Size

Kaplan-Meier survival curves of overall survival time in 95 dogs receiving PTE for confirmed IHPSS (median survival time, 2,204 days; A) and for dogs with an excellent outcome (median survival time, 2,435 days) versus all other outcomes (median survival time, 649 days; B).

Citation: Journal of the American Veterinary Medical Association 244, 1; 10.2460/javma.244.1.78

• Download Figure
• Download figure as PowerPoint slide

Kaplan-Meier survival curves of overall survival time in 95 dogs receiving PTE for confirmed IHPSS (median survival time, 2,204 days; A) and for dogs with an excellent outcome (median survival time, 2,435 days) versus all other outcomes (median survival time, 649 days; B).

Citation: Journal of the American Veterinary Medical Association 244, 1; 10.2460/javma.244.1.78

• Download Figure
• Download figure as PowerPoint slide

Patients with excessive pressures not receiving procedures—Five dogs did not satisfy criteria to receive the PTE procedure because of excessive resting portal pressures or portal venous pressure–CVP gradient. Four of these dogs had complex (multiple) shunts present: 2 were primarily left-sided shunts and 2 were primarily right-sided shunts; the fifth dog had a single right-sided shunt. Excessive resting portal pressures present in 3 dogs were 16, 17.5, and 20 mm Hg. Excessive resting portal venous pressure–CVP gradients present in 4 dogs ranged from 6 to 10 mm Hg. Two dogs had both elevated resting portal pressures and an excessive portal venous pressure–CVP gradient. Only 1 dog had a repeated angiography procedure approximately 3 months later, in which the PTE procedure was not performed owing to the presence of the multiple IHPSS. Those pressures were not recorded, but the values had reportedly diminished over time according to the discharge instructions. One dog was lost to follow-up; of the remaining 4 dogs, 1 died 49 days later and 3 remained alive on medical and nutritional management.

Discussion

In the present study, in which 95 dogs received endovascular treatment for IHPSSs over a 10-year period (2001 to 2011), results indicated that this minimally invasive treatment may result in lower morbidity and mortality rates, with similar patient outcomes, compared with previously reported open surgical procedures. To our knowledge, this is the largest single collection of nearly consecutive dogs with IHPSS reported to date. The age, breed, sex, clinical signs, and biochemical abnormalities described here are not dissimilar to those reported in previous studies.^2,4,7,8,12 The high incidence of pre- and postprocedural gastrointestinal hemorrhage or ulceration (approx 15% to 21%) in this population has not been previously reported. Following this discovery approximately in 2005, proton-pump inhibitors were prescribed that, in our estimation, dramatically reduced the incidence of this complication. Evaluation of the efficacy of this and other risk factors was beyond the scope of this study. We chose to include consecutive cases to provide information about the natural distribution, spectrum, and variety of intrahepatic portovenous anomalies, some of which differ from those previously reported.

Most dogs in the present study had a large, single IHPSS with a minor to absent portal venous pressure–CVP gradient and little to no portal perfusion apparent on preoperative imaging. Preoperative cross-sectional imaging is currently recommended in all patients when possible because it was found to be helpful in identifying the shunt anatomy preoperatively and enabling cross-sectional caudal vena cava measurements for stent diameter determination to be obtained. This allows procedure time to be reduced. Due to the contrast load necessary for dual-phase CT angiography, the PTE procedure is routinely performed under a different anesthetic episode, often the following day. With diluted contrast agents, multidetector CT technology, or MRA, it is likely that these could be combined into a single procedure if necessary.

The distribution of single left (n = 33), right (38), and central (19) divisional shunts in the patients of the present study is similar to that reported previously,^2,5,9 with fewer centrally located shunts reported. It is possible that fewer of these cases are referred for treatment because of perceived worse prognosis. There was no significant difference in outcome or survival time on the basis of shunt location in this population. However, 80% of the patients with central divisional shunts had an excellent outcome, compared with 67% for patients with right shunts and 48% for those with left shunts. One possible explanation could be that the development of intrahepatic collateral vessels prevents excellent outcomes and these collaterals may form more readily when more cranially located hepatic veins are occluded (as the left hepatic veins are most cranially located and the central [middle] hepatic veins are located most caudad). The number of complex (multiple) shunts (10%) identified was higher than anticipated from previous reports, although multiple intrahepatic shunts have been reported.²⁰ Multiple (complex) IHPSSs may actually remain underestimated in these cases because subtle additional shunts now identified through experience may have been missed in the early phases of the study. The overall high number of multiple shunts may be due in part to improved preoperative cross-sectional imaging; however, direct angiography often demonstrated more shunts in these cases than expected even with the use of multislice CT. Direct angiography offers higher local concentrations of contrast material injected under pressure. In addition, following shunt attenuation and increased portal pressures, previously unidentified portal branches are more clearly demonstrable (Figure 3).

Lateral digital subtraction venographic images of 2 central IHPSSs before and after PTE. A—Images obtained in a 6-month-old sexually intact male Nova Scotia Duck Tolling Retriever prior to shunt attenuation demonstrating minimal portal perfusion (top; white arrow) and following vena cava stent (black arrows) placement only (bottom). Notice the improved portal vein perfusion (white arrows). No coils were necessary in this patient. B—Images obtained in a 7-month-old spayed female mixed-breed dog demonstrating minimal portal perfusion (white arrow) prior to PTE (top) and following PTE with stent and coils (black arrows) demonstrating immediately enhanced portal perfusion (bottom; white arrows).

Citation: Journal of the American Veterinary Medical Association 244, 1; 10.2460/javma.244.1.78

• Download Figure
• Download figure as PowerPoint slide

View Full Size

Lateral digital subtraction venographic images of 2 central IHPSSs before and after PTE. A—Images obtained in a 6-month-old sexually intact male Nova Scotia Duck Tolling Retriever prior to shunt attenuation demonstrating minimal portal perfusion (top; white arrow) and following vena cava stent (black arrows) placement only (bottom). Notice the improved portal vein perfusion (white arrows). No coils were necessary in this patient. B—Images obtained in a 7-month-old spayed female mixed-breed dog demonstrating minimal portal perfusion (white arrow) prior to PTE (top) and following PTE with stent and coils (black arrows) demonstrating immediately enhanced portal perfusion (bottom; white arrows).

Citation: Journal of the American Veterinary Medical Association 244, 1; 10.2460/javma.244.1.78

• Download Figure
• Download figure as PowerPoint slide

Lateral digital subtraction venographic images of 2 central IHPSSs before and after PTE. A—Images obtained in a 6-month-old sexually intact male Nova Scotia Duck Tolling Retriever prior to shunt attenuation demonstrating minimal portal perfusion (top; white arrow) and following vena cava stent (black arrows) placement only (bottom). Notice the improved portal vein perfusion (white arrows). No coils were necessary in this patient. B—Images obtained in a 7-month-old spayed female mixed-breed dog demonstrating minimal portal perfusion (white arrow) prior to PTE (top) and following PTE with stent and coils (black arrows) demonstrating immediately enhanced portal perfusion (bottom; white arrows).

Citation: Journal of the American Veterinary Medical Association 244, 1; 10.2460/javma.244.1.78

• Download Figure
• Download figure as PowerPoint slide

Clinically normal dogs typically have a portal venous pressure–CVP gradient, and affected dogs have an IHPSS that typically decompresses the portal system, resulting in similar portal and systemic venous pressures prior to shunt attenuation. Most IHPSS shunts have large openings into the vena cava, so a low (or zero) resting portal venous pressure–CVP gradient would be anticipated. Along with the shunt itself, the reduced portal pressure results in even more diminished perfusion to whatever portal vessels may be present naturally. The goal of shunt attenuation is to reestablish a portal venous pressure–CVP gradient to encourage portal perfusion without resulting in excessive portal pressures and subsequent dangerous portal hypertension.

Interestingly, the dogs with multiple intrahepatic shunts in this study tended to have elevated portal venous pressure–CVP gradients (significantly higher gradients recorded than the single IHPSS dogs) precluding embolization according to the study guidelines. These guidelines were based on those currently used for EHPSS shunt attenuation and historical guidelines initially generated from EHPSS in dogs.²¹ We often appreciated more difficulty accessing these shunts because of perceived narrow openings from the shunt into the caudal vena cava; these were not always identified during preoperative cross-sectional imaging, likely as a result of the CT slice thickness. It is our impression that these narrow openings created a pressure gradient (as if the shunt was already attenuated naturally), resulting in increased portal pressures and a combination of improved portal perfusion (appreciated angiographically in some cases) as well as development of acquired intrahepatic shunts (dilated interlobar hepatic veins) into adjacent liver lobes (Figure 4). This occurs naturally following ligation of a hepatic vein²² in dogs and has been demonstrated to occur following both surgical and interventional attenuation of the left hepatic vein.^12,23 We appreciated this in some cases in which repeated angiography was performed, and in our current practice, these acquired intrahepatic shunts are being coiled when identified if pressures permit. It is unclear at this time whether new shunts simply develop subsequently.

Computed tomographic and venographic images obtained from 2 dogs with multiple or complex IHPSSs. A—Axial CT image of the liver in a young female dog demonstrating multiple intrahepatic venous collaterals prior to PTE (black arrows). B—Magnified axial CT image of the same patient in image A demonstrating very narrow shunt orifice (white arrow) at the junction of the left hepatic vein and vena cava causing pressure gradient and subsequent collaterals. C—Ventrodorsal digital subtraction portogram and cavagram in a 9-month-old castrated male Miniature Poodle demonstrating multiple intrahepatic venous collaterals (black arrows) prior to PTE. A pressure gradient was already present and the shunt was difficult to access through a small window.

Citation: Journal of the American Veterinary Medical Association 244, 1; 10.2460/javma.244.1.78

• Download Figure
• Download figure as PowerPoint slide

View Full Size

Computed tomographic and venographic images obtained from 2 dogs with multiple or complex IHPSSs. A—Axial CT image of the liver in a young female dog demonstrating multiple intrahepatic venous collaterals prior to PTE (black arrows). B—Magnified axial CT image of the same patient in image A demonstrating very narrow shunt orifice (white arrow) at the junction of the left hepatic vein and vena cava causing pressure gradient and subsequent collaterals. C—Ventrodorsal digital subtraction portogram and cavagram in a 9-month-old castrated male Miniature Poodle demonstrating multiple intrahepatic venous collaterals (black arrows) prior to PTE. A pressure gradient was already present and the shunt was difficult to access through a small window.

Citation: Journal of the American Veterinary Medical Association 244, 1; 10.2460/javma.244.1.78

• Download Figure
• Download figure as PowerPoint slide

Computed tomographic and venographic images obtained from 2 dogs with multiple or complex IHPSSs. A—Axial CT image of the liver in a young female dog demonstrating multiple intrahepatic venous collaterals prior to PTE (black arrows). B—Magnified axial CT image of the same patient in image A demonstrating very narrow shunt orifice (white arrow) at the junction of the left hepatic vein and vena cava causing pressure gradient and subsequent collaterals. C—Ventrodorsal digital subtraction portogram and cavagram in a 9-month-old castrated male Miniature Poodle demonstrating multiple intrahepatic venous collaterals (black arrows) prior to PTE. A pressure gradient was already present and the shunt was difficult to access through a small window.

Citation: Journal of the American Veterinary Medical Association 244, 1; 10.2460/javma.244.1.78

• Download Figure
• Download figure as PowerPoint slide

We hypothesize that the threshold for development of acquired intrahepatic collateral circulation through adjacent hepatic veins due to postsinusoidal hypertension is lower than the threshold for development of extrahepatic collateral circulation subsequent to presinusoidal portal hypertension. This makes sense experimentally as described,²² given that the dogs developed intrahepatic collateral circulation rather than acquiring extrahepatic shunts. If this theory was true, alternative pressure gradient guidelines may be necessary to guide embolization or attenuation at the level of the hepatic vein (postsinusoidal). Alternatively, attenuation at the level of the portal vein side of the shunt may be performed (presinusoidal), although a different technique would be necessary, posing an increased risk of occluding some naturally present portal branches deeper within the liver parenchyma. The acquired intrahepatic collaterals identified may also help explain persistent positive scintigraphy and bile acids concentration results in some cases reported previously.^2,4,5 Improved portal perfusion can coincide with these collaterals, explaining the improvement in clinical signs often identified. It remains unclear whether a portal venous pressure–CVP gradient is sufficient to improve clinical or biochemical signs when there is no apparent improvement in grossly evident portal perfusion. It remains conceivable that the presence of a pressure gradient could provide adequate resistance to portal blood flow and subsequently enhanced portal perfusion that may be beyond the ability of current imaging modalities to identify. Interestingly, 2 of 4 dogs with multiple IHPSSs and excessive resting pressures that did not receive endovascular treatment continued to do well on medical management alone 539 and 1,277 days after angiography. The presence of the portal venous pressure–CVP gradient may be sufficient to increase portal perfusion enough to maintain reasonable liver function.

Only 3 of 100 (3%) dogs were identified to have dramatic portal perfusion present on preoperative cross-sectional imaging or direct angiography, which was < 9% as described previously.² Although these patients have the best portal perfusion and many may do well without surgery, these patients are also likely have the best prognosis following complete shunt attenuation owing to the extremely low risk of portal hypertension or intrahepatic collateral development. Endovascular treatments involving placement of a vascular plug, vena cava–covered stent (stent-graft), or atrial septal defect occluder¹⁹ were safe and effective treatment methods for complete, acute shunt occlusion in these patients. Theoretically, the presence of normal or near-normal portal vasculature could increase the risk of hemorrhage during liver dissection if these shunts are approached surgically from a prehepatic or portal approach, potentially supporting endovascular treatment for this population of cases as well.

The transition from the mesh SEMS to the laser-cut SEMS greatly facilitated stent placement. Whereas the former is reconstrainable and repositionable prior to deployment, the foreshortening that occurs makes ultimate stent positioning difficult to anticipate because it expands into the shunt opening and shortens. This issue is more apparent for left-sided IHPSSs because of the left hepatic vein's cranial location and proximity to the right atrium. For this reason, in certain cases, the former stent was placed by means of a femoral vein approach to leave the leading end of the stent just caudal to the right atrium. The location of the caudal aspect of the stent is less important owing to the relatively long length of caudal vena cava, with few draining veins between the hepatic and renal veins. For both stent types, both PPV and nonventilation angiography were performed to confirm maximal caudal vena cava size for stent sizing and then compared with prior CT vena cava measurements when available. Stent oversizing of at least 20% was performed to ensure proper wall apposition to minimize stent migration. Often the stents were more dramatically oversized to fill the dilated nature of the vena cava at the level of the diaphragm and shunt entrance. No problems with postoperative stent migration or oversizing were identified in these cases. Malpositioning of the stent occurred in 5 PTE procedures; this was primarily due to the initial unavailability of appropriately sized (diameter and length) veterinary caval stents. As such, human stents^j (donated) were used in diameters and lengths not ideally suited for each individual case.

Stent placement across major venous ostia of the vena cava is performed routinely in human patients and has not been considered problematic in previous reports^24,25 or in these patients. Similar venous stents evaluated in the jugular vein, cranial vena cava, and caudal vena cava of dogs were identified to have incorporated into the vessel's tunic intima within 7 days after placement.²⁵ One histopathologic example from this study was available in a dog that died approximately 2 years after the procedure following severe gastrointestinal hemorrhage (Figure 5). Complications associated with central venous endovascular stent placement can include stent fracture, migration, occlusion, and misplacement.^22,25–27 Non–life-threatening arrhythmias (2 transient and 1 dog that later resolved on its own following stent placement at the right atrium entrance) and jugular vascular access-site hemorrhage (2 dogs) due to the large diameter stent delivery systems can also occur.²⁷ In 3 cases, a coil embolized beyond the stent and into the pulmonary circulation. Two of these 3 patients were treated early in the study when the mesh SEMSs^j were being used. Following coil placement and migration, an orthogonal angiogram confirmed the stents did not completely cover the shunt entrance. The coil was removed endovascularly in 1 dog, and a longer stent was placed within the first stent in each dog.

Histologic sections including caval stent and coils obtained at postmortem examination in a 33-month-old spayed female Labrador Retriever euthanized approximately 2 years following PTE for a perforated gastric ulcer and subsequent peritonitis. A—Section through the vena cava and caval stent (*) demonstrating endothelial lining (white arrows) covering the incorporated stent with no evidence of inflammation. B—Another section containing the caval stent (black arrows) and thrombogenic coils (white arrows) within organized fibrous tissue. C—Close-up section through coils further demonstrating permanent surrounding tissues rather than simple thrombus.

Citation: Journal of the American Veterinary Medical Association 244, 1; 10.2460/javma.244.1.78

• Download Figure
• Download figure as PowerPoint slide

View Full Size

Histologic sections including caval stent and coils obtained at postmortem examination in a 33-month-old spayed female Labrador Retriever euthanized approximately 2 years following PTE for a perforated gastric ulcer and subsequent peritonitis. A—Section through the vena cava and caval stent (*) demonstrating endothelial lining (white arrows) covering the incorporated stent with no evidence of inflammation. B—Another section containing the caval stent (black arrows) and thrombogenic coils (white arrows) within organized fibrous tissue. C—Close-up section through coils further demonstrating permanent surrounding tissues rather than simple thrombus.

Citation: Journal of the American Veterinary Medical Association 244, 1; 10.2460/javma.244.1.78

• Download Figure
• Download figure as PowerPoint slide

Histologic sections including caval stent and coils obtained at postmortem examination in a 33-month-old spayed female Labrador Retriever euthanized approximately 2 years following PTE for a perforated gastric ulcer and subsequent peritonitis. A—Section through the vena cava and caval stent (*) demonstrating endothelial lining (white arrows) covering the incorporated stent with no evidence of inflammation. B—Another section containing the caval stent (black arrows) and thrombogenic coils (white arrows) within organized fibrous tissue. C—Close-up section through coils further demonstrating permanent surrounding tissues rather than simple thrombus.

Citation: Journal of the American Veterinary Medical Association 244, 1; 10.2460/javma.244.1.78

• Download Figure
• Download figure as PowerPoint slide

Following stent placement, portal pressures were remeasured, considering that the stent alone can occasionally raise the pressure gradient. During the approximately first half of the PTE procedures in this series, a mean of 3.4 ± 1.9 coils (range, 0 to 6 coils) were placed while in the remaining cases, the mean number of coils placed was 8.1 ± 5.6 (range, 0 to 30 coils); this was a significant difference. Considering that portal hypertension was uncommonly encountered, we became more comfortable adding additional coils. There is no evidence at this time that the addition of more coils will improve the outcome. In addition, there is no evidence that 8-mm-diameter, 5-cm-long coils are preferred to either smaller- or larger-diameter coils. We chose this size and continued to use the same size coils to help create guidelines for the procedure. Youmans and Hunt⁶ reported that thrombogenic coil placement in the femoral vein was nondurable, given that recanalization occurred in 3 of 4 dogs with initial complete occlusion. Although it is conceivable that repetitive motion of the limb contributed to thrombolysis in those experimental dogs, it remains unclear why thrombogenic coil embolization recanalization does not seem to be a problem in patent ductus arteriosus occlusion and presumably not routinely in the cases of this study in which incorporation into surrounding tissues was anticipated (Figure 6). Coil embolization of an IHPSS provides not only acute occlusion (as demonstrated by acute increases in portal pressures) but also perceived progressive attenuation over an unknown and likely variable period of time as the coil-thrombus-stent combination become organized and permanent (Figures 5 and 6).

Serial gross postmortem specimens in 3 dogs that underwent PTE of an IHPSS. A—Intraluminal view of the caval stent and IHPSS opening (black circle) in a young female Labrador Retriever. B—Close-up view of the same patient demonstrating fibrous tissue around coils reducing the shunt opening. C—Shunt view of the same patient showing coils, original shunt opening diameter (white circle), and remaining shunt patency (black circle). D—Intraluminal view showing the patent hepatic vein (HV) and narrowed shunt opening due to coil fibrous tissue in a 14-month-old sexually intact female Weimaraner. E—Shunt view of the same patient showing incorporated coils, original shunt opening diameter (white circle), and minimal remaining patency. F—Intraluminal view demonstrating patent hepatic vein (HV), original shunt opening (white circle), and remaining shunt patency (black circle) in a 15-month-old sexually intact male Basset Hound.

Citation: Journal of the American Veterinary Medical Association 244, 1; 10.2460/javma.244.1.78

• Download Figure
• Download figure as PowerPoint slide

View Full Size

Serial gross postmortem specimens in 3 dogs that underwent PTE of an IHPSS. A—Intraluminal view of the caval stent and IHPSS opening (black circle) in a young female Labrador Retriever. B—Close-up view of the same patient demonstrating fibrous tissue around coils reducing the shunt opening. C—Shunt view of the same patient showing coils, original shunt opening diameter (white circle), and remaining shunt patency (black circle). D—Intraluminal view showing the patent hepatic vein (HV) and narrowed shunt opening due to coil fibrous tissue in a 14-month-old sexually intact female Weimaraner. E—Shunt view of the same patient showing incorporated coils, original shunt opening diameter (white circle), and minimal remaining patency. F—Intraluminal view demonstrating patent hepatic vein (HV), original shunt opening (white circle), and remaining shunt patency (black circle) in a 15-month-old sexually intact male Basset Hound.

Citation: Journal of the American Veterinary Medical Association 244, 1; 10.2460/javma.244.1.78

• Download Figure
• Download figure as PowerPoint slide

Serial gross postmortem specimens in 3 dogs that underwent PTE of an IHPSS. A—Intraluminal view of the caval stent and IHPSS opening (black circle) in a young female Labrador Retriever. B—Close-up view of the same patient demonstrating fibrous tissue around coils reducing the shunt opening. C—Shunt view of the same patient showing coils, original shunt opening diameter (white circle), and remaining shunt patency (black circle). D—Intraluminal view showing the patent hepatic vein (HV) and narrowed shunt opening due to coil fibrous tissue in a 14-month-old sexually intact female Weimaraner. E—Shunt view of the same patient showing incorporated coils, original shunt opening diameter (white circle), and minimal remaining patency. F—Intraluminal view demonstrating patent hepatic vein (HV), original shunt opening (white circle), and remaining shunt patency (black circle) in a 15-month-old sexually intact male Basset Hound.

Citation: Journal of the American Veterinary Medical Association 244, 1; 10.2460/javma.244.1.78

• Download Figure
• Download figure as PowerPoint slide

Individual animals likely have different thrombogenic potential. Dogs with a PSS commonly have prolonged coagulation profiles,²⁸ and dogs with hepatic disease may have various alterations in hepatic production of anticoagulation, procoagulation, and fibrinolytic proteins.²⁹ Individual variation may affect ultimate thrombosis of the coils; however, the relative improvement in clinical signs and biochemical parameters generated suggested most animals had persistent shunt attenuation. Whereas human patients often receive anticoagulant therapy for similar endovascular procedures, we chose to avoid systemic anticoagulation in these canine patients during the procedure because of the current incompletely defined nature of the clinical coagulation status in dogs with PSS.

Median procedure times of 90 minutes (range, 30 to 340 minutes) were not significantly different between the first and second half of the cases performed. Each patient's first PTE procedure was performed in the presence of the primary author (CW), but a number of individuals performed the procedures, including residents, fellows, and staff veterinarians. Procedure durations were not compared among individuals of differing experience levels because the procedures were often shared among the participants involved in each case.

Complications were relatively minor and uncommon, compared with the historical complication rate associated with surgery for IHPSS that has been reported in the literature to be as high as 55% to 77%.^2,4 Major life-threatening intraoperative complications were uncommon, compared with the historical surgical literature.^4,8–10 Portal hypertension is believed to occur uncommonly following PTE for a number of reasons. First, as described, it seems postsinusoidal pressure increases occur within the liver parenchyma and are relieved through intrahepatic venous collaterals, prior to developing portal hypertension and subsequent ascites, hemorrhagic diarrhea, gastrointestinal congestion or ischemia, and possible death. A second possible explanation may be the relative weakness of a coil thrombus to limit blood flow through the shunt orifice. It is possible that as the portal pressures increase, the portal blood simply displaces the coil thrombus rather than resulting in excessive portal pressures. This could conceivably occur more easily with a coil thrombus than with a surgically attenuated shunt in which the diameter is fixed, resulting in a narrowed opening that is less amenable to increased blood flow.

Perioperative mortality rates were also less than reported previously for most surgical treatments for IHPSS that range up to 28%.^4,7–9 Early postoperative mortality rates in this study were 5% within the first week; life-threatening postoperative complications in this study occurred in 13% of procedures, half of which were due to seizures or neurologic sequellae. Postoperative neurologic sequellae (6%) were similar to that reported previously for dogs undergoing surgery for PSS (12%; range, 5% to 18%) but may have been underrepresented initially during the study, given that we now monitor much more closely for any subtle signs of postprocedure neurologic abnormality, including minor facial or extremity twitching.³⁰ During this study, we did not routinely use antiseizure medications preoperatively unless the patients continued to have neurologic clinical signs while on the appropriate PSS medications or the patients were already taking these medications as prescribed by the referring veterinarians.

Repeated blood work was requested in all patients at the time of follow-up and was available for most over a variable time frame. Increases in albumin concentration, TS concentration, BUN concentration, cholesterol concentration, Hct, and MCV are all consistent with improvement in liver function. Increased dietary protein (high-protein content and reduced medications limiting protein absorption) could result in improvement in albumin, TS, and BUN concentrations but would not likely contribute to increases in the other values. Globulin concentration increases could be associated with improved liver function, reduced gastrointestinal dysfunction or loss (following proton-pump inhibition), or a combination of both.

Although improvements (reductions) in bile acids concentration were minimal, the pre-PTE values (n = 22 dogs) and post-PTE values (16) were not routinely obtained in patients responding well to the procedure. The PTE procedure is meant to reduce the morbidity and mortality rates associated with more invasive treatments for IHPSSs. A complete attenuation of the shunt is not always the goal. As such, continued shunting may be anticipated in many cases and elevated bile acids concentration should therefore be anticipated. Surgical outcomes of dogs with PSSs have been reported to be better when complete shunt occlusion is performed,^21,31 although this may be a result of the preexisting portal perfusion in that population of dogs rather than the ultimate technique used to attenuate the shunt. Furthermore, studies^2,10,32 have found no significant difference in outcomes for partial versus complete surgical attenuation of PSSs. Upon review of the literature in which incomplete occlusion of IHPSSs was performed, overall outcomes appear to be sufficient to preclude further shunt attenuation in most patients.^2,8,12 Papazoglou et al² reported 29 of 32 (91%) IHPSSs could not be completely attenuated, and none of the surviving dogs required a second surgery. It is likely that these animals develop sufficient but often reduced hepatic function to maintain an otherwise normal life. This suspicion is supported by the fact that 70% of a normal canine liver can be removed acutely without severe consequences,³³ and some minority of dogs with PSSs (and presumably some degree of portal perfusion) can be managed well for years with medication (or no treatment) alone.^34,35 Dogs with portal vein hypoplasia without portal hypertension (previously termed hepatic microvascular dysplasia) and reduced hepatic function can often live free of clinical signs without medical management, and animals with portal vein hypoplasia with portal hypertension (previously termed idiopathic noncirrhotic portal hypertension) and resulting multiple acquired EHPSSs can have a favorable prognosis with supportive medical care.³⁶ In addition, other conditions that result in elevated bile acids concentration such as tracheal collapse,³⁷ gastrointestinal disease,^38–40 and glucocorticoid or anticonvulsant treatment^38–40 are often subclinical for reduced hepatic function as well.

The persistently elevated bile acids concentration could be associated with persistent shunting, acquired shunting, underlying concomitant liver dysfunction, or some combination of these factors. Regardless, most of these patients improved clinically following the PTE procedures as demonstrated by reduction or elimination of medical management without the return of prior clinical signs. This is not an uncommon outcome with traditional surgery as well. A recent report⁴¹ described 94% good or excellent results for treating EHPSS with ameroid constrictors in patients in which 21% of cases evaluated were determined to have persistent shunting. Thirteen of the 16 dogs in that study with persistent shunting had good to excellent outcomes. Following hydraulic occluder placement, IHPSS-associated clinical signs resolved; however, 5 of 10 dogs had positive scintigraphy scans (persistent shunting) 2 weeks following complete occlusion with a hydraulic occluder, and 3 of 8 dogs had persistently elevated bile acids concentration values at the 1-year follow-up examination.⁵ These examples further suggest that many of these patients can do well with incomplete shunt attenuation, acquired shunting, or persistent liver dysfunction. Bile acids concentrations have been repeatedly demonstrated to have no prognostic influence in a number of IHPSS reports.^{2,5,8,10,12,32} This study was consistent with previous studies in that bile acids concentrations were not prognostic for outcome.

Interestingly, preoperative bloodwork (albumin concentration, cholesterol concentration, Hct, and MCV) was not prognostic for these patients and was instead changes in albumin concentration, globulin concentration, cholesterol concentration, BUN concentration, TS concentration, Hct, and MCV (both absolute values and percentage changes) that were associated with patient outcome in univariate analyses. Only a postoperative increase in concentration of globulin and TS remained significant for patient outcome in the multivariate analysis. This is important because it reminds clinicians that severe preoperative bloodwork abnormalities do not imply a poor prognosis for these dogs, and each patient should have the opportunity to be treated.

Medical management (lactulose and antimicrobials) or low-protein diet was still being occasionally used in 28 of 85 (33%) cases and 19 of 84 (23%) cases, respectively, despite our recommendations. A number of dogs remained on special diets (eg, novel protein or intestinal diets) because of gastrointestinal signs. It remains unclear whether these clinical signs are due to the underlying liver disease or other medical conditions, given that additional diagnostic tests were not routinely available. It is also likely that an unknown proportion of these animals no longer required medical or dietary management; however, many owners continued ≥ 1 treatments for fear of clinical sign recurrence. Any medical or dietary treatments were included in the statistical analysis of these dogs regardless of our clinical impressions of the necessity of these treatments, likely overestimating the number of dogs requiring persistent medical management.

Although lifelong acid suppression treatment was ultimately recommended, at follow-up examination, only 68 of 84 (81%) animals were receiving antacid treatment, reminding clinicians to maintain contact with owners whom otherwise may become complacent. Nineteen of 90 (21%) dogs had postprocedure clinical signs associated with gastrointestinal hemorrhage at some point during the follow-up period. We believe that dogs with IHPSS have a gastrointestinal disorder that is likely a separate, concomitant condition. Unpublished data on 51 dogs in this study determined a high incidence (17%) of preprocedural gastrointestinal ulceration present on endoscopy and biopsy results, consistent with moderate to severe inflammatory bowel disease. In this same study, dogs treated prior to lifelong-antacid treatment had approximately 50% of deaths due to gastrointestinal bleeding, whereas following life-long antacid treatment, approximately 4% of deaths were due to gastrointestinal bleeding. This finding led to the use of life-long proton-pump inhibitors beginning around 2005 and careful management and monitoring of gastrointestinal disease in this population of dogs. Even though the relative risks of gastrointestinal complications in dogs with IHPSS are now becoming known, the consequences of lifelong proton-pump inhibition are not clear at this time. Evidence of gastrointestinal bleeding such as melena (13/100), hematochezia (1/99), or blood in vomit (1/99) was apparent prior to surgery. These numbers are likely underestimates of the true incidence, considering that we grew more diligent in identifying these signs throughout the study. Performing the PTE procedure (or any shunt attenuation technique) can increase portal pressures and exacerbate previously subclinical gastrointestinal bleeding. In patients with evidence of preoperative gastrointestinal bleeding, aggressive medical management is recommended prior to shunt attenuation.

Repeated procedures were performed in 15 of 95 (16%) dogs. The first 3 cases had additional procedures scheduled until we decided to alter the protocol and only repeat the procedure if necessary. Performing repeated PTE was at the discretion of the owner; it was recommended if clinical signs recurred or did not resolve, but some owners avoided a repeated procedure because of cost or contentment with their dog's current quality of life. The 16% occurrence of repeated PTE is likely an underestimation if left to the authors, but the true need for repeated PTE remains unknown, likely not dissimilar to partial suture ligation in many cases. During long-term follow-up in 88 dogs, 29 (33%) did ultimately demonstrate 1 or more clinical signs possibly associated with liver dysfunction. Retrospectively, it is difficult to determine the underlying cause of these signs in all cases, but it is more likely that between 16% and 33% may benefit from additional PTE procedures.

Median and mean follow-up times available for 98 dogs were 962 and 1,041 ± 821 days (range, 0 to 3,411 days), respectively. This is the longest follow-up time reported to date for a large series of dogs with IHPSSs. Median survival time was 2,204 (95% CI, 1,295 to 2,461) days (range, 0 to 3,411 days; 95% CI, 1,295 to 2,461 days), which is longer than survival times previously reported for other treatments of dogs with this condition. Dogs with an excellent outcome had a longer median survival time, compared with the outcome of other dogs (2,435 days [95% CI, 1,723 days and indeterminate] vs 649 days (95% CI, 211 and 2,442 days, respectively). On the basis of the presence of clinical signs, current medications (if present), and serum biochemical results, PTE outcome was determined by the authors to be excellent (66%), fair (15%), or poor (19%). The individual owners' interpretation of patient outcome was intentionally removed from the outcome measurements in this study to maintain consistency among all of the patients.

Five dogs did not receive PTE procedures because of elevated resting pressures. These decisions were subjective in nature, and some dogs with similar pressure gradients were ultimately treated later in the study owing to the author's comfort with the procedure following more experience with the technique. These 5 patients were included in the study to conform as closely as possible to the goal of 100 consecutive IHPSS dogs. We felt it was important to include a population of IHPSS dogs that had a concurrent resting portal venous pressure–CVP gradient or elevated resting portal pressures. With the advent of progressive vascular occlusion devices such as the ameroid constrictor, the cellophane band, and the more staged, acute attenuation of the hydraulic occluder, it is likely that fewer surgeons have been measuring portal venous pressure or CVP during shunt surgery. Even with more progressive attenuation devices, shunt attenuation in these cases may be detrimental at least, and dangerous at worst, in this subset of patients. Given that the threshold for acquiring intrahepatic shunts or interlobular hepatic venous drainage appears to be lower than that necessary to acquire multiple extrahepatic shunts, these guidelines may need to be revised or reconsidered for IHPSS when attenuated in a postsinusoidal location. This lower threshold and subsequent development of intrahepatic alternative drainage likely also explain at least partially the reason life-threatening portal hypertension was uncommon in the PTE patients. It may be safer (less portal hypertension) to perform IHPSS attenuation from the postsinusoidal approach; however, the risk of acquired intrahepatic drainage is likely to be higher. Alternatively, the presinusoidal approach may result in higher risk of portal hypertension because of the relative poor compliance of the canine portal system to increased portal pressures; however, the historical guidelines used for EHPSS attenuation may be more appropriate. In addition, presinusoidal attenuation could arguably be preferred to encourage portal perfusion.

In conclusion, the results of this retrospective study suggested that PTE is a safe, fast, and potentially effective treatment for canine IHPSS associated with lower morbidity and mortality rates and equal (or even improved) long-term outcomes, compared with traditional open surgical techniques, when performed by appropriately trained personnel. Prospective, randomized studies are necessary to confirm these findings. Because the patients in this series appeared to have an increased risk of gastrointestinal hemorrhage prior to, and following, treatment for the IHPSS, we currently recommend lifelong proton-pump inhibition because no adverse effects associated with their use have been identified to date. Approximately 10% of these dogs will have multiple IHPSS and approximately 5% will have prior elevated portal pressures or portal venous pressure–CVP gradient. The implications of the latter finding remain unknown at this time. Preoperative bloodwork values do not appear to be significantly correlated with patient outcome; however, changes in these values after shunt embolization were significant. Postsinusoidal IHPSS attenuation is associated with a currently unknown risk of intrahepatic venous collateral development that likely reduces the risk of subsequent portal hypertension and associated complications as well as potentially reduces development of portal perfusion. The results of this study suggested that although 16% to 33% of dogs with IHPSS treated by means of the endovascular procedure described may require an additional procedure, PTE results in fair to excellent outcomes in 81% of patients with a median survival time of > 6 years.