Acute Portal Hypertension Models in Dogs: Low- and High-Flow Approaches

Abstract

Effective animal models are needed to evaluate the feasibility of new techniques to assess portal hypertension (PH). Here we developed 2 canine models of acute PH by increasing intrasinusoidal resistance and by increasing the portal vein (PV) flow volume to test the efficacy of a noninvasive technique to evaluate PH. The acute low-flow PH model was based on embolization of liver circulation by using a gelatin sponge material. The acute high-flow PH model was based on increasing the PV flow volume by using an arteriovenous (A-V) shunt from the femoral artery and saline infusion. PV pressures and diameters were assessed before and after inducing PH. Pressure values and diameters were obtained from the inferior vena cava in 3 unmanipulated controls. The low-flow model of PH was repeatable and successfully increased PV pressure by an average of 16.5 mm Hg within 15 min. The high-flow model of PH failed to achieve increased PV pressures. However, saline supplementation of the portal circulation in the high-flow model led to mean increases in PV pressures of 12.8 mm Hg within 20 min. Pulsatility in the PV was decreased in the low-flow model and increased in the high-flow model relative to baseline. No changes in PV diameter were noted in either model. These acute PH models are relatively straightforward to implement and may facilitate the evaluation of new techniques to assess PH.

Portal hypertension (PH) marks the silent progression of cirrhotic liver from a preclinical to a clinical phase, when symptoms associated with common complications like ascites, variceal bleeding, and encephalopathy manifest and can be fatal (mortality rates of 20% to 70%).^7,12,31,34 The hepatic venous pressure gradient is used clinically to diagnose PH, but the technique is invasive and is not routinely used for screening potential PH cases.^26,39 Development of a noninvasive technique to identify nascent PH has great clinical value.³⁹ The long-term goal of our project is to develop and evaluate a contrast-enhanced ultrasound imaging-based noninvasive pressure estimation technique^{9,10,17,20,37} as a means to identify PH. Before clinical testing of this approach, we conducted feasibility studies in animal models to confirm the accuracy of this new technique. Here we developed 2 canine acute PH models to test the efficacy of our noninvasive pressure estimation technique to evaluate PH.

Current animal PH models include those in mice, rats, pigs, rabbits, and dogs.^{1,5,11,13,21,23,27,33,38,41} Guidelines for selecting a particular animal PH model (or for other hepatic abnormalities) have also been published.³⁰ In view of these guidelines, we selected dogs for our study, because the blood vessels of smaller animals may be difficult to catheterize and because some species, like swine, develop marked pulmonary hypertension during contrast-enhanced ultrasound imaging.³² Moreover, our team has extensive experience with canine experiments, and we have noted no side effects in dogs after the administration of the contrast material for ultrasound imaging.^9,10,17 Furthermore the physiology of dogs closely approximates that of humans.²³

Animal models and clinical studies have shown that a pathophysiologic increase in intrahepatic vascular resistance ultimately leads to PH,^12,31 whereas etiologically the origin of PH may be pre-, intra-, or postsinusoidal.⁶ Overall, a modified form of Ohm's law for fluid flow (∆P = Q × R, where ∆P is the change in pressure, Q is flow, and R is resistance) explains increases in portal vein (PV) pressures,¹ either by an increase in PV flow, an increase in resistance to PV flow, or a combination of these 2 mechanisms. Therefore, we sought to develop 2 acute PH models in dogs, with one resulting from increased resistance to PV flow (a low-flow model of PH) and the other from an increase in the volume of PV flow (a high-flow model of PH).

For our acute low-flow model of PH, we increased the intrasinusoidal resistance in the liver parenchyma by injecting a gelatin sponge material (Gelfoam, Ethicon, Somerville, NJ) via the PV. This product has been used previously in dogs²⁸ and does not elicit foreign-body reactions or inflammation.²² We hypothesized that PV pressures would increase and the pulsatility of PV flow would decrease (due to the decrease in flow) after administration of the gelatin material. For the acute high-flow PH model, we used an arteriovenous (A-V) shunt, which was created by connecting the splenic or femoral artery to the PV. We hypothesized that, in this model, both PV pressure and pulsatility would increase (due to the superimposition of arterial flow) after initiating the high-flow conditions in the PV.

Materials and Methods

This research protocol was approved by the IACUC of Thomas Jefferson University and conducted in accordance with NIH guidelines. We obtained 22 mongrel canines (Canis familiaris; weight, 20.8 to 25.0 kg; age, 6 to 12 mo; 20 male; 2 female) from Marshall BioResources (North Rose, NY). The vendor vaccinated the dogs against bordetella, parainfluenza, adenovirus, leptospirosis, distemper, parvovirus, and rabies; these vaccinations were updated as needed. Before their arrival at the study site, the dogs were dewormed with pyrantel pamoate anthelmintic (Med-Pharmex, Pomona, CA) and ivermectin (Merck, Whitehouse Station, NJ). The dogs in this project required no additional medical treatment throughout the study.

When the dogs arrived, their general appearance was recorded and a detailed physical examination, including temperature, pulse rate, and respiratory rate, was performed. The canines were housed in the AAALAC-accredited facilities of our university and were cared for and monitored by using practices that comply with the principles for the utilization and care of vertebrate animals used in testing, research, and training.² The dogs were housed in large, well-ventilated animal pens and were allowed to acclimate for at least 72 h before they were used in the research study. They were fed a mix of canned dog food (Science Diet) and laboratory chow (Purina Mills, Gray Summit, MO). After the experiments, the dogs were euthanized by intravenous injection of Beuthanasia (Euthasol; Virbac, Fort Worth, TX; dose, 0.25 mg/kg), in accordance with the recommendations of the Panel on Euthanasia of the American Veterinary Medical Association.³

Experimental design.

The dogs were assigned randomly to 1 of 3 groups. One group comprised of 3 dogs used to obtain control data from the inferior vena cava (IVC). Another group comprised of 13 canines used to create and evaluate the low-flow PH model; the remaining group involved 6 dogs in which the high-flow PH model was used. The dogs were fasted for at least 12 h prior to experiments, to reduce postprandial effects on PV flow and pressures.¹⁶

Anesthesia protocol and animal care during experiments.

Intravenous propofol (dose, 7 mg/kg; Abbott Laboratories, Chicago, IL) was used for initial sedation. The dogs were intubated, and anesthesia was maintained with 0.5% to 2% isoflurane (Iso-thesia, Abbott Laboratories) via endotracheal tube throughout the experiment. For this purpose, a calibrated anesthesia ventilator (model 2000, Hallowell EMC, Pittsfield, MA) was used with the anesthesia unit (Quatiflex VMC, Matrx Medical, Orchard Park, NY). Throughout the experiment, the dogs’ anesthesia, respiration, and ventilation were monitored (Capnomac Ultima, Datex, Ohmeda, Finland) by certified veterinary technicians. During the course of the experiment, the dogs rested on a warming blanket connected to a heated water pump (TP 400, Gaymar Industries, Orchard Park, NY) to maintain body temperature. The dogs’ electrocardiogram, temperature, and respiration and pulse-oximeter data were monitored continuously (Vet/OX Plus 4700, Sensor Devices, Lancaster, PA).

Experiment procedure.

All experimental procedures were performed when the dogs were fully anesthetized (stage 3, plane 2).⁴⁰ An 18-gauge catheter was placed in the cephalic vein for infusion of Sonazoid (GE Healthcare, Oslo, Norway) microbubbles at 0.015 to 1.5 μL/kg/min; these microbubbles act as a contrast agent for ultrasound imaging and have been previously used in dogs^9,10,17 and humans¹⁸ without any side effects. For the purposes of this study, Sonazoid microbubbles helped to delineate the PV when scanned with a Logiq 9 ultrasound scanner (GE Healthcare, Milwaukee, WI) and using a 4C probe.¹⁵

For obtaining control data from the IVC (n = 3), a calibrated 5-French pressure catheter (SPR 350S/ SPR 350, Millar Instruments, Houston, TX) was introduced into the IVC through the femoral vein. The catheter was advanced into the IVC under ultrasound guidance. The presence and localization of the pressure catheter in the IVC was confirmed by using color Doppler imaging (Figure 1). The IVC diameter was measured from the sonograms.

Figure 1.

Color Doppler image depicting flow (blue) in the inferior vena cava. Solid arrows indicate the pressure catheter in the inferior vena cava; the dotted arrow indicates the diameter of the vessel.

For obtaining data from the PV, a midline abdominal incision was created by using sterile technique to provide access to the main PV or one of its branches (Figure 2 A). A PV branch 15 cm distal to the main PV was identified, into which a calibrated 5-French pressure catheter was introduced (Figure 2 A) and then advanced into the main PV trunk under ultrasound guidance (Figure 2 B). The pressure catheter was fixed with 2-0 silk suture at the entry point of the PV branch (Figure 2 A) to maintain the catheter position during baseline and PH data acquisition. Next, a 5-French catheter was inserted into another branch of the PV to access the main PV for gelatin sponge injection or A-V shunting when needed. Patency of the PV was confirmed by ultrasound imaging after these surgical procedures.

Figure 2.

(A) Surgical procedure to introduce pressure catheter (green arrows) into the main portal vein. The blue arrow indicates the direction toward the liver. (B) Grayscale ultrasound image of the portal vein of a canine. The solid arrow indicates the pressure catheter; the dotted arrow indicates the diameter of the vessel.

The acute low-flow model of PH (n =13) was induced by embolization of the liver circulation by injecting gelatin sponge material (Gelfoam). Sterile sheets of this material (100 cm²) were cut into small pieces (less than 5 mm in each dimension) and immersed into sterile saline solution. The resulting material-containing solution was introduced into the PV through the 5-French catheter. The acute high-flow PH model (n = 6) was created surgically by connecting the femoral artery to the PV by using a 3-way stopcock with extension tubing, thereby creating an A-V shunt. High-pressure saline infusion generated by an inflatable pressure cuff was used to supplement the flow volume when needed. For both PH models, PV pressures were monitored continuously by using the pressure catheter, and the patency of the main PV was confirmed by imaging. The normal PV pressures in dogs vary from 7 to 11 mm Hg, whereas normal mean IVC pressures vary from 5.6 to 6.6 mm Hg.^{4,5,23,33,38,41} Therefore, we defined the PH endpoint for our study as a stabilized elevation (tracked for more than 2 min) in PV pressure of at least 5 mm Hg above the baseline value. The IVC pressures obtained from the 3 control dogs were used for cross-verification of induced PH, because PH also is defined as being present when the difference between the IVC and PV pressures is greater than 5 mm Hg.³⁹

The pressure catheter was connected to an oscilloscope (LeCroy, Chestnut Ridge, NY) by using the catheter's transducer control unit (TCB 500, Millar Instruments). The oscilloscope was configured to acquire data from the pressure catheter data onto a computer via a general-purpose interface bus by using LabVIEW software (National Instruments, Austin, TX).

PV data acquisition.

Baseline pressures were monitored for 15 to 45 min, and pressure signals were recorded on the computer for 27 to 36 acquisitions, with each acquisition containing 5 s of data. The PV diameters were obtained from the ultrasound images. PH then was induced; pressures after PH induction were measured for 15 to 45 min, and PV pressure and diameter data were acquired as previously.

IVC and PV pressure data were analyzed by using Matlab software (The Mathworks, Natick, MA). The mean pressure signals from each acquisition were averaged to obtain IVC and PV pressures for each dog. The pulsatility index was calculated as the ratio of the difference between maximal and minimal pressures to the mean pressure value for each canine (similar to the use of the pulsatility index for characterizing velocities;¹⁹ but here, this index was modified to characterize pressures).

Statistical analyses.

For all statistical analyses, a P value less than 0.05 was considered significant. The pressures, pulsatility index, and diameter of the PV before and after inducing PH were compared by using 2-tailed paired t tests. All analyses were performed by using SPSS Statistics (release 19.0.0, IBM, Armonk, NY).

Results

PH induction in canines.

The baseline IVC pressure (mean ± 1 SD) obtained from the 3 control dogs was 5.7 ± 1.2 mm Hg (range, 4.4 to 6.7 mm Hg); the baseline PV pressure from the dogs in the PH groups was 10.1 ± 2.7 mm Hg (range, 7.1 to 15.8 mm Hg). For the dogs in the low-flow PH group, one-third to 4 sheets of gelatin material were needed to reach the PH endpoint, and the time needed to attain the PH endpoint ranged from 2 to 15 min (Table 1). There was no correlation between the quantity of gelatin sponge used and time needed to induce PH (r = 0.22) or a change in PV pressure (r = 0.05). The PH endpoint was achieved in 10 of the 13 dogs in this group (Table 1); data acquisition was hampered in the remaining 3 dogs, as explained in Table 1.

Table 1.

Induction of portal hypertension in the low-flow model

Dog	No. of sheets of gelatin sponge used	Time (min) to induce portal hypertensiona
1	8	not applicableb
2	4	8
3	3	4
4	2	2
5	2	6
6c	not applicable	not applicable
7	1	8
8	2	6
9	0.33	5
10	0.5	5
11	1	15
12d	not applicable	not applicable
13	0.5	10

^aAfter injection of gelatin sponge until the PH endpoint

^bNo increase in PV pressure was seen, and the experiment was terminated at 60 min after injection of the gelatin material

^cBecause the pressure sensor was damaged, data from this dog were not considered

^dAcute embolization was noted on ultrasound images, and the experiment was terminated

Table 2 summarizes the details regarding the acquired PH status of the 6 dogs in the high-flow (A-V shunt) group. In dogs 1 and 2 (Table 2), when increased flow volume was obtained from the femoral artery alone, the pulsatility index increased, but the defined PH endpoint was not reached. In dog 3, no rise in PV pressure was observed after 45 min, and the experiment was terminated without acquiring any data for the postPH phase. For dogs 4 through 6 (Table 2) the PV flow volume due to the A-V shunt was augmented by saline infusion into portal circulation via a 3-way stopcock connection; in these cases, the PH endpoint was reached after 20 min.

Table 2.

Induction of portal hypertension in the high-flow model

Dog	Source of flow to the portal vein	Time (min) to induce portal hypertensiona
1	Femoral artery	not applicableb
2	Femoral artery	not applicablebc
3	Femoral artery	not applicabled
4	Femoral artery + saline infusion	20
5	Femoral artery + saline infusion	20
6	Femoral artery + saline infusion	20

^aFrom opening the 3-way stopcock in the PV until reaching the PH endpoint

^bNo increase in PV pressures were seen, but data were acquired during the postPH phase

^cThe pressure signal showed high pulsatility

^dNo rise in PV pressure was observed after 45 min, and the experiment was terminated; no data were collected for the postPH phase

IVC and PV data.

Figure 3 shows the pressure data from a representative dog in each group. Figure 3 A shows the pressure trace from the IVC, whereas Figure 3 B and C depict sample PV pressures under baseline and PH conditions from dogs in the low-flow and high-flow groups, respectively. Note that the PH pressure trace from the dog with the A-V shunt (Figure 3 C) shows more pulsatility than do the control and low-flow PH pressure tracings. The pressures, pulsatility indices, and diameters obtained from the IVC and PV from all dogs are summarized in Table 3.

Figure 3.

(A) Pressure trace obtained from the inferior vena cava (IVC) of a canine. (B) Pressure traces obtained from the portal vein (PV) under baseline and portal hypertension (PH) conditions in a dog in the low-flow group. (C) Pressure traces obtained from the PV under baseline and PH states in a dog from the high-flow group. Note that the high pulsatility seen in the PV pressures after PH in the high-flow group relative to the pressures in panel B and the baseline pressure in panel C.

Table 3.

Data (mean ± 1 SD) obtained before (baseline) and after induction of PH

Experimental condition: location	Pressure (mm Hg)	Pulsatility index	Diameter of portal vein (mm)
Baseline: inferior vena cava	5.7 ± 1.2	1.6 ± 0.3	11.4 ± 3.3
Baseline: portal vein	10.1 ± 2.7	0.6 ± 0.3	9.3 ± 1.7
Low-flow PH: portal vein	26.5 ± 7.5	0.3 ± 0.2	9.6 ± 1.6
High-flow PH: portal vein (n = 6)	19.1 ± 6.5	0.8 ± 0.4	9.7 ± 1.6
High-flow PH: portal vein (n = 3)a	23.2 ± 4.3	0.6 ± 0.1	9.8 ± 1.5

^aSubgroup of dogs that received flow from femoral artery and saline infusion

For all dogs in the PH model groups, PV data were acquired before and after PH (Figure 4). For the low-flow group, note the increase in PV pressures and decrease in pulsatility index after PH induction (Figure 4 A and B). For the high-flow group, PV pressures showed an increase only when saline was added to increase the flow volume (Figure 4 C); the pulsatility index values increased after the induction of PH (Figure 4 D). The statistical comparisons of PV pressures, pulsatility index, and PV diameter before and after inducing PH are provided in the following section.

Figure 4.

(A) Pressures and (B) pulsatility index of the portal vein (PV) before and after inducing portal hypertension (PH) in all dogs in the low-flow group. (C) PV pressures and (D) pulsatility index before and after inducing PH in all dogs in the high-flow group. In panels C and D, the open symbols denote the dogs in which PH was not achieved (that is, the increased flow in the PV arose only from the femoral artery).

Statistical analyses comparing parameters before and after inducing PH.

The low-flow PH model showed a statistically significant (P < 0.001) difference between PV pressures before and after efforts to induce PH (Table 4). In addition, the pulsatility index was significantly (P = 0.012) lower after PH induction, suggesting a low and damped flow relative to baseline conditions (Table 4). The diameter of the PV did not differ between baseline and PH conditions.

Table 4.

Comparison of portal vein data (mean ± 1 SD) obtained before (baseline) and after PH

	PH model	Difference (95% confidence interval)	P
Pressure (mm Hg)	Low-flow	16.5 ± 7.6 (11.5 to 21.6)	<0.001
	High-flow	7.8 ± 7.6 (0.9 to 16.6)	0.067
	High-flowa	12.8 ± 2.2 (7.4 to 18.3)	0.010
Pulsatility index	Low-flow	−0.3 ± 0.4 (−0.6 to −0.1)	0.012
	High-flow	0.4 ± 0.3 (0.1 to 0.7)	0.031
	High-flowa	0.3 ± 0.2 (−0.2 to 0.8)	0.109
Diameter (mm)	Low-flow	−0.1 ± 1.6 (−1.8 to 1.6)	0.886
	High-flow	−0.4 ± 1.7 (−2.5 to 1.7)	0.638
	High-flowa	−1.1 ± 1.7 (−5.4 to 3.1)	0.368

^aSubgroup of dogs that received flow from femoral artery and saline infusion

For the high-flow model, the difference in pressures before and after inducing PH approached significance (P = 0.067) when all dogs in the group were considered; this difference reached significance (P = 0.010) for the subgroup of dogs in which saline infusion was added to increase the PV flow volume (Table 4). In contrast, the pulsatility index differed significantly (P = 0.031) before and after PH when all dogs were considered, but this difference was not significant (P = 0.109) when the subgroup of dogs with the pressure saline infusion was considered separately (Table 4). Portal vein diameters did not differ before and after inducing PH (Table 4).

Discussion

We here studied 2 canine acute PH models to investigate the efficacy of a noninvasive pressure estimation technique for the assessment of PH. After consideration of published guidelines, we selected a canine model for our feasibility studies.³⁰ Portal hypertension can result due to increased resistance to PV flow, increased PV flow volume, or a combination of these.^1,6,12,31 Therefore, we assessed the robustness of this noninvasive technique to evaluate PH under both conditions. Consequently, 2 acute PH models were developed: a gelatin sponge material was introduced surgically to mimic an increased resistance (low-flow) scenario, and an A-V shunt was created to increase flow volume (the high-flow model). The PH endpoint was defined as an increase in PV pressure of at least 5 mm Hg above baseline values. Baseline IVC pressure was 5.7 ± 1.2 mm Hg (Table 3), which agrees with the published mean of 5.6 to 6.6 mm Hg and range of 2.7 to 10 mm Hg.³⁶ Baseline PV pressure was 10.1 ± 2.7 mm Hg (Table 3), which is consistent with reported mean values ranging from 7 to 11 mm Hg.^4,5,23,33,41

For the low-flow model, the PH endpoint was achieved in 10 of 11 dogs (success rate, 91%); experimental difficulties not associated with the PH model hampered data acquisition in the 2 remaining dogs in this group. The data obtained before and after PH in the 10 dogs in which the model was created successfully supported the hypotheses that PV pressures would increase and the pulsatility index would decrease after embolization of the hepatic circulation by using the gelatin material. The resistance to PV flow and induction of PH occurred within 2 to 15 min after injection of the gelatin material. However, there was no correlation between the quantity of gelatin sponge used and the magnitude of PH or time needed to induce PH, suggesting that high variability in PV anatomy and liver size led to the intersubject differences.

PV pressure failed to increase in the 3 dogs in the high-flow group in which the flow from the femoral artery was integrated with the PV flow (Figure 4 C), but the pulsatility index increased in 2 of these dogs. For the remaining 3 dogs in the high-flow group, saline infusion was used to supplement the flow from the femoral artery, leading to successful induction of PH. These results imply that when the femoral artery was coupled to the PV, the compliance in the PV may be sufficient to compensate for the increased flow volume and led to the high pulsatility index (by 0.4 or 40% higher relative to baseline values, Table 4). When the flow from the femoral artery was supplemented with saline, the rise in the pulsatility index was relatively low (0.3 or 30%), perhaps because the PV had reached maximal compliance or because the pulsatility effect of the direct femoral artery flow was muted by the diluted inflow. When all 6 dogs in this group were considered, the hypothesis associated with a rise in PV pressures failed to reach statistical significance, whereas that associated with an increase in pulsatility index was accepted. In comparison, for the subgroup of 3 dogs that received saline, the hypothesis associated with a rise in PV pressures was accepted but that dealing with an increase in the pulsatility index would be rejected; these findings reflect the effect of a small sample size.

Canine PH models traditionally have aimed to elucidate the mechanism of PH development^1,5,23,27,41 (which is still not well understood³⁹), and in these studies, chronic PH typically is induced to evaluate the associated physiologic changes. For our application, we required relatively straightforward and acute low- and high-flow PH models, wherein the rise in PV pressures occurred within 30 min after acquiring baseline data and ultrasound scanning after contrast administration was feasible.

Various canine PH models have been proposed.^{5,23,27,33,41} A whole-liver compression PH model using a polypropylene mesh resulted in the death of 1 dog and we therefore excluded this model from our consideration.⁴¹ In addition, because that approach involved wrapping the whole liver with mesh or gauze,⁴¹ it was not applicable for evaluation of our ultrasound technique, because of possible scattering of the ultrasound beam by the mesh or gauze. Another approach is the use of portal and splenic vein stenoses to study the relationship between hypersplenism and hemodynamics in PH.⁵ However, the constriction of portal and splenic veins would affect the amount of ultrasound contrast that reached the PV, and therefore this model had limited applicability for our scenario. In another study, authors used anti-E. coli serum to induce portal fibrosis and tracked PH related changes over 2 wk,³⁸ but this time-frame to observe PH changes was unsuitable for our application. An approach involving the use of microspheres for inducing PH has been shown to be successful in dogs, but the cost associated with these microspheres is relatively high.²³ We briefly considered an approach for developing a multiple portosystemic shunt model by using dimethylnitrosamine,²¹ but the time taken to induce these shunts varies between 8 to 14 wk (maximum, 56 wk) and therefore was not practical for our purpose. Vascular abnormalities subsequent to natural A-V shunts between the hepatic artery and PV have been reported in literature,^24,29,35 but none were noted in the dogs that were used for our experiments.

We opted to use a gelatin sponge material to create our low-flow PH model because this product has previously been used in canines with no reported side effects,^22,28 is known for its hemostatic properties, is inexpensive, and is readily available in hospital and animal laboratory settings. Our success rate in inducing PH was 91%. For chronic PH models, the applicability of this product to create a low-flow PH model is limited, mainly due to recanalization.¹⁴ Therefore, the choice of our low-flow PH model is application-specific. Similarly, it is a known that a hyperkinetic syndrome (due to initial increase in resistance to PV flow) ultimately results in increased PV flow and contributes to PH.¹ We used an A-V shunt model in our study because, physiologically, the hyperkinetic syndrome and resulting increase in PV flow volume may require considerable time (that is, weeks) to develop after resistance to PV flow increases. The A-V shunt model we used here simulated high-flow conditions but did not contribute to increases in PV pressures until saline infusion was used to supplement the flow. This scenario suggests that A-V shunting alone may not be sufficient to produce an acute high-flow PH model, as is consistent with findings of a previous study,²⁵ which showed that A-V shunts can induce modest distal-vein hypertension because the high pressure at the arterial end may be dissipated before affecting pressures in a distal vein.

A limitation of these models in our study was that the resulting hemodynamic changes in the PV were not tracked longitudinally. Nonetheless the data obtained from these dogs may help in the development of a noninvasive technique to identify PH and therefore may be of immense clinical value. Another limitation of this study is that conclusions regarding our high-flow model were derived from only 6 dogs; a minimum of 15 dogs would have been required to obtain greater than 80% power. Therefore data from the high-flow model likely require additional validation. Finally, both of the acute PH models we used here may be applicable for inducing PH in other species (for example, swine). The development and capability of our contrast-ultrasound–based noninvasive pressure estimation technique in detecting PH is addressed elsewhere.⁸

In conclusion, we have developed 2 canine models of acute PH. Our low-flow PH model was repeatable and successful in increasing PV pressures by 16.5 mm Hg on average within 15 min. A high-flow PH model developed by creating an A-V shunt (between the femoral artery and PV vein) was unsuccessful in increasing PV pressures. However, when the flow from the femoral artery was supplemented by infusing saline, mean increases in PV pressures of 12.8 mm Hg were achieved. The pulsatility decreased after PH was induced in the low-flow model and increased after PH was induced in the high-flow model. No changes in PV diameter were noted after inducing PH in either model. These PH models are relatively straightforward to implement and may be used for evaluation of new techniques for detecting PH. Our canine models have limited applicability to the study of chronic physiologic changes associated with PH.