Selection of an appropriate animal model is critical to the success of any scientific study in which the goal is to translate biomedical information across the species barrier for human application. Although numerous anatomic and biochemical differences often produce species specific responses to medical interventions, there remains an acute need to use these suboptimal techniques until practical alternatives can be developed. In the case of surgical interventions, the most important consideration is often the degree to which physiologic structures in the animal model mimic human anatomy.
Just as the shape and contour of molecular binding sites activate certain drugs in some species while blocking them in others, 1,2 subtle differences in body size, shape, orientation, and vasculature can prove problematic when using large animals to test certain surgical procedures and implantable devices. This is especially true when the device under study is designed to work in conjunction with patient anatomic structures to perform a specific function.
This retrospective study summarizes experience in testing one such device (a substernal ventricular compression balloon) in two large animal models commonly used in thoracic surgery research. Results show how model selection can shape test outcomes and clearly demonstrate that choice of test species can profoundly influence the predictive value of implant studies. Knowledge gained through these experiments has yielded valuable insights regarding the strengths and limitations of both these animal models, as well as information that can be used to improve future study designs and help guide selection of appropriate animal models for biomechanical implant testing.
Materials and Methods
Studies in both pigs and dogs were performed to test the efficacy of a cardiac compression device (CCD) designed to assist the heart via pneumatic actuation of a balloon placed between the right ventricle (RV) and the chest wall. 3 The impetus driving CCD development was the need for a quick and simple means to support the pulmonary circulation in response to the rapid onset of RV dysfunction. A quick-response capability is not provided by the armamentarium of mechanical assist devices available to surgeons today, which include centrifugal and axial flow devices, positive displacement pumps, and pulmonary counterpulsation balloons.
The prototype devices tested here all consisted of a disc shaped silicone balloon attached to a pneumatic drive line as shown in Figure 1. In pigs, a patch of Teflon felt was secured to the underside of the balloon using a silicone-based adhesive for the purpose of device anchoring. A patch of medical grade silastic sheeting with similar mechanical properties was used in an analogous manner in dogs. Standard intraaortic balloon pump (IABP) drive line tubing was fastened to the balloon sidewall port (1.5 cm outer diameter) via a straight ⅜″ to 1/2″ connector. In pigs, the connector was fitted with a rigid domed mesh to reduce wall stresses at the connector balloon interface. In dogs, the throat of the balloon was stabilized by fixing it to the silastic backing with silicone rubber. In both experiments, attachment sites were pneumatically sealed using ultraviolet curable adhesive (Loctite # 3311 Locite Corp., Rocky Hill, CT).
Unstressed, the balloons measured 5 cm in diameter and 2 cm in height and occupied an internal volume of 40 mL. Because of their high compliance, the balloon volumes could be repeatedly expanded to twice this value without damage. Two lateral convolutions allowed the balloons to lie flat under negative pressures while presenting a smooth, flat surface to effect uniform displacement of the RV free wall.
In both studies, device actuation was controlled using a commercial IABP pump console (System 83, Datascope Corp., Montvale, NJ) modified to deliver a preset volume of helium gas to the balloon during cardiac systole. To meet volume requirements, a second safety chamber was connected in parallel with the original to double driver capacity to 65 mL. Cardiac synchronization was achieved by disabling the inflation delay feature, thereby allowing balloon inflation to occur immediately after detection of the QRS complex. Apart from these two alterations, drive console operating conditions were identical for both RV copulsation and aortic counterpulsation.
The surgical procedures described as follows were performed in compliance with the Guide for the Care and Use of Laboratory Animals prepared by the National Academy of Sciences and published by National Academy Press (ISBN 0–309–05377–3, 1996). Both projects were approved by the Institutional Animal Care and Use Committee of Allegheny-Singer Research Institute (Pittsburgh, PA).
Experimental Procedure in Pigs
Six female Yorkshire pigs, weighing 35–45 kg, were used in a pilot study to determine the feasibility of supporting right ventricular function using this external cardiac compression technique. Each animal was sedated with intramuscular injections of ketamine (615 mg) and xylazine HCl (80 mg) before induction of general anesthesia. Anesthesia was maintained with 1–2% isoflurane delivered through an endotracheal tube. A Swan-Ganz catheter and arterial pressure line were placed via the external jugular vein and carotid artery respectively. Right ventricular stroke volume was calculated based on Doppler velocity profiles measured across the pulmonary valve via transesophageal echocardiography (TEE).
Acute RV failure was induced by graded right coronary artery ligation. 4,5 RV dysfunction was considered significant when a 50% reduction in stroke volume was observed. Bretylium (50 mg), pronestyl (600 mg), and lidocaine (160 mg/hour) were administered to prevent arrhythmias. The CCD was subsequently placed over the heart so that the deflated balloon was sandwiched between the RV free wall and Teflon backing. The device was fixed in place by suturing the Teflon backing to the edges of the opened pericardium. The sternum was then reapproximated in preparation for CCD activation. Balloon inflation was timed to begin on detection of the QRS complex, whereas the deflation point was adjusted manually to coincide with cardiac diastole (on the basis of RV and balloon pressure waveforms).
Hemodynamic data in pigs were collected under the following conditions: (1) before RV infarction (baseline), (2) after RV infarction and balloon placement (postinfarct), and (3) after balloon placement with the device inflated with every heartbeat (balloon on). CCD driveline pressures were also recorded during intervals of cardiac assist. Right ventricular copulsation was continued for periods up to 30 minutes. Upon completion of the study, all animals were euthanized with 20 meq KCl while remaining under general anesthesia.
Experimental Procedure in Dogs
The canine model was chosen for follow-up experiments wherein device function would be tested for longer periods in a setting of dilated cardiomyopathy. Eight coonhounds of mixed variety (25–30 Kg) were used as an alternative to pigs based on prior experience with rapid ventricular pacing in this animal species. Each dog was sedated with intramuscular injections of Acepromazine (10 mg) and Atropine (0.4 mg) before induction of general anesthesia. Anesthesia was maintained with 1–2% isoflurane delivered through an endotracheal tube. A Swan-Ganz catheter and arterial pressure line were placed via the external jugular vein and carotid artery, respectively. The chest was entered through a left anterior thoracotomy, and the pericardium was opened. A single epicardial pacing lead was placed on the left ventricle and connected to a single chamber implantable pacemaker (Prevail 8086, Medtronic, Minnepolis, MN) to induce heart failure via rapid pacing. 6,7 Implantable flow probes (model 24N, Transonic Systems, Inc., Ithaca, NY) were secured around both the pulmonary artery and aorta for beat to beat stroke volume measurement, with the aortic probes acting primarily as backups in case of pulmonary artery (PA) probe failure. The silastic backing of the CCD was then sutured to the pericardium directly over the RV, with the balloon facing the heart and the backing material beneath the sternum.
The animals were allowed to recover for 2–3 days before activating the pacemaker. The heart was paced at a rate of 240 beats per minute (bpm) until moderate to severe heart failure was detected via transthoracic echocardiography—typically 2 to 3 weeks. Four of the dogs used in this study failed to complete the protocol as a result of sudden fatal complications caused by heart failure. The remaining dogs were again placed under general anesthesia so that the pneumatic drive line could be externalized and the device activated. As with the pig study, balloon inflation was timed to coincide with ventricular systole, and the deflation point was adjusted manually to match cardiac diastole.
Hemodynamic data in these dogs were collected at times analogous to the staged protocol used in the acute pig study, namely (1) before device placement (baseline), (2) after heart failure with the device inactive (heart failure), and (3) after heart failure with the device inflated with every heartbeat (balloon on). Device activation was continued for up to 4 hours. Upon completion of the study, all animals were euthanized with 20 meq KCl while under general anesthesia.
Data Collection and Statistical Analysis
In both experiments, data were collected under closed chest conditions to mimic physiologic conditions anticipated during clinical application of a catheter based CCD. Pressure and electrocardiogram waveforms were digitized at a rate of 100 samples/second for periods of 1–2 minutes and stored using an IBM 300PL PC and commercial data acquisition software (WinDAQ, Dataq Instruments, Akron, OH). Data sets were collected upon establishment of steady state hemodynamics for each condition tested (defined as stable arterial pressures maintained for 30 seconds or more). Paired Student's t-tests were performed using StatView5 software (version 5.0.1, SAS Institute, Inc., Cary, NC) to determine the significance of differences between treatment groups. Two-sided p values < 0.05 were considered significant. All summary data are expressed as means ± standard deviation.
In the pig studies, CCD activation produced marked increases in right heart pressures and flows, restoring these parameters to near normal levels (Table 1). RV and PA systolic pressures increased 69 and 56% respectively over postinfarct baseline, whereas RV diastolic pressures were lowered by 30%. Cardiac stroke volumes increased dramatically from 14.7 ± 1.9 to 37.8 ± 9.2 mL (p < 0.005), and mean PA pressures improved from 12.7 ± 2.4 to 15.8 ± 2.9 mm Hg (p < 0.01). Systolic arterial pressures were also markedly improved with balloon activation, increasing from 66.3 ± 11.6 to 78.8 ± 8.7 mm Hg (p < 0.02), but did not return to normal levels (103.7 ± 10.5 mm Hg). Heart rate remained unchanged from the postinfarct baseline rate of 76 bpm, and no significant cardiac arrhythmias were noted. Substantial displacement of the RV free wall toward the ventricular septum was observed via TEE during balloon inflation. Doppler measurements made across the tricuspid valve revealed that RV compression did not induce or exacerbate regurgitation.
In the dog studies, rapid ventricular pacing yielded substantially lower arterial blood pressures and a 30% reduction in overall stroke volume (p < 0.05), although right heart pressures remained unchanged from baseline values (Table 2). Unlike results obtained using the pig model, device inflation in dogs produced no appreciable change in hemodynamic parameters. Proper timing of balloon inflation was confirmed via transthoracic echocardiography and viewed directly on subsequent exploration of the chest cavity. Device position was adjusted several times in each experiment to ensure proper orientation over the RV, but no improvement in function was achieved. In each instance, echocardiographic studies revealed minimal RV free wall motion resulting from balloon inflation. Both intraoperative and postmortem exploration of the chest cavity in all four animals confirmed that the balloon had been placed over the RV free wall as intended. After the fourth such study yielded negative results, the series was discontinued pending further analysis to determine why successful device actuation proved ineffective in this experimental model.
The degree to which our understanding of biology, anatomy, and physiology has improved in recent decades is difficult to overstate, yet there remains a great deal to be learned regarding the intricacies that govern complex biologic systems. As a result, efforts to simulate the human body for biomedical research purposes have thus far produced few practical alternatives to animal experimentation, forcing researchers to continue using imperfect animal models to help develop and test new medical therapies. 8,9,10
Indeed, the choice of test species is often the key factor in determining how well a given experiment will presage clinical outcomes. Animal model selection is particularly important when evaluating implantable devices designed to operate in concert with anatomic structures, as is often the case with orthopedic, vascular, and cardiac prostheses. It is hard to imagine a better proof of this axiom than the evidence cited in this report, in which similar heart assist devices were tested in two distinct large animal models (dogs and pigs) with opposite outcomes.
Upon considering the characteristics that distinguish porcine from canine anatomy, the reasons behind these disparate results appear linked to the shape of the thoracic cage and the orientation of the heart within the chest cavity. Because the amount of cardiac compression achieved is purely a function of translation mechanics, differences in the way heart failure was produced (infarction versus rapid pacing) and the way the balloons were placed into position (sternotomy versus thoracotomy) were discounted as possible confounding factors in these analyses. Indeed, despite the use of identical balloon pumps placed in the same anatomic position and actuated under similar closed chest conditions, effective ventricular free wall displacement was plainly evident in all porcine studies and almost completely absent in dog experiments.
In the particular breed of dog used in these studies (coon-hounds), the depth of the chest cavity can be nearly double its width as measured at the level of the heart, 11 whereas the heart itself typically lies parallel to the sternum. Yorkshire pigs, however, generally exhibit a more circular chest wall profile 12 with the cardiac apex angled toward the sternum and slightly to the left, an arrangement more akin to human anatomy. Therefore, when lying in the dorsal recumbent position as dictated by the test protocol, these dogs displayed greater separation between the sternum and right ventricular free wall than pigs. Because the CCD relies on the sternum to provide a stable anchor point against which to expand toward the heart, cardiac compression via balloon expansion was much less effective in these dogs (Figure 2). It would appear, then, that the porcine model provides the better option for device testing in this instance and that the canine model should generally not be used for studies of this type.
Owing to the serendipitous nature of these findings, several caveats should be taken into account when interpreting their significance. First, it is important to recognize that, because the test protocols used in the two groups were not identical, factors other than anatomic differences may have contributed to the antipodal outcomes reported here. Second, because this device has never been tested in the clinical setting, it is impossible to say with certainty which of these two species better mimics the human situation with regard to device effectiveness. Finally, it should be noted that because chest shape varies considerably from breed to breed in dogs, it is likely that certain varieties would prove more applicable to humans than the one used here.
There is a growing recognition in the scientific community of the need to minimize animal use in medical research for reasons both practical and ethical. One important way to achieve this goal without risking human health is through the judicious use of animal models selected to mimic human physiology to the greatest possible extent. These studies clearly demonstrate how subtle anatomic differences among animals can influence experimental outcomes and lead to false conclusions regarding the potential efficacy of a given therapy. It is hoped that this report will increase awareness of the potential drawbacks associated with the rote application of established animal models to novel research applications and thereby encourage added diligence in the evaluation of candidate test species
The authors thank Drs. Chong Park, Deepak Singh, Leah Teekell-Taylor, and Uday Dasika for surgical assistance and Drs. Parvizi and Sunil Mankad for echocardiographic analyses of heart failure and device function. This work was supported by Allegheny-Singer Research Institute (Pittsburgh, PA) Grant 96–027–2P.