Introduction

Large animal research is a major component of preclinical testing and validation of new surgical techniques, medications and devices for the treatment of cardiovascular disease1,2,3,4. Swine are an excellent model for the study of cardiovascular disease due to their rapid growth, and anatomical and physiological similarities to human hearts and have thus become the favored large animal model for the study of cardiovascular disease5. Echocardiography is a non-invasive cardiac imaging modality that is easily accessible and has multiple advantages including real-time imaging, ease of transportation and lower cost compared to other modalities such as computed tomography or magnetic resonance imaging.

While the anatomy and physiology of the heart in swine is very similar to the human heart, the positioning of the heart relative to the other structures differs significantly between swine and humans. These differences are primarily due to the quadruped stance of swine as well as the unusual sternal structure of humans compared to other mammals. While mammals generally have laterally compressed thoraces, the human thorax is dorsoventrally compressed6. Due to these thoracic structural differences, the commonly used echocardiography views of humans are not optimal use in swine models. Echocardiography in the piglet has been shown to be feasible but there is currently no standardized transthoracic echocardiography (TTE) protocol7. Standardized landmarks are important to ensure reproducibility of echocardiographic measures. Our group is studying partial heart transplantation (PHT) using piglet models1,2,8,9, which involves monitoring of transplanted semilunar valves for growth and function over time, with piglet TTE being the most important surveillance technique for the transplanted valves. Normative data on semilunar valve growth rate over time in healthy piglets has not been reported in the literature and is important for studying and extrapolating the effects of therapies such as valve transplants in translational research.

The goal of this study was to: (1) create a standardized piglet TTE protocol with defined areas on the piglet’s chest wall to evaluate and measure semilunar valves; (2) report normal semilunar valve and left ventricular dimension growth rates in piglets.

Methods

Standardized piglet TTE protocol

This study used purpose-bred Yorkshire pigs (8–15 kg, 5–8 weeks old), obtained from the National Swine Resource and Research Center (Columbia, MO) and Oak Hill Genetics (Ewing, IL). All animal procedures were approved by the Institutional Animal Care and Use Committee of the University of Arkansas Medical Sciences (Protocol # IPROTO202300000160) and conformed to standard Office of Laboratory Animal Welfare on the housing and medical care of swine used in research. This study was conducted in compliance with relevant guidelines and regulations and is reported in accordance with the ARRIVE guidelines.

Our standardized protocol (Table 1; Figs. 1, 2, 3 and 4) was developed by a trained pediatric cardiologist and pediatric cardiac sonographer. Piglets were scanned under anesthesia using 5 MHz and 6 MHz probes on the Phillips EPIQ 7 ultrasound machine [Epiq7c Circular Edition] to obtain views like those acquired in human echocardiography. Landmarks were recorded for obtaining the views.

Fig. 1
figure 1

Subcostal view.

Full size image
Fig. 2
figure 2

Apical 4 chamber view.

Full size image
Fig. 3
figure 3

Parasternal long axis (PLAX) view.

Full size image

The details of housing animals in our facility between echocardiograms as well as anesthesia techniques are included in Supplemental material.

Study population and data collection

Serial TTEs were performed on eight piglets under anesthesia using our standardized echocardiography protocol (Table 1), each evaluated between 4 and 8 times from the approximate age of 4 weeks to 5 months at intervals of 1–2 weeks. Additional echocardiographic evaluation was performed for preoperative piglets (Table 2). Every session was performed under general anesthesia. At each imaging session, measurements of the aortic and pulmonary valve annuli were recorded in millimeters. The aortic valve was measured in end-systole from the parasternal long axis view (Fig. 4b), and the pulmonary valve was measured in end-systole and diastole from the parasternal short axis view (Fig. 3g). Ventricular cavity dimensions and wall thickness measurements were obtained using M-mode from the parasternal short axis view (Fig. 3b and c). Ventricular ejection fraction and shortening fraction were calculated using M-mode measurements. Simultaneously, age (in days), weight (in kilograms), and length (in centimeters), were collected on the day of the echocardiogram.

Fig. 4
figure 4

Parasternal short axis (PSAX) view.

Full size image

Statistical analysis

To characterize how valve size changes with growth, generalized additive mixed models (GAMM) were employed. To explore the association individually, we fit a model for each combination of two valves’ size (aortic and pulmonary valves measured in centimeters) and cardiac dimensions (left ventricular end diastolic dimension [LVEDD], left ventricular end systolic dimension [LVESD] and left ventricular posterior wall dimension in systole [PWs]) as response and the three predictors (age, weight and length), resulting in a total of 15 models. For each predictor \(\:X\), each response \(\:Y\) was modeled as:

$$\:Y\:=\:{\beta\:}_{0}+\:f\left(X\right)\:+Zu+\epsilon$$

where \(\:{{\upbeta\:}}_{0}\) is the intercept, \(\:f\) represents a penalized cubic regression spline function, \(\:Z\) is the known desgin matrix for the random effect, u is the random intercept subject to each piglet, and \(\epsilon\) is the error term.

In addition, a multivariable GAMM was also fitted to account for the combined effects of age, weight and length to each valve size and cardiac dimensions, which takes the following form:

$$\:Y\:=\:{\beta\:}_{0}+\sum\:_{i=1}^{3}{f}_{i}\left({X}_{i}\right)\:\:+Zu+\epsilon$$

where \(\:{X}_{i}\) is age, weight and length respectively and \(\:{f}_{i}\) is the corresponding penalized cubic spline function.

The use of penalized cubic spline relaxes the assumption on the actual relationship between our responses and predictors. The upper limits of knots are set to be 10 by default and the actual number of knots that reflects the flexibility is determined by the penalization process. Model-based marginal prediction and its 95% confidence band were used to visualize the growth for each combination of valves and predictors. Predictions were computed over the observed range of each predictor, from its minimum to maximum values in the data. P-values < 0.05 were considered statistically significant. For model fitting, we used the mgcv package in R. All data were analyzed using R version of 4.4.3.

Table 1 Standardized piglet transthoracic echocardiography protocol [RA: right atrium; LA: left atrium; RV: right ventricle; LV: left ventricle; aov: aortic valve; PV: pulmonary valve; MPA: main pulmonary artery; R: right; L: left].
Full size table
Table 2 Additional echocardiographic evaluation prior to cardiac surgery [RA: right atrium; LA: left atrium; RV: right ventricle; LV: left ventricle; aov: aortic valve; PV: pulmonary valve; MPA: main pulmonary artery; R: right; L: left].
Full size table

Results

A total of 59 readings from 8 piglets were included in the analysis, with a median of 8 readings per piglet (minimum: 4, maximum: 10). The overall ejection fraction has a mean of 53.8% (95% CI: [49.0%, 58.8%]) and the overall shortening fraction has a mean of 27.6% (95% CI: [24.6%, 30.6%]). The growth trends generated for valve size against age, weight and length are shown in Fig. 7, and those for cardiac dimensions are shown in Fig. 8. Full model summaries are provided in the Supplementary Material.

Fig. 5
figure 5

Patent ductus arteriosus (PDA) view.

Full size image
Fig. 6
figure 6

Bubble study.

Full size image

Univariate analysis showed age was the strongest variable predictor (adjusted R2 = 0.848, Fig. 7A) and length was the poorest predictor (adjusted R2 = 0.723, Fig. 7C) for pulmonary valve size. For aortic valve size, length was the strongest predictor (adjusted R2 = 0.892, Fig. 7F) and age was the poorest predictor (adjusted R2 = 0.876, Fig. 7D). All terms in the univariate models were significant. For cardiac dimensions, weight was the best predictor for LVEDD (adjusted R2 = 0.819, Fig. 8B) and PWs (adjusted R2 = 0.709, Fig. 8H), whereas age was the best predictor for LVESD (adjusted R2 = 0.638, Fig. 8D).

Fig. 7
figure 7

Generalized additive mixed models for pulmonary and aortic valve size in relation to age, weight and length. Solid lines denote marginal predictions from the generative additive mixed models. Pink and blue shaded areas denote 95% confidence interval. Circle points represent echocardiographic measurements, with dotted lines linking repeated observations from the same piglet to show within-subject trajectories.

Full size image
Fig. 8
figure 8

Generalized additive mixed models for left ventricular end diastolic dimension (LVEDD), left ventricular end systolic dimension (LVESD) and left ventricular posterior wall dimension in systole (PWs) in relation to age, weight and length. Solid lines denote marginal predictions from the generative additive mixed models. Green, orange and red shaded areas denote 95% confidence interval. Circle points represent echocardiographic measurements, with dotted lines linking repeated observations from the same piglet to show within-subject trajectories.

Full size image

The multivariate GAMM with age, weight and length outperform all the univariate models for both valve sizes and cardiac dimensions. The adjusted R² values were 0.877 for the pulmonary valve and 0.918 for the aortic valve, and 0.877, 0.918, and 0.918 for LVEDD, LVESD, and PWs, respectively. Weight was not statistically significant in the multivariable model for either the pulmonary or aortic valve when all three growth characteristics were included as predictors. For cardiac dimensions, weight was the only predictor significantly associated with LVEDD and PWs, whereas none of the growth characteristics were significantly associated with LVESD in the full model.

Discussion

This is the first description of a standardized piglet echocardiography protocol for the assessment of porcine semilunar valve growth over time as well as the first report of normative porcine semilunar valve growth trends. Piglet echocardiography has been reported to be feasible7 and has been used to assess ventricular function and pulmonary hypertension in porcine models10,11,1213. There were no complications in the 59 serial sedated echocardiograms performed for this study in control piglets. We have reported our emergency protocol for potential emergencies in the supplementary materials.

We found that to obtain views like parasternal short and long axis, we had to scan the piglet from the right side of the chest and have reported the best way to acquire the images in this protocol. We have used this protocol successfully to generate serial comprehensive echocardiographic assessments of dozens of piglets undergoing partial heart transplantation as well as several controls8,9. Preoperative assessment for presence of patent ductus arteriosus and intracardiac shunts (using a bubble study) is helpful for procedural planning prior to cardiac surgery.

Secondly, data on serial semilunar valve evaluation and normative semilunar valve growth trends was lacking prior to this study and would be important for studying cardiovascular models like partial heart transplantation in the piglet1,2,3,4. This study provides a useful protocol as well as normative data for all researchers conducting long term studies utilizing porcine donors as it allows for accurate and reproducible evaluation of the valves that are being studied.

Additionally, while prior work has reported feasibility of ventricular function assessment in piglets7,12 our data is the first to report serial interventricular dimensions over time and ventricular ejection fraction over time in piglets.

While this is a small sample, the repeated measurements over time allowed us to develop normal growth curves. This is important in advancing the piglet model specifically for pediatric cardiovascular disease as it will allow researchers interested in a variety of diseases and treatments to better estimate the valve growth and identify outliers within their cohort. This advancement could also prove useful to future xenotransplantation efforts as data on the growth and size of piglet cardiac features in relation to weight, age and body size will be useful for selecting donors with appropriate sizes for the desired recipient. Our data show that age is the strongest predictor of pulmonary valve growth and length is the strongest predictor of aortic valve growth in control piglets. In our experience, in piglets undergoing cardiac surgery that have poor post-operative growth, correlation with weight is the most useful. These growth trends serve as a comparison for our future experimental models.

Conclusions

We report our standardized piglet echocardiography protocol and normative semilunar valve and left ventricular growth trends by age, weight and length.

Institutional review board statement

The animal study protocol was conducted in accordance with the Institutional Animal Care and Use Committee of the University of Arkansas Medical Sciences (Protocol # IPROTO202300000160) and conformed to standard Office of Laboratory Animal Welfare on the housing and medical care of swine used in research. The study is reported in accordance with ARRIVE guidelines.