Isolated pulmonary regurgitation causes decreased right ventricular longitudinal function and compensatory increased septal pumping in a porcine model

Abstract Aim Longitudinal ventricular contraction is a parameter of cardiac performance with predictive power. Right ventricular (RV) longitudinal function is impaired in patients with free pulmonary regurgitation (PR) following corrective surgery for Tetralogy of Fallot (TOF). It remains unclear whether this is a consequence of the surgical repair, or whether it is inherent to PR. The aim of this study was to assess the relationship between longitudinal, lateral and septal pumping in a porcine model of isolated PR. Methods Piglets were divided into a control (n = 8) group and a treatment (n = 12) group, which received a stent in the pulmonary valve orifice, inducing PR. After 2–3 months, animals were subjected to cardiac magnetic resonance imaging. A subset of animals (n = 6) then underwent percutaneous pulmonary valve replacement (PPVR) with follow‐up 1 month later. Longitudinal, lateral and septal contributions to stroke volume (SV) were quantified by measuring volumetric displacements from end‐diastole to end‐systole in the cardiac short axis and long axis. Results PR resulted in a lower longitudinal contribution to RV stroke volume, compared to controls (60.0 ± 2.6% vs. 73.6 ± 3.8%; P = 0.012). Furthermore, a compensatory increase in septal contribution to RVSV was observed (11.0 ± 1.6% vs. −3.1 ± 1.5%; P < 0.0001). The left ventricle (LV) showed counter‐regulation with an increased longitudinal LVSV. Changes in RV longitudinal function were reversed by PPVR. Conclusion These findings suggest that PR contributes to decreased RV longitudinal function in the absence of scarring from cardiac surgery. Measurement of longitudinal RVSV may aid risk stratification and timing for interventional correction of PR in TOF patients.

concentric contractions of both the lateral wall and the septum in the short-axis planecontribution to cardiac pumping. We have demonstrated that on average, 60% of left ventricular pumping and 80% of right ventricular pumping are attributable to longitudinal contraction by means of AVP displacement in the healthy human heart. 1 These differences reflect an adaptation to the divergent ventricular and atrial filling processes and haemodynamic characteristics of the systemic and pulmonary circulation. 1 It has become apparent that the analysis of the longitudinal contractile function is of clinical relevance in the assessment and prediction of cardiac pathologies, such as myocardial infarction, dilated cardiomyopathy, congenital heart disease and cardiac arrhythmias. [2][3][4][5][6][7][8][9][10] Changes in longitudinal pumping can be observed before overall cardiac performance is compromised. 10 Recently, cardiac magnetic resonance imaging (CMR) has emerged as an accurate approach for assessment of longitudinal function and its predictive nature. 7 Thus, the characterization of the relationship between longitudinal and radial contribution to stroke volume has not only improved our understanding of cardiac physiology, but has also become an important clinical tool for better risk stratification in cardiac patients.
The pathophysiological changes underlying cardiac dysfunction in patients with Tetralogy of Fallot (TOF; pulmonary stenosis, overriding aorta, ventricular septal defect, right ventricular hypertrophy) have been the subject of ongoing investigation, as the optimal timing of follow-up surgeries in these patients remains to be clearly defined. 11 In particular, the timing and method for pulmonary valve replacement in patients with free pulmonary regurgitation (PR) after surgical repair using transannular patch remains controversial, as it requires a balance between maintaining right ventricular (RV) function and size on the one hand, and limiting the patient's exposure to multiple interventions throughout lifetime on the other. 11,12 The development of tools to further characterize cardiac function with the aim to aid in the clinical management of TOF patients with PR is therefore of importance and has been explored in several studies. For example, investigators observed reduced right ventricular AVP movement in patients with PR following corrective surgery for TOF. 13,14 Riesenkampff et al. recently reported impaired right atrial filling, which is functionally linked to AVP displacement, in TOF patients. 14 However, the changes in atrial function were not observed in non-TOF patients with isolated PR, that is patients who did not undergo open surgery, suggesting that the pathological cardiac parameters observed in TOF patients could be a by-product of corrective surgery, rather than being inherent to PR. 15 To address this controversy, the aim of this study was to investigate the effects of isolated PR on longitudinal and radial contributions to ventricular pumping in a porcine model. A better understanding of the pathophysiological changes underlying PR may lead to improved risk stratification and may aid in the clinical management of TOF patients with regard to timing of pulmonary valve replacement.

Calculation of pulmonary and tricuspid regurgitation
To quantify the degree of PR, phase-contrast imaging was conducted over the pulmonary artery. The average regurgitant fraction over the pulmonary valve was 42.9 AE 2.7% in the PR group (Table 1). Tricuspid regurgitation (TR) was observed in seven of 12 pigs in the PR group. In these pigs, TR amounted to 28.1 AE 6.47% (spread between 9.3% and 62.4%).

Correlation between corrected right ventricular ejection fraction and longitudinal left ventricular stroke volume
Animals with PR exhibited a correlation (R 2 = 0.623; P = 0.002) between the corrected RV ejection fraction (EF), that is net pulmonary forward flow/ RVEDV and the longitudinal contribution to LVSV (Fig. 4).

Summation of individual stroke volumes
As an internal quality control, the sum of calculated individual contributions to SV, that is longitudinal SV, lateral SV and septal SV, was compared to the SV obtained by delineation of the ventricles in end-diastole (ED) and end-systole (ES

Invasive pressure measurements
Invasive pressure measurements were conducted in seven of eight animals in the control group (except for pulmonary capillary wedge pressure (PCWP) measurements: 6/8), eight of 12 pigs in the PR group and six of six animals in the PPVR group. Data are summarized in Table 2.
Correlation between longitudinal contribution to right ventricular stroke volume and pulmonary capillary wedge pressure Animals in the PR group demonstrated a correlation between the longitudinal contribution to RVSV and mean PCWP (R 2 = 0.940; P < 0.001) (Fig. 5a). No correlation could be found between PCWP and the degree of pulmonary regurgitation (R 2 = 0.020; P = 0.738; Fig. 5b).

Interobserver variability
The average interobserver variability for both ventricular volumes combined was 0.6 AE 5.1% and for AVP measurements 4.9 AE 13.7% (data presented as mean AE SD).

Discussion
This study analyses the individual contributions of longitudinal, lateral and septal contraction to LVSV and RVSV in pigs suffering from isolated PR. The results suggest that decreased longitudinal function in the RV is a result of PR and corresponding RV volume overload and not only a consequence of cardiac surgery.
Severe PR is a common feature following corrective surgery for TOF and entails numerous long-term complications, including exercise impairment, ventricular dysfunction, arrhythmias and sudden cardiac death. 16,17 As long-term survival increases and increasing numbers of primarily operated children with TOF advance in age, strategies for the long-term management of PR in this population are being developed. The challenge hereby rests in preserving RV function, while keeping the number of corrective interventions for PR in the course of the patient's life at a minimum. To this end, CMR cut-off volumes for the RV are being proposed to determine the optimal time point of pulmonary valve replacement, although the exact volume threshold for intervention is still under debate. 18 The use of cut-off volumes as a factor supporting decision-making, however, does not encompass a functional perspective of the RV. This problem is highlighted by recent longitudinal observations, demonstrating that RV dilation does not necessarily progress in TOF patients and reaches a stable plateau. 19,20 Recently, it has become evident that assessment of AVP displacement or longitudinal contraction, whether by CMR or echocardiography, serves as a potent predictor of adverse events in a range of cardiac pathologies including TOF. 3,7,9 For example, it has been demonstrated that longitudinal LV dysfunction assessed by echocardiography is a predictor for severe arrhythmias and death in TOF patients. 3 Despite the lack of prospective data, it is also apparent that TOF patients with free PR suffer from  reduced longitudinal contribution to RVSV and exhibit pathological right atrial filling and a decreased right atrial reservoir function, as evidenced by recent insight by Riesenkampff et al. [13][14][15] The authors argue that decreased atrial reservoir function observed in TOF patients is a by-product of surgery, as evidenced by the lack of atrial abnormalities in non-TOF patients with isolated PR. 14,15 The data presented in this manuscript, however, support the idea that isolated PR also results in reduced longitudinal RV pumping and is therefore independent of surgery or scarring. Atrial filling and longitudinal ventricular pumping are closely intertwined processes. 1,2,14,[21][22][23] The displacement of the AVP during ventricular systole draws blood into the atria, similar to the piston aspirating liquid into a syringe. Given the interdependence between the reservoir function of the atria and the extent of displacement of the AVP in ventricular systole, the discrepancy between the observations of Riesenkampff et al. and this study remains to be elucidated. Alas Riesenkampff et al. 15 15 It thus remains to be clarified whether patients with close to normal RVEDV and isolated PR harbour a latent atrial or longitudinal pumping dysfunction that may be uncovered as the enlargement of the RV progresses.
The degree of reduction in longitudinal RVSV in this porcine model is consistent, albeit to a lesser extent, with our observations made in TOF patients with PR. 13 While reduced RV global longitudinal strain has recently been described in an echocardiographical study in TOF patients with PR, it should be noted that no reduction in RV strain or tricuspid annular plane systolic excursion (TAPSE) has been noted in this  animal model of free PR previously. [24][25][26] While TAPSE is an adequate measure of AVP excursion, this study assessed the volumetric contribution of longitudinal pumping to overall stroke volume. Hence, TAPSE may be unaffected, while longitudinal pumping from a volumetric perspective may well display alterations. A similar rationale may be applied to strain measurements. We have previously postulated that the reduction in RV longitudinal pumping may be an adaptive mechanism with the aim of more adequately matching right atrial (RA) inflow to net RVSV, that is SV minus regurgitant volume, as a preserved longitudinal function would result in overfilling of the atria and a concomitant pendulum volume. 13 The observations made in this study further substantiate this hypothesis.
We have also observed a correlation between the degree of RV longitudinal pumping and PCWP (Fig. 5a). Although conclusions with regard to this effect remain speculative, we hypothesize that the mechanisms for this observation are an underfilling of the LV in this model of PR. The LV demonstrates signs of underfilling, as suggested by a low LVSV. Low PCWP is regarded as a proxy for left atrial pressure, and thus is directly coupled to LV filling. The fact that there was no correlation between the degree of PR and PCWP, but a very strong correlation between RV longitudinal pumping and PCWP, indicates that RV longitudinal pumping has added value as a functional parameter over the degree of valvular lesion.
To compare ventricular volumes between different experimental groups, we have indexed ventricular volumes as fractions of THV. As THV remains almost constant during the cardiac cycle and the four chambers of the heart are volumetrically coupled, THV can be used as a volumetric control accounting for different animal growth rates. 27 We observed a marked increase in RVEDVi and RVESVi due to RV volume overload. Both parameters reversed to normal levels after restoration of valve function. Conversely, LVEDVi and LVESVi decreased during conditions of PR, most likely reflecting LV underfilling. This is further substantiated by the pronounced paradoxical spetal movement, resulting in a septal contribution to RVSV (Fig. 2). This observation is in line with previous findings in TOF patients. 13 Paradoxical septal movement is a common sign of RV volume overload. 28 It has been suggested that the diastolic bulging of the septum into the LV due to RV distension causes a decrease in LVEDV and therefore results in an adverse functional coupling of both ventricles as RV overload progresses. 29 A reduction in LVEDV has been reported in TOF patients. 13 Interestingly, paradoxical septal movement has also been observed in TOF patients without PR and RV overload. However, this abnormality is attributable to changes in excitation propagation, specifically a right bundle branch block. 30 While paradoxical septal movement in TOF patients with PR is therefore most likely a combination of both RV overload and abnormal conductivity, we attribute the observed paradoxical septal movement in this model primarily to volume overload.
As septal contraction contributes to LVSV in the healthy heart, paradoxical septal movement will negatively affect LVSV. We have previously observed an increase in lateral contribution to LVSV in TOF patients to compensate for the loss of septal contribution to LVSV. 13 In contrast to our previous findings, we here report a compensatory increase in longitudinal and lateral contribution to LVSV. This discrepancy may be attributable to a more acute adaptation to RV overload (2-3 months) compared to TOF patients, where adaptation evolves over several years, or may alternatively be unique to the porcine model (see Limitations). Ultimately, the underlying cause of this difference cannot be conclusively elucidated at this point; however, the correlation between the amount of right ventricular impairment and the degree to which the left ventricle increases its longitudinal contribution to SV may point towards a functional importance of this adaptive mechanism in this model where there is an acute onset of PR. A potential explanation for this observation may be found in the physiology of systolic left atrial filling, which is primarily driven by ventricular longitudinal pumping. With progressing right ventricular functional impairment, the left ventricle may support left atrial filling by increasing its longitudinal contribution to SV, with the overall aim of maintaining energetically favourable constant total heart volume. 27 This hypothesis would also be in line with the observed indicators of impaired LV filling, such as low LVSV. Alternatively, the increase in longitudinal pumping may simply reflect a compensation for the loss of septal contribution to SV. This is supported by previous observations in patients with pulmonary arterial hypertension (PAH). These patients exhibit LV underfilling and an intact septal contribution to LVSV. 31 Contrarily to our observation, LV longitudinal pumping in PAH is, however, reduced, suggesting that the increased longitudinal contribution to LVSV in PR may in part be a compensation for lost septal SV due to paradoxical septal movement. 31 Despite our speculative understanding of the pathophysiological mechanisms underlying LV pumping adaptation to PR, it is worth stressing the fundamental observation that the LV has the capacity to adapt its pumping mechanics to RV impairment and isolated loss of LV function, that is septal pumping, and maintain effective SV in the process.
This study further demonstrates a general reversibility of the changes in cardiac pumping induced by PR and allows speculation about the nature of the adaptive mechanism. The observed changes may be a result of cardiac remodelling or alternatively a purely mechanical response to altered volume/pressure relationships. While this study cannot conclusively delineate between both mechanisms, it can be noted that 1 month of restored valve function, which in turn can be considered a short to medium time period for reversal of remodelling effects, almost entirely normalized cardiac pumping. Previous histological studies in this animal model further demonstrate that the muscle hypertrophy resulting from PR is still present 1 month after valve insertion. 32 We therefore speculate that changes in longitudinal pumping are mostly mechanic, rather than structural adaptations; however, more acute measurements after restoration of valve function will be necessary to further elucidate this question.

Study limitations
We observed TR in seven of 12 pigs with PR. It can be assumed that TR manifests in these cases secondarily to PR-induced RV distension. Subgroup analysis between TR vs. non-TR animals revealed a statistical difference in RVESVi in the PR group (TI: RVESVi = 11.7 AE 0.6%; non-TI: 14.0 AE 0.6%; P = 0.026). We attribute this difference to the need for generation of a larger SV during conditions of TR, in order to maintain a constant net pulmonary forward SV. While other isolated effects of TR could not be identified, an influence on pumping mechanics cannot be conclusively excluded due to the small number of animals available for subgroup analysis.
The acute onset of free PR in the present study reflects the clinical situation when using a transannular patch in TOF repair, and the degree of PR in the animal model is generally comparable to the TOF patient population (e.g. 39 AE 3% in 13 vs. 42.9 AE 2.7% in this study). However, in this animal model, the RV was prior to the intervention not pressure overloaded compared to pressure overload occurring in TOF cases. Also the limited duration (2-3 months) was much shorter than the decades often seen in humans with free PR. However, these few months represent a growth from piglet to adult pigs and, furthermore, caused a more pronounced increase in RV volume (factor 2.3 AE 0.3 compared to LVEDV) than typically occurring in patients (e.g. factor 1.7). This change may therefore evoke different adaptive mechanisms than in the clinical context, where PR and RV volume overload evolves gradually and persists over years. 33 A further limitation of this study is that not all animals have been exposed to the same duration of PR (2 or 3 months). This is attributable to the fact that the original experimental protocol aimed at answering a different scientific question, namely the effectiveness of PPVR to rescue different degrees of RV dysfunction after induction of PR. 34 In this analysis, the subgroups were pooled to increase sample size. Furthermore, it has been documented that the largest dynamic increase in parameters, such as RVEDV, occurs within the first month of PR, implying that functional differences between 2 and 3 months may not be as pronounced compared to earlier time points. 34 While insertion of a stent is an effective way of inducing isolated PR an influence of the stent on AVP movement or other pumping mechanics cannot be excluded.

Conclusions
This study underlines the value of characterizing the individual contributions to SV, as functional impairments, particularly in longitudinal pumping, are present before more global parameters, such as EF, are affected. While not offering prospective information, this study shows the value of longitudinal contractile function as a routine marker in the assessment of RV function, particularly in TOF patients. Measurement of longitudinal RVSV may support risk stratification and facilitate timing for interventional correction of PR. Furthermore, this study demonstrates that the observed changes are inherent to PR and therefore occur independently of previous corrective surgery.

Animals
The design of the experimental study has been previously published. 24,32,34 In short, piglets (cross-breeds of Danish Landraces, Yorkshire and Duroc) were divided into control (n = 8) and treatment (n = 12) groups. At baseline, piglets were 8-9 weeks old. Animals in the treatment group received a bare metal stent in the pulmonary valve orifice via percutaneous right heart catheterization, inducing free PR, while animals in the control group underwent sham interventions. PR was maintained for 2 or 3 months, before re-examination. A subset of animals in the treatment group (n = 6) underwent PPVR after 2-3 months of PR and was subjected to re-examination 1 month after valve repair. The animal study was approved by the Animal Experiments Inspectorate

Interventions
Before intervention, animals received sedation with midazolam 0.5 mg/kg. Subsequently, anaesthesia was induced and maintained with intramuscular injection of a mixture of tiletamine, zolazepam, xylazine, ketamine and methadone. Animals breathed spontaneously (40% O 2 ); however, intubation was performed to secure the airways. During intervention, amiodarone 5 mg/kg was administered intravenously to reduce occurrence of cardiac arrhythmias. Heparin was given as an anticoagulant (start dose 100 IU/kg, followed by 50 IU/kg/30 min maintenance). Percutaneous right heart catheterization was performed through the femoral vein. Following angiography (Omnipaque; GE Health Care, Piscataway, NJ, USA), a bare CP â stent (NuMED, Hopkinton, NY, USA) mounted on a 22-mm BIB â dilation catheter (NuMED) was advanced to the pulmonary valvular annulus inducing free PR. Invasive pressure measurements (RA, RV, pulmonary artery (PA) and PCWP) were conducted during catheterization using a fluid-filled transducer. For animals in the PPVR group, a Medtronic Melody â valve (Medtronic, Minneapolis, MN, USA) was placed inside the CP â stent following 2-3 months of free PR. Adequate valve function was confirmed with pulmonary angiography and echocardiography.

Cardiac magnetic resonance imaging
Cardiac magnetic resonance imaging was conducted on a 1.5T scanner (Siemens MAGNETOM â Vision, Erlangen, Germany), equipped with a standard phased array chest coil. Imaging was performed after two (n = 7) and three (n = 5) months of PR. Control animals (n = 8) were imaged at corresponding time points. For animals in the PPVR group (2 months of PR: n = 4; 3 months of PR: n = 2), imaging was repeated 1 month after valve insertion. Short-and long-axis images were acquired with a cinematographic gradient echo pulse sequence (temporal resolution: 30 ms; slice thickness: 6 mm; matrix size: 256 9 256). To calculate pulmonary regurgitation volume, velocity-encoded phase-contrast cine images were acquired over the main pulmonary artery (temporal resolution: 24 ms; slice thickness: 6 mm; matrix size: 256 9 256, VENC 250 cm/s).

Image analysis
Image analysis was performed using the image analysis software Segment (v2.0, R5100; Medviso, www. medviso.com). 35 LV and RV volumes were calculated by manual delineation of endocardial borders in all short-axis slices in end-diastole (ED) and end-systole (ES). To allow for intergroup comparison, ventricular volumes were indexed to the respective THV and (c) (d) Figure 6 Long-axis view of the heart. Panels (a) and (b) show the heart of a control animal in end-diastole (a) and end-systole (b) respectively. Dashed line represents the AVP in end-diastole. Dotted line represents the AVP in end-systole. Panels c and d depict the heart of an animal suffering from PR (3 months) at corresponding imaging time points. Note the increased RV volume and the decreased RV AVP movement with a compensatory increase in LV AVP movement. expressed as percentages thereof. The individual contributions to ventricular SV, that is the volumes displaced by longitudinal, lateral and septal contraction (lateral and septal contributions can further be consolidated under the term radial contraction), were calculated, as described previously. 1,2 In brief, for calculation of longitudinal SV, the position of the AVP was determined in ED and ES by on average seven measurement points in long-axis views. Representative examples and corresponding AVP displacement are shown in Figure 6. The displacement of the AVP from ED to ES was multiplied by the average of the three largest areas enclosed by the ventricular epicardial short-axis contours, yielding the longitudinal contribution to SV. For the rationale underlying this method, please refer to the studies of Carlsson et al. 1,2 By tracing the epicardial contours and defining the ventricular septum, the volumes generated by lateral and septal displacement between ED and ES were calculated and their corresponding contribution to SV was determined. 1,2,13 PR was calculated as the percentage of retrograde blood flow of total forward flow in the pulmonary artery. Presence of TR was determined by visual signs in the MR images, such as visible flow artefacts over the tricuspid valve during ventricular systole, and quantified using the discrepancy between pulmonary forward flow volumes and delineated ventricular stroke volumes.
Interobserver variability was calculated by blinded comparison of ventricular volumes and AVP measurements in ten data sets between the authors S.K. and S.S.S.

Statistical analysis
Data are presented as means AE standard error of the mean (SEM). Welch's unpaired t-test was used to test for differences in means between two populations. A paired t-test was used to compare individual differences before and after PPVR in the PPVR subset. Interobserver variability was calculated as the mean bias AE standard deviation. Statistical significance was assumed at results with a P-value <0.05.