Hemodynamic effects of a dielectric elastomer augmented aorta on aortic wave intensity: An in-vivo study

Dielectric elastomer actuator augmented aorta (DEA) represents a novel approach with high potential for assisting a failing heart. The soft tubular device replaces a section of the aorta and increases its diameter when activated. The hemodynamic interaction between the DEA and the left ventricle (LV) has not been investigated with wave intensity (WI) analysis before. The objective of this study is to investigate the hemodynamic effects of the DEA on the aortic WI pattern. WI was calculated from aortic pressure and flow measured in-vivo in the descending aorta of two pigs implanted with DEAs. The DEAs were tested for different actuation phase shifts (PS). The DEA generated two decompression waves (traveling upstream and downstream of the device) at activation followed by two compression waves at deactivation. Depending on the PS, the end-diastolic pressure (EDP) decreased by 7% (or increased by 5 – 6%). The average early diastolic pressure augmentation ( P dia ) increased by 2% (or decreased by 2 – 3%). The hydraulic work ( W H ) measured in the aorta decreased by 2% (or increased by 5%). The DEA-generated waves interfered with the LV-generated waves, and the timing of the waves affected the hemodynamic effect of the device. For the best actuation timing the upstream decompression wave arrived just before aortic valve opening and the upstream compression wave arrived just before aortic valve closure leading to a decreased EDP, an increased P dia and a reduced. W H .


Introduction
Aortic counterpulsation devices (ACPs) are used to assist and unload the cardiovascular system especially in high-risk patients with acute cardiovascular disease (De Waha et al., 2014;Santa-Cruz et al., 2006).The ACP concept is to reduce the left ventricular afterload to unload the left ventricle (LV) and augment the early diastolic pressure to enhance coronary flow.ACPs can be either intra-aortic (De Waha et al., 2014;Santa-Cruz et al., 2006), para-aortic (Lu et al., 2011) or extra-aortic (Campos Arias et al., 2017;Davies et al., 2005).The intra-aortic balloon pump (IABP) is the most widely used ACP.However, IABP patients may suffer from serious complications (De Waha et al., 2014;Khan et al., 2023;Santa-Cruz et al., 2006), and pneumatic actuation might lead to unpredictable actuation delays (Bruti et al., 2015).
An augmented aorta based on a dielectric elastomer actuator (DEA) is a soft tubular cardiac assist device with purely electrical actuation (Almanza et al., 2021) suggested for ACP (Martinez et al., 2022).DEAs consist of a dielectric membrane placed between compliant electrodes.By applying a voltage between the electrodes, the dielectric membrane is compressed in the thickness direction and expanding in all other directions (Pelrine et al., 2000).The soft DEA (distensibility: 2⋅10 − 3 mmHg − 1 , pulse wave velocity (PWV): 7.9 m/s) replaces an aortic section (distensibility: 2.5⋅10 − 3 − 4.5⋅10 − 3 mmHg − 1 , PVW: 3.5 − 11 m/s (Redheuil et al., 2010)) and actively increases and decreases its diameter in response to voltage.The change in diameter is a result of the applied voltage and the internal pressure in the tube (Martinez et al., 2022).The purely electrical actuation, the lightweight, and the nonobstructive nature of the DEA are advantages compared to current ACPs.It might be possible to fully implant and power the DEA with trans-cutaneous wireless power transfer (Knecht et al., 2015).The nonobstructive DEA could be actuated at any time during the heart cycle to optimize the assistance.A detailed comparison of the DEA with other ACPs can be found in (Martinez et al., 2022).
Wave intensity analysis (WIA) uses the simultaneous measurement of pressure and flow velocity to study arterial wave propagation (Khir et al., 2001b;Parker, 2009).The wave intensity (WI) pattern in arteries typically consist of two forward traveling waves: one compression wave (FCW) at start of systole (created by blood ejection) and one decompression wave (FDW) at end of systole (created by blood flow deceleration).WIA has been used to study arterial hemodynamics (Sun et al., 2000) and the effect of cardiovascular diseases, surgery (Khir et al., 2001a;Sugawara et al., 2009), and drugs (Penny et al., 2008), and to analyze the impact and performance of ACPs (Campos Arias et al., 2017;Kolyva et al., 2009;Lu et al., 2012) and pulsatile ventricular assist devices (PVADs) (Jahren et al., 2014;Khir et al., 2006).Kolyva et al., 2009 identified two pairs of waves caused by the IABP.The inflation resulted in a backward traveling compression wave (BCW) and its reflected FCW which correlates with the pressure augmentation.The deflation resulted in a backward traveling decompression wave (BDW) and its reflected FDW which correlates with the decrease of end-diastolic pressure (EDP).
For PVADs (Jahren et al., 2014;Khir et al., 2006) and extra-aortic ACP devices (Campos Arias et al., 2017) similar waves have been reported.The hemodynamic effect of ACPs depends on the device synchronization and the interference between the device-generated and ventriculargenerated waves (Sugawara et al., 2009).
The hemodynamic effect of DEAs has not been analyzed with WIA before.In our previous study (Martinez et al., 2022), the DEA was tested as an ACP in-vivo and for actuation at different time points (phase shifts (PS)) through the heart cycle.The best ACP result was achieved with an activation just before the start of systole and deactivation just after systole.The aim of the current study is to investigate the hemodynamic impact of the different PS on the WI.We hypothesize that the DEAgenerated waves interfere with the ventricular-generated waves and that the wave's timing can be beneficial or detrimental to the device's performance.

Dielectric elastomer actuator.
The DEA is a soft multilayer tube that replaces a section of the descending aorta.The positive electrode is encapsulated between two negative electrodes to confine the electrical field within the device, and the electrical connections are insulated to support voltages higher than 10 kV.The DEA is activated by applying a high voltage (<7kV).The DEAs are 40 or 50 mm long with 20 mm inner diameter.For detailed manufacturing process and geometry see (Martinez et al., 2022).Fig. 1A shows the DEA in activated (ON) and deactivated (OFF) states.When activated, the DEA diameter increases and sucks blood into the device, decreasing the pressure upstream and downstream of the device.When deactivated, the DEA diameter decreases and pushes blood out of the device, increasing the pressure upstream and downstream of the device.Fig. 1B shows the activation voltage signal used in this study.
Animals.This study has been approved by the Commission of Animal Experimentation of the Canton of Bern, Switzerland (Approval number BE14/2021).DEAs were implanted in five pigs (~50 kg) via left-sided thoracotomy.A section of the descending aorta was excised, and the DEA was anastomosed to the descending aorta.The heart rate was paced at 100 bpm using temporary pacemaker leads attached to the right atrium (Martinez et al., 2022).The first two animals were needed for training and tuning of the measurement protocol, while animal three died before any DEA could be tested.The final protocols and instrumentation were available in the last two animals.
Instrumentation.All recorded signals were acquired using Powerlab (ADInstruments, Houston, USA) and logged continuously (sampling frequency 1000 Hz) using LabChart Pro (ADInstruments, Houston, USA).The measured signals (see Fig. 1A) were LV pressure and volume (pressure-volume catheter, Millar, ADInstruments, Houston, USA), aortic pressure upstream and inside of the DEA (Xtrans, CODAN Pvb Critical Care GmbH, Forstinning, Germany), aortic flow upstream and downstream of the DEA (COnfidence, Transonic Systems, Inc., Ithaca, NY, USA), the pacemaker signal, and the DEA actuation voltage.The DEA was actuated using a high voltage unit (Trek 20/20C, Advanced Energy, Denver, USA) and a CompactDAQ (National Instruments, Austin, USA) connected to a computer with LabVIEW (National Instruments, Austin, USA).
Device synchronization.The peak of the pacemaker signal was used as actuation trigger; the computer-generated actuation signal was sent to the high voltage unit (Fig. 1B) and the DEA was actuated with welldefined delays after the trigger.The aortic valve opening (mid-point of LV pressure increase) was used as reference point for start of DEA actuation (defined as 0% of heart cycle).The actuation length (Fig. 1B) was set to the systolic length, measured as the time between aortic valve opening and closure (aortic flow-equivalent to dicrotic notch), and the rising and falling times were fixed at 50 ms.The time of aortic valve opening, and the systolic length were assumed to stay constant in each animal during pacing and were measured before the start of the protocols.See previous publication for further detail (Martinez et al., 2022).
Actuation protocols.The DEA actuation was delayed, or phase shifted (PS), compared to the aortic valve opening.The PS was varied from 0 to 90% in steps of 10% relative to the heart cycle length, giving a total of 10 Fig. 1.A: Schematic of the DEA implantation in the descending aorta and the instrumentation (modified from (Martinez et al., 2022)).The DEA increases its internal diameter when activated (DEA ON) and returns to its passive state when deactivated (DEA OFF).B: The actuation voltage profile of the DEA.
PSs (Fig. 2A).To compensate for the time delay between the systolic pressure increase in the DEA compared to the LV (due to the distance), the DEA was actuated earlier (delay correction, Fig. 2B) to give time to the waves generated by the DEA to reach the aortic valve at the desired synchronization point.This delay was measured as the time between the mid-point of LV pressure increase and the foot of the pressure increase inside the DEA.
Each PS was maintained for 60 heart cycles.Baseline measurements (DEA off) were performed before and after each PS for at least 60 heart cycles.The actuation voltage was increased to 6-7 kV in steps of 0.5 kV leading to larger DEA deformation.The highest reached voltage level (before failing) was used in the analysis.Failing DEAs were replaced such that several DEAs were tested in each animal.Table 1 gives an overview of all animals, protocols, DEAs, and voltage levels used in the analysis.
Data processing and analysis.For each measurement, a stable interval consisting of 20 consecutive heart cycles (no arrhythmias, and stable pressure values) with DEA actuation and 20 consecutive baseline heart cycles were chosen.Hemodynamic parameters were calculated for each included heart cycle as well as the mean and standard deviation of each parameter.A single averaged heart cycle was calculated for the DEAactuated and the baseline heart cycles.
The total WI was calculated as (Parker, 2009): where I [W/m 2 ] is the total WI, P [Pa] is the aortic pressure, U [m/s] is the aortic blood flow velocity.dP and dU are the differences between two successive samples.U was estimated from the aortic flow (Q ao ) by assuming an inner aortic diameter of 10 mm with U = Q ao /A, where A ([m 2 ]) is the aortic cross section.The P and Q ao were assumed to be measured at the exact same location (∼ 5 cmupstream of the DEA).Therefore, the flow and pressure were synchronized using the onset (foot) of the systolic pressure and flow increase at baseline.The WI was separated into its forward (+) and backward (− ) components: where ρ [kg/m 3 ] is the blood density (1060 kg/m 3 ) and c [m/s] is the wave speed.ρc was estimated from the linear slope of the P-U-loop at early systole (Khir et al., 2001b;Parker, 2009) at baseline.The forward and backward components were further decomposed into compression and decompression waves according to (Parker, 2009): Forward traveling waves means that dI > 0. A FCW corresponds to dP > 0 and dU > 0. A FDW corresponds to dP < 0 and dU < 0. Backward traveling waves means that dI < 0. A BCW corresponds to dP > 0 and dU < 0. A BDW corresponds to dP < 0 and dU > 0.
The EDP and the mean early diastolic pressure (P dia ) (calculated from end of systole to 150 ms before start of next systole) were calculated as measures for ACP efficiency.The hydraulic work (W H [J]) in the aorta done by the heart (estimate of LV stroke work) was estimated by (Khir et al., 2001a;Kolyva et al., 2009): where T is the period of the heart cycle.
Statistics: The Wilcoxon rank sum test was used to assess if the DEA actuated heart cycles were significantly to baseline for all parameters.A p < 0.05 was considered significant.

Results
The hemodynamic conditions varied during the experiments and among animals.Average mean arterial pressure was 63 ± 6 mmHg (52 − 72 mmHg).Peak systolic pressure was 78 ± 6 mmHg (68 − 89 mmHg).Wave speed was 1.07 ± 0.4 m/s (0.6 − 1.9 m/s).Fig. 3A shows the averaged WI pattern for baseline and the total WI.The LV generated one FCW LV and one FDW LV , which were reflected (at the DEA) as BCW LV and BDW LV .Fig. 3B shows the corresponding pressure and velocity.
Fig. 3C and E show the averaged WI pattern seen for 0% and 50% PS compared to baseline.Fig. 3D and F show the corresponding pressure and velocity (see Supplementary Figs. 4 and 5 for consecutive beat to beat analysis).Upstream, the DEA generated a BDW DEA at start of actuation and a BCW DEA at the end of actuation traveling towards the heart.Downstream, forward waves were traveling towards the periphery (not further discussed in this study).The backward traveling waves reached the aortic valve and were reflected as FDW DEA and FCW DEA .At 0% PS (Fig. 3C), the reflected FDW DEA was just ahead of aortic valve opening and the FCW LV .The FCW LV was larger and earlier than baseline.The reflected FCW DEA arrived at end of systole and was partly superimposed to the FDW LV .The FDW LV was smaller and earlier than baseline.At 50% PS (Fig. 3E), the reflected FDW DEA arrived at end of systole and was superimposed to the FDW LV .The FDW LV was larger and later than baseline.The reflected FCW DEA arrived just before aortic valve opening and the FCW LV .The FCW LV was smaller and later than baseline.
Fig. 4 shows the WI pattern for baseline and the different PSs.The DEA-generated waves shift to the right for increasing PS and impact the LV-generated waves and the total WI pattern.Table 2 shows the average values, standard deviation, and significance of the FCW LV 's and the FDW LV 's peak values.The FCW LV increased (p < 0.05) for 0, 20, 40, 60, and 80% PS with the largest increase at 60% PS, and decreased for 10, 30, and 50% PS with the largest decrease at 10% PS.The FDW LV increased (p < 0.05) for 10, 50, 60, and 90% PS with the largest increase at 10% PS, and decreased for 0 and 40% PS with the largest decrease at 0% PS.Fig. 5 shows the average values for W H , P dia , and EDP (average values, standard deviation and significance in Table 2).The W H decreased for 0-20% and 80-90% PS (p < 0.05) and increased for 50-70% (p < 0.05).The P dia increased for 0-20% PS (p < 0.05) and decreased for 30-70% PS (p < 0.05).The EDP decreased for 0% and 70-90% (p < 0.05) and increased for 20-50% PS (p < 0.05).The strongest ACP effect was seen in animal 4 DEA 2 (yellow line in Fig. 5) with augmentation of P dia of 2% and decrease in EDP of 5-7%.

Discussion
In this study, wave intensity analysis (WIA) is performed for the first time to assess the hemodynamic effects of a soft, tubular dielectric elastomer actuator augmented aorta (DEA) implanted in porcine models (n = 2).DEAs were positioned in the descending aorta and actuated with ten different phase shifts (PS) in steps of 10%.
Actuating the DEA as an aortic counterpulsation (ACP) device (activation just before aortic valve opening (increase in diameter) and deactivation just after aortic valve closure (decrease in diameter)) gave the best overall results in terms of lower end-diastolic pressure (EDP) and augmented early diastolic pressure.This is in line with the optimal operation of intra-aortic balloon pumps (IABPs) (De Waha et al., 2014;Santa-Cruz et al., 2006).In comparison with IABPs, the non-obstructive DEAs can be actuated through the whole heart cycle without obstructing the flow.Moreover, the actuation time is much faster in a DEA (<20 ms (Martinez et al., 2022), current study 50 ms) than in IABPs (~200 ms), and actuation time is not affected by device angulation as in IABP (Bruti et al., 2015).This allows tuning of the DEA actuation timing to explore and optimize the assistance.WIA gives insight into the coupling between the device and the cardiovascular system and the interference between Fig. 3. Wave intensity (WI) pattern for baseline (A), DEA activation with phaseshift (PS) 0% (C) and 50% (E) seen in animal 4 DEA 2 at 7 kV for the averaged heart cycle decomposed into forward and backward compression and decompression, and baseline total WI.The corresponding pressure and velocity (B, D, F) waveforms for the averaged heart cycle.The WI is normalized to the baseline peak value of the forward compression wave of the left ventricle (FCW LV ).The vertical dashed lines mark the start of systole (onset of FCW LV in baseline at the measurement site), and the end of systole (end of FDW LV in baseline at the measurement site), respectively.FDW: forward decompression wave, FCW: forward compression wave, BDW: backward decompression wave, BCW: backward compression wave, LV: left ventricle.DEA: dielectric elastomer actuator, WI: wave intensity.

Table 2
Average values and standard deviation of the peak values of the left ventricular forward compression and decompression waves, hydraulic work, mean early diastolic pressure and EDP in % change compared to baseline for the different phase shifts for all animals and DEAs.waves.In the present study, we hypothesized that the DEA actuation timing changes the wave intensity (WI) pattern, and we used WIA to assess whether the timing of the DEA-generated waves can be beneficial or detrimental for the cardiovascular system.The baseline WI (Fig. 3A) consisted of one forward compression wave (FCW) and one forward decompression wave (FDW) generated by the LV in line with (Khir et al., 2001b;Parker, 2009).The FCW LV 's peak value has been linked to cardiac contractility, with a larger peak value indicating higher LV contractility (Sugawara et al., 2009).The physiological meaning of the FDW LV is more complex.A decrease in the peak value can be due to a decline in the cardiac muscle tension bearing ability or to a change in the inertial force of the aortic blood flow, or both (Sugawara et al., 2009).The DEA introduces two pairs of new waves (Fig. 3C and E): one backward decompression wave (BDW DEA ) together with the corresponding reflected FDW DEA at activation and one backward compression wave (BCW DEA ) together with the corresponding reflected FCW DEA at deactivation (The reflection coefficient at the aortic valve seems to be positive for all PS).These DEA-generated waves are similar to waves generated by other ACP devices and pulsatile ventricular assist devices (Campos Arias et al., 2017;Jahren et al., 2014;Khir et al., 2006;Kolyva et al., 2009;Lu et al., 2012).However, the waves of the IABP (Kolyva et al., 2009;Lu et al., 2012) are larger (relative peak values) compared to the DEA-generated waves.This is expected, due to the large volume displacement of the IABP (up to 50 cm 3 ) compared to the small of the DEA (<2 cm 3 ) and that the DEA does not obstruct the aortic lumen.
The DEA-generated waves interfere differently with the LVgenerated waves depending on the PS (Fig. 3 and Fig. 4).For 0% PS (Fig. 3C), the BDW DEA (DEA activation) reaches the aortic valve just before opening, and lowers the EDP (Fig. 5C).This leads to a larger and faster FCW LV compared to baseline, indicating a stronger LV ejection in line with (Sugawara et al., 2009).A similar increase in FCW LV was also observed by (Lu et al., 2012) in response to IABP deflation.The BCW DEA (DEA deactivation) arrives at the same time as the start of the FDW LV .The corresponding reflected FCW DEA and the FDW LV superimpose, and the FDW LV peak decreases and occurs earlier.The hemodynamic effect of this interference is unclear.It could indicate a reduction in the inertial force of the blood flow out of the LV (Sugawara et al., 2009) which might lead to an earlier closure of the aortic valve and interfere with the total load on the LV.For 0% PS, the DEA reduces the hydraulic work (W H ) and thereby unloads the LV.This is in line with the reduced EDP, and with the possible earlier valve closure.For 50% PS (Fig. 3E), the BCW DEA arrives just before aortic valve opening, and increases EDP (Fig. 5C).This leads to a lower and delayed FCW LV indicating a weaker LV ejection in line with (Sugawara et al., 2009).The BDW DEA arrives just before aortic valve closure and delays the FDW LV compared to baseline.The corresponding reflected FDW DEA superimposes with the FDW LV , and the peak of the FDW LV increases.A stronger FDW LV could indicate an increased inertial force of the blood flow and thereby a delay in valve closure.For 50% PS, the DEA increased W H , leading to increased LV load.This is in line with the increased EDP and the delayed FDW LV .The wave interferences differ for all PS.For 60% PS, the DEA increases the FCW LV the most.Here the reflected FCW DEA superimposes with the FCW LV and increases the W H , leading to increased LV load.Fine-tuning of the start and end of actuation could optimize the effect of the DEA on the different hemodynamic parameters and the load of the heart (see Supplementary Materials).This finding is in line with (Lu et al., 2012), who observed improved efficiency of IABPs and para-aortic ACP depending on the wave timing.
The intended use of the DEA is as an ACP device.The non-obstructive nature of the DEA (aortic lumen open at all times, even at electrical failure) and the pure electrical actuation are advantages compared to current ACPs (De Waha et al., 2014;Khan et al., 2023;Santa-Cruz et al., 2006), and could enable a fully implantable device, for medium to longterm use.For a future final medical device, a modified design wrapping around the aorta would be desirable to reduce the invasiveness of the implantation and keep the native aorta intact.The effect of the device in the current study is modest (1-6%) but is expected to be higher in stiffer aortas (ageing) and in patients with hypertension.Higher pressures would generate larger deformations of the DEA and larger displaced volumes.The damping of the aortic waves is less in a stiffer system and the pulse wave velocity is higher.Optimization of the design is expected to further improve the effectiveness of the DEA.
Limitations: The acute experiments affected the pig's hemodynamics leading to lower aortic pressure and flow.An implantation in the human ascending aorta is expected to further improve the DEA efficiency (Sales and McCarthy, 2010).The small number (n = 2) of animals is a drawback.However, five different DEAs were tested, and many actuation cycles (3000 cycles (+2700 in the Supplementary materials)) were performed.WIA depends on pressure and velocity measurements in the same location.Although we aimed to position the flow and pressure sensors in the same location, the final exact position of the pressure sensor was difficult to assess.This might have contributed to the large variation seen in the estimated wave speeds.This, however, did not change the qualitative findings of the WIA between the different measurements.Additionally, WIA in the intact aorta could not be provided, this will be implemented in the future.

Conclusion
The wave intensity analysis shows that the DEA generated similar waves as other ACPs, and that the timing of these waves affected the Fig. 5. Average values for the (A) hydraulic work (W H ), the (B) mean diastolic pressure (P dia ) and the (C) end diastolic pressure (EDP) compared to baseline for all phase shifts (PS) for all animals and DEAs.The values are represented in boxplots in which the red lines represent the median value, and the edges of the boxes represent the 25th and 75th percentile, respectively.The average value for all animals and DEA is shown in red, and the results of animal 4 DEA 2 showing the strongest impact on EDP and P dia is shown in yellow.(For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)DEA's efficacy.The non-obstructive nature of the DEA allowed to phase shift the device actuation throughout the full heart cycle, and we show that the DEA-generated waves were shifted and interfered differently with the ventricular-generated waves.The wave interference gave insight into why the actuation time was beneficial or detrimental for the cardiac load.The best actuation timing led to decreased hydraulic work and end-diastolic pressure, and an augmented early diastolic pressure.

Fig. 2 .
Fig. 2. (A).Left ventricular pressure and pacemaker signal together with DEA actuation signals for two different phase shifts (PS).PS is defined as the time between the start of DEA actuation and the aortic valve (AV) opening relative to the heart cycle in percentage.B: ventricular pressure and pacemaker signal together with DEA actuation signal for PS = 0% (at aortic valve opening).Delay correction: earlier DEA actuation to remove time delays in the pressure signal arising from the distance between the DEA and the left ventricle.

Fig. 4 .
Fig. 4. WI pattern normalized to the peak value of the baseline forward compression wave seen in animal 4, DEA 2 for baseline and the 10 different phase shifts.The black solid lines indicate the DEA backward decompression wave generated by DEA activation and the dotted black lines indicate the DEA backward compression wave generated by the DEA deactivation.PS: phase shift.DEA: dielectric elastomer actuator.WI: wave intensity.

Table 1
Overview of animals and protocols used in the analysis.Several DEAs were implanted in each animal and tested for different voltage levels.