Direct Cardiac Compression Device for Short-Term Mechanical Circulatory Support

Authors: Kristof Sarosi, Thomas Kummer, Stijn Vandenberghe, Stefanos Demertzis, Patrick Jenny

Abstract

Patients requiring short-term ventricular assistance have limited options. While not all interventions necessarily require blood-contacting support, all current devices are invasive and interact with the bloodstream. Direct cardiac compression (DCC) devices offer a potential solution by providing mechanical circulatory support (MCS) without contact with the bloodstream. This proof-of-concept study compares a novel DCC patch device for short-term MCS to open-chest cardiac compression (OCCC) using an ex vivo ovine cardiac arrest (CA) heart model. Performance is evaluated by assessing pressure, flow rate, and valve functionality, as well as any damage to the tissue. In the CA model, the device achieves 1.5 L/min cardiac output and a mean aortic pressure of 55 mm Hg. Despite direct epicardial contact, the DCC patch device maintains effective valve function. This study demonstrates that the DCC patch device achieves comparable performance to OCCC, supporting previous in vitro findings. Our results suggest the potential for this novel DCC patch device to be applied across a range of short-term MCS scenarios.

Heart failure (HF) and cardiac arrest (CA) patients requiring short-term ventricular assistance have limited options. In contrast to HF, there is no blood flow during CA, leading to emergency cases. Rapid medical intervention is key as blood circulation must be restored within minutes. Cardiac arrest can be triggered by HF, anesthesia in surgery, and many other factors.^1,2^ Cardiac arrest can be treated by cardiopulmonary resuscitation (CPR), defibrillation, or antiarrhythmics, but in certain cases, these techniques cannot be used.^2–4^ In the case of penetrating CA, CA after recent cardiac surgery, decompression of pericardial tamponade, and CA during open-chest surgery, open-chest cardiac compression (OCCC) is recommended.^4^ In many cases, OCCC is only a temporary solution before the patients’ circulation is supported by extracorporeal membrane oxygenation (ECMO).^5–7^

In OCCC, cardiac surgeons manually massage the heart, supporting circulation. Open-chest cardiac compression is an invasive technique, requiring resuscitative thoracotomy or an open chest to access the heart. D’Souza and Law^4^ provide a modern view on OCCC and emergency thoracotomy. As there are no standards for OCCC, only recommended practices, there is great variability in outcomes depending on the surgeons and the patients.^8,9^ Open-chest cardiac compression requires surgeons to maintain circulatory support until recovery or device support is established.

With the high costs and invasive nature of OCCC and current mechanical circulatory support (MCS) devices, alternative MCS solutions are still sought after. Direct cardiac compression (DCC) devices have been under investigation since before the time of left ventricular assist devices (LVADs) and ECMO devices. Direct cardiac compression devices are similar to OCCC as there is no direct contact with the bloodstream and the device is in direct contact with the epicardium. Even early DCC prototypes could outperform OCCC and closed-chest cardiac compression (CCCC) in organ perfusion and cardiac output (CO).^8,10^ The recent breakthrough in soft robotics fuels a renaissance of DCC devices.^11–15^ These devices could be used in prolonged total circulatory support (CS), resuscitation in CA, MSC following ischemia and infarction, in vivo organ preservation, MCS in hypothermia, and in drug overdose cases.^10^ Bonnemain et al.^11^ defined requirements for such systolic enhancement, low filling pressure, avoiding blood exposure, and adapting to varying conditions. Oz et al.^16^ described additional criteria for ideal DCC left ventricle (LV) and right ventricle (RV) stroke volume (SV) balance, safe control of RV pressure, simple and quick to apply, adjustable, MCS in CA, and easy removal. DCC devices could improve CS,^17,18^ bridge-to-decision, and bridge-to-device therapies.^12^

Limitations related to DCC device designs have so far prevented positive results on human subjects. Reducing diastolic filling pressure for better filling of the ventricles is an issue for most DCC devices as they do not provide active suction.^15,19^ As DCC devices compress the heart, and with that deform the muscle tissue and the support structures (chordae, annulus) of valve apparatus, they can cause valve regurgitation.^19^ Mechanical depolarization of the heart due to direct mechanical compression could hinder cardiac recovery.^16^ Increased filling pressure^16,19,20^ and epicardial surface adhesion^21^ could lead to complications with DCC devices.

There is no single MCS or DCC solution solving every problem of patients with HF or CA. We designed a DCC device for emergency scenarios during open-chest surgeries or following scheduled or emergency thoracotomy, providing short-term CS. We aim to increase the available options for cardiac surgeons, potentially replacing OCCC. Our research is based on the study of Kummer et al.,^22^ who published their work on their simulation framework, supporting the development of early prototypes. Detailed information on the DCC patch device design and in vitro results can be found in our previous publication.^23^ In this study, we present the results of our ex vivo study on an ovine CA model, comparing the performance of the DCC patch device to OCCC.

Materials and Methods

This proof-of-concept study aims to showcase the performance of the DCC patch device. We compare the performance of the DCC patch device to OCCC to study whether it could replace OCCC in clinical practice. Other use cases are discussed in the Discussion section. Our study is conducted in an ex vivo mock circulation loop (MCL) on a single ovine CA model in the Cardiac BioSimulator^24^ at LifeTec Group, Eindhoven.

The DCC patch device consists of two patches, shaped like the cups of a hand, a driveline, and a vacuum line. Figure 1 shows the DCC patch device. The patches are in direct contact with the epicardium, compressing directly on the ventricles. The shape of the two patches is different as the LV patch has less curvature than the RV patch. This design difference should help in balancing the pressure in the LV and RV. The vacuum is required to keep the device attached to the surface of the heart through the suction cups in the center of each patch. A magnetic locking mechanism between the two patches allows for quick deployment in emergency scenarios. Further details regarding the DCC patch device can be found in a previous publication.^23^

Using the DCC patch device enabled us to tune certain parameters, such as the orientation of the device, the initial distance between the patches, the rate of actuation, and the systolic fraction. Despite having a fully automated actuation system, we decided to share the results with manual operation to demonstrate the full potential of the DCC device. The power output from motorized actuation was deemed insufficient compared with manual actuation. The operator was instructed to push the lever mechanism at a given rate of actuation with a given speed of contraction (slow/quick). An audio signal at the desired 100 bpm rate of actuation helped the operator of the DCC patch device to perform regular compression. The ex vivo test bench is set up as an open MCL with the ability to control preload and afterload values for both the left and right heart circulations individually. Pressure values are recorded directly at the aorta, pulmonary artery (PA), left atrium (LA), and right atrium (RA). The ultrasonic flow meters were attached to Tygon tubes downstream of compliant silicone tube sections. The results presented in this study are achieved with a preload of 16 mm Hg on each side.

The cardiac surgeon was instructed to perform OCCC just as in the operating room (OR) (Figure 2). As there is no standard technique, OCCC is unique to the surgeon. With our experimental setup, the surgeon had more direct and real-time information and feedback on the performance of the OCCC than in an OR, as live pressure and flow rate data, cardiac output, and videos of the heart valves were shown live on a screen. This might have already resulted in improved performance of OCCC compared with performing OCCC in an emergency scenario in an OR.

As for this proof-of-concept study, using the DCC device in an ex vivo setup for the first time, we adjusted the settings of the patches in a trial-by-error manner. This study does not include a detailed description of this process, only the most important findings, and results using the settings as a conclusion of the initial trials. In total, 38 tests were conducted in 2 hours using the same ovine model. These tests included OCCC studies, parameter studies using the DCC patch device, and studies with a modified DCC patch design (not included in this study). Along with pressure and flow rate measurements, we also recorded videos of the tricuspid and mitral valves, and the experiments themselves. Pressure and flow rate data were recorded for approximately 30 seconds at 1,000 Hz for each test. This generated enough data points to create statistically significant measures of performance.

Post-processing includes filtering the high-frequency noise in the signal, segmenting and averaging the data over cardiac cycles, calculating the standard deviation in time series, mean and peak pressure, SV, CO, and rate of actuation. We used Welch’s t-test to compare the cases of OCCC and the DCC device, as the analysis of variance (ANOVA) homoscedasticity assumption was not met. We used one-way ANOVA for certain sensor data in the parameter studies, as both the Shapiro-Wilk test and the Levene test indicated that we met the conditions for using one-way ANOVA. In any other case, we used Welch’s t-test.

Results

In this section, we present raw data from the pressure sensors and flow meters and statistically evaluated measures of performance such as average heart cycle, consistency of output, SV, CO, and peak and mean pressure in the aorta and the PA. The results are compared between OCCC and the DCC patch device. We find that the DCC patch device performs similarly to OCCC, providing the required CO and pressure levels for MCS while maintaining safe operation. The performance of the device with related results is presented in the following paragraphs.

Figure 3 shows raw data using OCCC and the DCC patch device. Both OCCC and DCC patches demonstrate significant pressure level differences between the atria and the outflow tracts. This indicates that the flow is driven by the compression, not by the preload.

We segmented and averaged the collected data over every cardiac cycle to investigate the expected output and consistency. The results of the averaging are shown in Figure 4. The difference in the length of the cardiac cycles is due to the different rates of actuation. As the surgeon was not aided by an audio signal, the mean rate of actuation of the OCCC was 80 bpm (Figure 5). Figure 4 shows similar values in pressure and flow rate for both OCCC and the DCC device, but the consistencies of the two actuation modes are different. Our DCC patch device shows a smaller standard deviation, that is, a more consistent mode of actuation.

Both OCCC and the DCC patch device were able to provide around 15 ml of SV despite the different rates of actuation. Figure 5 shows SV and rate of actuation data for the two modes of actuation. Stroke volume is an important measure as it can be limited by the design of the patches and reduced by insufficient filling pressure or time. As shown in Figure 5, the rate of actuation can vary greatly for OCCC. Long-term OCCC especially suffers from this phenomenon as the surgeon is getting fatigued by the extensive pumping action. Given that the total CO is given by the SV and the rate of actuation, the performance can be better measured by CO. Figure 6 depicts CO with mean arterial pressure (MAP) and mean pulmonary arterial pressure (mPAP) as the main indicators of performance. Importantly, the DCC device has significantly different pressure levels in MAP and mPAP. The difference between OCCC and the DCC patch device regarding CO is nonsignificant. Both modes of actuation deliver around 1.5 L/min CO on average. In terms of mean pressure, the DCC patch device delivers significantly higher pressure levels compared to OCCC, but still in the safe region for hypertension^25–28^ (references show the following mPAP >25 mm Hg, aortic pressure >140/90 mm Hg).

To obtain the results, we have performed preliminary tests on different actuation frequencies, systole length, and patch orientation. We share the most important findings regarding rate of actuation and using different systolic fractions. Figure 7 shows that higher rate of actuation leads to significantly higher CO, MAP, and mPAP values. Regarding the systolic fraction, CO and MAP are significantly higher with a smaller systolic fraction, while mPAP is less affected.

Discussion

The DCC patch device was designed to suit the needs of emergency scenarios in open-chest surgeries or following thoracotomy. Valve regurgitation was monitored by cameras capturing the mitral and tricuspid valves. Valve functions were maintained during device operation. Regarding hypertension, we evaluated mean and peak pressures in the aorta and the PA. Using the manual actuation of the device, we did not eliminate the human factor and thus the variability of the output. The operator of the device was not a medical professional, unlike the cardiac surgeon performing OCCC. Using simple audio pacing, we not only managed to decrease the variability in the output of the device but also increased the cardiac output, something to note for the clinical audience. Issues concerning long-term use such as increased filling pressure and tissue-device adhesion are not discussed in this study.

Assessing hypertension is crucial for both the systemic and the pulmonary circulation. The pulmonary circulation is especially prone and sensitive to hypertension as the wall of the RV is much thinner than that of the LV. According to the literature, peak and mean pressure in both the LV and the RV are good indicators of hypertension.^25,26,29^ The pressure values are below the hypertension levels for both the aorta and the PA.

Sensitivity to the position of the device could be of interest for surgeons. Should they be concerned about positioning the device perfectly in an emergency? As the two patches of the DCC device are different in size and curvature, this information could be crucial for surgeons. There was indeed a significant difference regarding the different positions of the DCC device on the heart. The intended orientation has a lower output on the pulmonary side than with different orientations. To avoid pulmonary hypertension (PH), we used the device in the intended position. With this, we sacrificed performance for safety. Hypertension is an important factor in assessing the safety of the device. As shown in Figure 6, the aortic and PA pressure levels are in the safe range using the DCC patch device.

Reaching high pressures on the systemic side is crucial for organ perfusion. Some studies show improved performance in organ perfusion with pulsatile DCC and MCS devices over continuous ones,^10,30^ but the broader literature shows mixed results. Organ perfusion is crucial for the long-term survival of patients. Our device generated pressure increments in the LV and the RV at different levels, crucial for organ perfusion while avoiding RV hypertension.

Another important aspect of DCC device feasibility is the interaction of the tissue with the device. We used our DCC patch device on the ovine model for 2 hours (with pauses). In this time frame, the DCC patch device caused visible damage to the epicardium (Figure 8). We must add that this study was a proof-of-concept study only for the performance of the device, and not for biocompatibility and long-term use. The surface of the device was not treated after it was manufactured with powder bed fusion (PBF) 3D printing, leaving a rough contact surface for the epicardium.

According to the literature, most of the recent DCC devices aimed to replace VADs. Ventricular assist device treatments have improved greatly in the past decades with the third-generation devices, reaching more than 80% survival in the first year.^31^ Accessibility is still an issue with VADs, requiring improvement, but we see the application of our device in different treatments. Anstadt et al.^10^ made a comprehensive review of DCC devices (for direct mechanical ventricular actuation [DMVA] devices) and their applications. Anstadt’s device was a breakthrough in the field of DCC, achieving astonishing results in animal studies, but their human studies did not succeed. According to the publications of the CorInnova device, DCC devices might be able to replace ECMO devices in certain scenarios.^20^ We aim to concentrate on different applications, with the focus on short-term MCS. This study was meant to show new possibilities for resuscitation in CA. We aim to use our device as short-term support in open-chest surgeries or following thoracotomy. With that in mind, our device was designed to allow for quick deployment in emergency scenarios. As additional use cases for our device, prolonged support in fibrillation, use in in vivo organ preservation, and support in hypothermia could also be explored.

Reviewing the studies published for the CorInnova DCC device, we can compare our results with one of the most established DCC projects of recent times. In their latest publication on a caprine CA model for pediatric patients, they generated 0.4 L/min CO, and 6 mm Hg increase in both MAP and mPAP.^20^ Their study on the ovine HF model shows similar results.^32^ For comparison, our DCC patch device delivered 1.5 L/min CO, 36 mm Hg increase in MAP and a 7 mm Hg increase in mPAP. We must mention that the CorInnova studies were conducted in vivo, while our study uses a single ex vivo ovine model. However, we believe that the results from the CorInnova publications serve as reasonable references for our study.

One of the biggest flaws of previous DCC devices is that unlike in physiological conditions or during OCCC, the generated pressure increment for the LV and the RV are similar. Initially, in our measurements, the pressure measured at all four measurement points is equal. Any pressure difference between measurement points at the same time instance is created by the OCCC or the device. Results in Figures 4 and 6 show the pressure increment differences generated in the LV and RV by the DCC device.

This study presents an ex vivo study on a single ovine CA model. We used an open MCL provided by LifeTec Group. This MCL allows users to independently control preload and afterload for the LV and the RV. To improve on this study, a closed MCL, a cadaver, and later an in vivo model should be used. Translating the results from an ex vivo model to an in vivo one is not straightforward but could prove as a valuable reference. Unlike the CorInnova device, our DCC patch device cannot be used in a minimally invasive procedure. The DCC patch device requires thoracic surgery or an open chest for deployment. Our design is currently not biocompatible as this and our previous study^23^ assess only the output created by the DCC patches.

Considering the results, the DCC patch device performed similarly to OCCC carried out by a trained surgeon. Figures 3 and 4 show similar flow rates and pressures for the DCC device to OCCC. Regarding the dampened flow rate values, both the tricuspid and the mitral valves are closed after systole. As there are no signs of regurgitation, the only contributing factor to this effect is the compliant tubing upstream of the flow meters. This limitation is given by the Cardiac BioSimulator’s design. This proof-of-concept study shows the potential of providing life-sustaining support with the DCC patch device for patients in emergencies. The device should be placed on the heart by a surgeon but it could be operated electronically or manually by other members of the medical team. Replacing OCCC would free surgeons from the burden of supporting the circulation of the patient manually, allowing them to carry out other tasks in the OR, stabilizing the patient for bridge-to-decision. The DCC patch device could also be used bedside in intensive care units (ICUs) for emergencies where OR readmission time could put the patient’s life at risk.

Conclusions

In this study, we present a novel, fully mechanical DCC patch device developed at ETH Zürich. The device consists of two patches, similar to clam shells or human hands, directly attached to the surface of the heart. The DCC device was designed for emergency scenarios where quick deployment is crucial. Our aim is to provide short-term CS, to avoid contact with the bloodstream, reduce costs, and free surgeons from manual cardiac compression. Results show that our device can outperform a trained surgeon by generating matching flows and significantly higher MAP and mPAP, sufficient to sustain the life of a CA patient while further treatment options are considered.