10.2021-Sense.nano-Ellen-Roche

ELLEN ROCHE: Hello, my name is Ellen Roche. I'm an associate professor in the Department of Mechanical Engineering and in the Institute for Medical Engineering and Science. My research is focused on implantable devices to augment organ function and repair tissue deficits, and the development of experimental and computational models to guide their design and reveal insights into the device's mechanical interaction.

Today, given the theme of the SENSE.nano symposium, I'm going to talk about a project that involves sensing in an instrumented benchtop setup and is moving towards a clinical study. This work is towards interventions for improving hemodynamics in single ventricle physiology patients. Let me start by explaining the single ventricle physiology, and how it motivates our work.

Two to three babies per 10,000 live births have a condition called hypoplastic left heart syndrome, which is one type of single ventricle disease. A healthy heart has two ventricles, or large pumping chambers, as shown on the left. In the hypoplastic left heart, the patients have a severely underdeveloped left ventricle, the chamber that usually pumps oxygenated blood to the body. This is corrected by a series of surgeries named the Norwood, Glenn, and Fontan surgeries, occurring within the first few months of life and about two and five years of age respectively.

The resulting surgically reconstructed heart is shown in the animation on the right. You will note that owing to the surgical modification, the oxygenated, or red blood, is pumped to the body by the functioning ventricle. And the deoxygenated blood returns to the lungs from the upper and lower body through a non-pumping passive conduit, or shunt, which I will refer to from now on as the Fontan shunt.

So after surgery, there is no pump to return blood to the lungs. This results in unfavorable hemodynamics which I'll describe in more detail. And these, in turn, result in multiple deleterious effects. Now in these patients, instead of having two parallel pumps and circuits, patients have left and right circuit in series and just one pump. The color blocks here on the upper left respond to the pressure diagram on the lower left. These pressures are very high for the venous side.

So patients have systemic venous hypertension, or high blood pressure. Another notable feature of the hemodynamics is that there is retrograde flow, or flow returning to the abdomen and not making it forward to the lungs. These hemodynamics can lead to a backing up of blood and some of the abdominal organs. They can cause things like liver fibrosis, low blood oxygenation, and poor exercise tolerance.

Recently it has been shown that the Fontan flow pattern on the venous side is governed by respiration, in contrast to in a biventricular physiology where it is governed by the cardiac cycle. As you can see on the left, during inspiration when the diaphragm lowers, abdominal pressures rise, pushing blood that is pooled on the venous side into the thorax where the pressure is falling. This leads to a surge of venous flow of up to 140%.

Conversely, during expiration, the pleural pressure increases, pushing blood out of the pulmonary vasculature. And without a right heart, blood is pushed all the way down into the abdomen where the pressure falls. This leads to pronounced retrograde, or backward flow, of up to about 30% and increased pressure burden and less oxygenated blood.

We can use ultrasonic arterial flow sensors to measure flow in a typical inferior vena cava, here shown in blue. Here we can see that it is synchronized with the cardiac cycle with a frequency of close to 70 beats per minute. When we measure the flow in the Fontan shunt using similar ultrasonic flow probes, we can see it's governed by breathing, and the frequency is more like 20 breaths per minute. You'll note in the shaded area the retrograde, or back flow, also.

So we've established the link between breathing and flow in these patients. And this is clear. What we want to do in the lab is to overlay the breathing mechanics with single ventricle physiology flow dynamics in experimental, computational, and clinical models. Why do we do this? Today, benchtop systems have not recapitulated the impact of respiration on flow, and there are no good, closed-chest, animal models of this physiology that maintain the negative pleural pressure that is important in cardiorespiratory interdependency.

We believe that recreating hemodynamics is a key building block to better understand single ventricle flow dynamics and to develop non-invasive, and invasive, clinically relevant support strategies for Fontan patients. I'm going to tell you about two ways in which we model and investigate the biomechanics of single ventricle physiology. Specifically, this link between breathing mechanics and single ventricle hemodynamics.

My goal is to share the work, and hopefully spark some ideas for advance sensing technology that could help us with our benchtop and clinical endeavors. First, we create a benchtop mimic of the anatomical structures pertinent to respiration, as you can see on the left. We model the abdominal and chest cavity with acrylic pressure chambers that were separated by a diaphragm.

The diaphragm anatomy was segmented from patient data and reconstructed with polyurethane. The diaphragm is controlled pneumatically with artificial muscles that lower the diaphragm into the abdominal cavity and recreate positive abdominal and negative thoracic pressures that occur during inspiration. This video shows how we can inflate, and deflate, organic lungs due to changes in transdiaphragmatic pressure in our respiratory simulator.

We can vary the compliance of the lungs, the surrounding cavities, or the abdominal cavity, and look at the effect each of these changes has on breathing mechanics. So we took this respiratory simulator, and next we look to integrate what we call a Fontan flow loop with this simulator. We modeled the venous vasculature of the Fontan physiology using resistance and compliance elements.

For the head, which is not directly subjected to breathing pressures, we used a classic compressed air wing capsule model. For the lungs and the thorax, we used two silicon reservoirs which approximated the physiological compliance of the pulmonary vasculature, and the geometry and elasticity mimicked by silicon rubber. Similarly, we used a lumped parameter model for the compliance of all abdominal venous vasculature with one large silicon compliance chamber.

The advantage of silicon is that they expand and contract according to physiological breathing pressures. By adding patient specific image derived 3D-printed vessel anatomy, we can make the model patient specific. This culminated in our finalized respiratory and flow model. We are using a surrogate blood fluid, which is optimized to have the density and viscosity of blood at room temperature. The overlaid cardiorespiratory model for the Fontan flow loop is shown on the right here.

Of course, pertinent to this talk, we instrument the rig heavily with sensors. We use ultrasonic flow sensors on the outside of the mock vessels and flow conduits, and we measure pleural and abdominal pressure with the integrated sensors. We also sense and regulate the input pressure to the artificial muscles that move the diaphragm. With this setup, we were able to recreate the flow patterns published previously in clinical MRI studies.

You can see the patient data in green, and the flow data from our simulator in red. During inspiration, the flow increases, and it falls during expiration and causes retrograde flow, which is shaded here in purple. This is typical for Fontan hemodynamics. Then we can start probing parameters. For example, what happens when we take deep or shallow breaths, and how does that affect respiratory flow?

Here we see a decrease as we move from deep breathing to shallow breathing. I just started that slide again, sorry. Then we can start program parameters. What happens when we take deep or shallow breaths, and how does that affect flow in the Fontan conduit? Here we see a decrease as we move from deep breathing to shallow breathing. Similarly, we look at how does vessel compliance, or stiffness, affect flow.

We see less forward, and less retrograde, flow with stiffer vessels as they are not as responsive to the external pressure changes. With this tool built, we are aiming to complete an in vivo study to further understand biomechanics and to validate our benchtop models. Their protocol is to do an eight minute breathing exercise with Fontan patients in the MRI machine, and this is done with our collaborators at Boston Children's Hospital.

During these studies, we will measure 4D flow in the Fontan conduit and in the aorta. We plan to do a number of different breathing exercises. First, we will do normal breathing, then shallow, heavy, slow, and fast breathing, for at least two to four cycles each. We include Fontan patients that are over the age of 14, and that have no fenestrations in their Fontan.

A visual and audio instructions help patients in the MRI machine to breathe according to the protocol. These visuals show them when to breathe in and when to breathe out. So that they can really time their breathing with the patterns that we want them to achieve. We record chest wall motion, flow in the Fontan conduit or the inferior vena cava, and aortic flow.

We retrieve the data from the image set. Look at anterograde and retrograde flow at each of the aforementioned anatomical locations, and then we segment chest wall motion to calculate breathing effort and respiratory rate. Here is where sensors could also help us a lot with our data collection. In the future, we will complete this study. And results will help to validate our benchtop models and elucidate the interplay between different types of breathing and flow in a clinical study, which could lead to non-invasive interventions.

We would love to identify and develop implantable sensors for measuring flow in the Fontan shunt. So we could continuously monitor this flow and correlate it with wearable breathing sensors. Finally, we are concurrently developing therapeutics consisting of active devices to augment the flow to the lungs in these patients, like a replacement right ventricle. Ideally, these devices will be triggered from breathing pressures, or from changes in the direction of flow in the shunt.

So their implantable biocompatible sensors would be critical for synchronization and closed loop operation. With that, I would like to thank the team in my lab working on this project in our fantastic collaborations, collaborators at Boston Children's Hospital. If you'd like to read more about our work, the publications are listed on the right. Finally, I would like to acknowledge the funders of this work. And I thank you for your time. I will be answering questions in the live panel shortly. Thank you.