Waveform Analysis for Vascular Evaluation and Cardiovascular Advancements in Research and Enhancement

Valvular heart disease (VHD) is frequent in industrialized countries. Aortic
stenosis (AS) and mitral regurgitation (MR) are the two most common types of
VHD affecting millions of people worldwide. Its prevalence is estimated to be
between 5-20%, increasing with age and other risk factors such as
cardiovascular disease and hypertension. The consequences of VHD can include
heart failure, stroke, arrhythmias, and infective endocarditis, which can lead
to significant morbidity and mortality. Early diagnosis and treatment are
important to prevent or minimize these outcomes..

The treatment of VHD depends on the type and severity of the condition. In
cases of mild VHD, lifestyle changes and medications may be sufficient to
manage the symptoms, while in others, surgery may be required. The most common
surgical treatments for valvular heart disease include valve repair and valve
replacement. Minimally invasive procedures, such as transcatheter aortic valve
replacement (TAVR) or percutaneous interventions to treat MR such as
transcatheter edge to edge repair (TEER) with MitraClip* are also available for
certain types of VHD. The choice of treatment will depend on the individual
patient's age, overall health, and the specifics of their condition.

When left untreated, VHD is associated with several serious and potentially
life-threatening complications, including heart failure, stroke and arrhythmias
. It is possible that the treatment of VHD after the manifestation of symptoms
is suboptimal in some patients, as pathophysiological changes could be
irreversible. Early diagnosis could prove helpful in expediting treatment,
potentially preventing irreversible changes such as decline in left ventricular
function. Currently, trans-thoracic echocardiography (TTE) is the gold standard
to confirm diagnosis and assess the severity of VHD. However, TTE is not
feasible for screening asymptomatic patients and furthermore known to be highly
dependent on the skills of the operator. A more simple, non-invasive and
feasible way to detect (early) VHD would therefore be very valuable and could
potentially be used in the first line and outpatient clinics to screen patients
for major valve abnormalities.

VHD is associated with changes in the distal arterial pressure waveforms. For
example, AS causes a delayed pressure rise in the aorta, and an increase
systolic ejection period, reflected by a prolongation of left ventricular
upstroke time, and a decline in steepness of the pressure slope. This affects
the general shape of the pressure waveform as measured in the central aorta,
but this change in morphology is transmitted and can be measured more distally
in the vascular tree as well. Recently, we performed a feasibility study (CoArt
study) in which we created a diagnostic machine learning model to detect
patients with AS, based

on non-invasive blood pressure features. The model was able to distinguish none
to mild AS from moderate to severe AS with a sensitivity of 0.81, specificity
of 0.75, and AUROC of 0.82 in a highly predefined population. Further
validation in a more general population is important as well as continued
improvement of the model towards the diagnosis of other VHD.

Aortic aneurysms (AA) also lead to changes in the pressure wave form by
affecting its propagation through the vascular tree. AA of the ascending aorta
or abdominal aortic aneurysms (AAA) are usually asymptomatic until rupture or
aortic dissection occur, both of which are associated with extremely high
mortality. Consequently, the early detection and monitoring of AA is of major
importance in reducing mortality and morbidity. Currently, AA are most often
detected as coincidental findings by ultrasound, computed tomography, or
magnetic resonance imaging, requiring expert knowledge. Developing a machine
learning based algorithm for early detection of AA has been shown feasible in
silico. To develop a clinical model based on non-invasive blood pressure
waveforms may serve as a non-invasive, easily applicable, early screening tool
and thus, save lives.

Primary Objective:

Our overall aim is to develop an easy-to-use, and interpretive method for the
early detection of valvular heart disease and thoracic aortic aneurysm based on
arterial blood pressure waveforms, collected non-invasively. Therefore, we will
develop and validate (internally and externally) a machine learning algorithm
that can detect major cardiovascular pathologies (moderate to severe - mild to
none) in patients scheduled to have surgical correction. We identified the most
relevant pathologies, with respect to prevalence and potential detectability.

a) aortic valve stenosis

b) aortic valve insufficiency

c) mitral valve insufficiency

d) thoracic aortic aneurysm

e) healthy controls (matched on age)

Secondary Objective(s):

To determine whether there are differences in accuracy of the model between
males and females, and recalibrate the model accordingly

To determine whether there are differences in the accuracy of the model related
to the cause of thoracic aortic aneurysm such as genetic (tissue elasticity
deficits) or age related (calcification/stiffening of the vasculature)

To compare the waveform characteristics in supine and upright position.

We will conduct a prospective cohort (data collection study) in two phases. The
different phases of the study are depicted in figure 1. Based on our experience
in previous algorithm development we expect to analyse up to 10 variables per
model, per predicted deviation. Following the widely employed *10 events per
variable* rule of thumb, we will include 100 patients per studied deviation.

- Phase 1: Procedure specific, internal validation. In this phase the algorithm
is developed including a highly selective population of patients already
diagnosed with the disease and scheduled for surgery. Patients with aortic
valve stenosis or insufficiency, mitral valve insufficiency, thoracic AA and a
healthy control group of 100 patients each.

- Phase 2A: external validation of the algorithm; 150 patients suspected of VHD
or AA and scheduled for diagnostic evaluation are included.

- Phase 2B: external validation in another center; 150 patients suspected of
VHD or AA and scheduled for diagnostic evaluation are included.

Arterial pressure waveforms will be measured in patients at hospital admission
to obtain waveform data whilst being awake. Electronic data collection of
continuous non-invasive arterial pressure waveform signals takes place with the
CS/EV1000 system. In addition, we will require de-identified patient medical
records, including data from TTE and/or computed Tomography/ Magnetic Resonance
Imaging. No interventions will be done in phase 1 and 2.

The patient medical records charts will include demographic information and
intermittently

recorded pre, intra, and postoperative data.

Demographic information will include but not be limited to the following:

- Age, Height, Weight, Sex, BMI

- Procedures

- Comorbidities

- Admission Status, etc.

- Patient history

- Premedication

- Medication

- Pre-admittance blood pressure

Intermittently recorded pre, intra, and postoperative clinical data will
include but not be limited to the following (based on the monitoring devices
used as a standard of care for a specific patient):

- All hemodynamic parameters including, heart rate, blood pressure, SpO2, etc.

- All induction and post-induction medication used, dosing and speed of
injection, including vasopressors

- All ventilator variables including respiratory rate and ventilator settings

- Laboratory and microbiology tests

- Patient outcomes

- Hemodynamic and respiratory adverse events

The data collection will take place in the Amsterdam UMC, location AMC in the
Netherlands. The external validation dataset will be collected in a hospital to
be determined later.

We estimate that inclusion will take up to 36 months from initiation of study
(aim January 2023)

Participation in the study will involve non-invasive measurements of blood
pressure waveforms. The procedures have no additional risks or benefits. There
are no investigational devices used in this study. There are no additional
risks associated with the use of the CS/EV1000/HemoSphere monitor other than
described in the Instructions for Use. There are also no risks associated with
the study procedures.

The study's findings hold potential benefits for both individuals and
healthcare systems. The development of a non-invasive, easily applicable
screening tool for VHD and AA could expedite early diagnosis and intervention,
leading to improved patient outcomes and reduced morbidity and mortality