Improving Hypertension Treatment in African Americans Using Computational Modeling and Predictive Analytics - ABSTRACT
As compared to whites, African Americans (AA) develop hypertension (HTN) at an earlier age, have a
greater frequency and severity of HTN, poorer control of blood pressure (BP), and have twice the mortality
rate from HTN. For 47 years our department has been developing computer simulations of integrative
physiology for research purposes. The current model, HumMod, is comprised of 14 organ systems, and
includes neural, endocrine, circulatory, and renal physiology. We have created tools that generate and
analyze large cohorts of computer-generated (virtual) patients. With these techniques HumMod has been
used for hypothesis generation and for understanding underlying physiological mechanisms that are not
able to be determined in either whole animal or human experiments. This proposed work will use these
tools and this mathematical model of human physiology to develop a realistic AA virtual population for
studying antihypertensive therapies that have well-known (diuretic or salt reduction), variable
(angiotensin converting enzyme, or ACE inhibition), or unclear (renal denervation (RDX), and baroreflex
activation therapy (BAT) therapeutic efficacies in AA. Published data from our laboratory show that our
model is robust and can realistically simulate salt sensitivity, multiple types of HTN, and device-based
antihypertensive therapy. As shown in our preliminary data, we have successfully created a virtual
population that was similar to the clinical data (AA population with resistant HTN) in 5-dimensions (blood
pressure, heart rate, glomerular filtration rate, cardiac output, and peripheral resistance) and have
conducted in silico trials for new device-based therapy currently being evaluated for the treatment of
resistant HTN—namely RDX, BAT, and arteriovenous fistula. Based on these preliminary data, we
hypothesize that these techniques will allow us to investigate the physiological mechanisms responsible
for the variation in response to therapy in a wide range of AA patient types and predict the likelihood of
success for a particular treatment. Aim 1 of the proposal will test the hypothesis that a virtual AA
population with resistant HTN can be successfully calibrated and validated. Aim 2 of the proposal will test
the hypothesis that in silico trials using the calibrated populations from the first Aim can be used for testing
and predicting mechanisms of nonresponse to device-based antihypertensive therapies. Aim 3 will test
the hypothesis that our predictive analytic techniques can be used to identify mechanisms and proxy
markers of therapeutic resistance in hypertensive AA. These proposed studies have clinical relevance
because they address a leading cause of morbidity and mortality as well as potential mechanisms of
therapeutic resistance in an underserved and understudied minority. Furthermore, these applications and
the potential insights gleaned from our physiological model and predictive analytic tools may have broad
implications for BP control in other resistant hypertensive populations.
