Predicted Cardiac Hemodynamic Consequences of the Renal Actions of SGLT2i in the DAPA‐HF Study Population: A Mathematical Modeling Analysis

Abstract

The Dapagliflozin and Prevention of Adverse Outcomes in Heart Failure (DAPA-HF) study demonstrated that dapagliflozin, a sodium-glucose cotransporter-2 inhibitor (SGLT2i), reduced heart failure hospitalization and cardiovascular death in patients with heart failure with reduced ejection fraction (HF-rEF), with and without type 2 diabetes mellitus. Multiple potential mechanisms have been proposed to explain this benefit, which may be multifactorial. This study aimed to quantify the contribution of the known natriuretic/diuretic effects of SGLT2is to changes in cardiac hemodynamics, remodeling, and fluid homeostasis in the setting of HF-rEF. An integrated cardiorenal mathematical model was used to simulate inhibition of SGLT2 and its consequences on cardiac hemodynamics in a virtual population of HF-rEF patients generated by varying model parameters over physiologically plausible ranges and matching to baseline characteristics of individual DAPA-HF trial patients. Cardiovascular responses to placebo and SGLT2i over time were then simulated. The baseline characteristics of the HF-rEF virtual population and DAPA-HF were in good agreement. SGLT2i-induced diuresis and natriuresis that reduced blood volume and interstitial fluid volume, relative to placebo within 14 days. This resulted in decreased left ventricular end-diastolic volume and pressure, indicating reduced cardiac preload. Thereafter, blood volume and interstitial fluid volume again began to accumulate, but pressures and volumes remained shifted lower relative to placebo. After 1 year, left ventricle mass was lower and ejection fraction was higher than placebo. These simulations considered only hemodynamic consequences of the natriuretic/diuretic effects of SGLT2i, as other mechanisms may contribute additional benefits besides those predictions.

Predicted Cardiac Functional Responses to Renal Actions of SGLT2i in the DAPACARD Trial Population: A Mathematical Modeling Analysis

“…32 Overview of DAPACARD Simulation Procedure Figure 2 summarizes the process used to simulate the DAPACARD study. Following the approach we have previously described, 24 we first generated a virtual population of participants with T2DM by sampling a subset of model parameters over a range of values with uniform distribution, summarized in Table 1 and described later. The parameter ranges were calibrated to ensure that the distribution of simulated baseline characteristics matched the DAPACARD clinical trial population, 23,32 including baseline estimated glomerular filtration rate (eGFR), blood glucose, systolic blood pressure (SBP), and diastolic blood pressure (DBP), covered the range of observed baseline values in DAPACARD.…”

“…This matching process ensures that correlations or lack of correlations between variables in the actual population were maintained in the virtual population. Although DAPACARD did not include participants with HFrEF, a second virtual population of participants with both T2DM and HFrEF was generated by inducing HFrEF, as described previously 24 in the DAPACARD T2DM population. The 2 sets of virtual participants (DAPACARD and DAPACARD-HFrEF) were then simulated on 10 mg of dapagliflozin or placebo treatment for 6 weeks.…”

“…We have previously validated that the model reproduces clinically observed effects of SGLT2i and ACEis on end points including glomerular filtration rate, blood pressure, sodium and water excretion, and LV mass (for ACEis). 24 Simulating T2DM Virtual Participants With and Without HFrEF Table 1 summarizes the model parameters varied to produce states of diabetes, hypertension, and HFrEF and their physiologically plausible ranges. The state of T2DM was generated by increasing average blood glucose concentration from 7 to 11.75 mmol/L, or hemoglobin A 1c from 6% to 9%, corresponding to the total range observed in the DAPACARD study population.…”

“…32 To reproduce this range from normotension to hypertension, we varied the parameters listed in Table 1 and described previously, which represents hypertension resulting from vascular disease, glomerulosclerosis and nephrosclerosis, impaired responsiveness to natriuretic signals, and/or renin-angiotensin-aldosterone system activation. 24 To produce the HFrEF population, additional parameters were varied to simulate functional and structural changes that occur in HFrEF as we have described previously. 24 These included reduced cardiac contractility representing ischemic injury, increased cardiac stiffness representing myocardial fibrosis, changes in myocyte geometry representing hypertrophy and eccentric remodeling, and alterations in the venous and arterial vascular systems.…”

“…The DAPACARD clinical study aimed to fill this gap by investigating the effect of 6 weeks of treatment with dapagliflozin, an SGLT2i, on myocardial function and metabolism, including measuring changes in left ventriclular (LV) global longitudinal strain and myocardial efficiency in subjects with T2DM. 23 We have previously used a quantitative systems pharmacology (QSP) model of cardiorenal function to model the renal natriuretic/diuretic mechanisms of SGLT2i, 24 and to predict the effect of these mechanisms on cardiac function and fluid status (blood volume and interstitial fluid volume [ISF]) in virtual participants with HF with reduced ejection fraction (HFrEF) matched to baseline characteristics of the DAPA-HF clinical trial. 25 In the analysis presented here, we aimed to complement the DAPACARD clinical trial by prospectively simulating the expected effects of the natriuretic/diuretic mechanisms of SGLT2i on global longitudinal strain and myocardial efficiency in virtual participants matched to the DAPACARD study population, as well as in a second population generated by inducing a state of HFrEF in the DA-PACARD T2DM virtual participants.…”