Coronary microvascular dysfunction and left heart filling pressures in patients with type 2 diabetes

Abstract

Aims

Patients with type 2 diabetes (T2D) have a high prevalence of diastolic dysfunction and heart failure with preserved ejection fraction (HFpEF), which in turn leads to an increased risk of hospitalization and death. However, the factors of risk and their relative importance in leading to higher left ventricular filling pressures are still disputed. We sought to clarify the associations of a wide range of invasive and non‐invasive risk factors with cardiac filling pressures in high‐risk T2D patients.

Methods and results

Patients with T2D at high risk of cardiovascular events were prospectively enrolled in this study. Participants were thoroughly phenotyped including right heart catheterization at rest and during exercise, echocardiography, urinary excretion of albumin (UACR), and quantification of their myocardial blood flow rate (MFR) using cardiac ⁸²Rb‐PET/CT. Of the 37 patients included in the study, 22 (59%) patients met invasive criteria for HFpEF. Only 2 out of 39 variables emerged as independent factors associated with left ventricular filling pressure as assessed by pulmonary capillary wedge pressure (PCWP) at rest; history of hypertension (coefficient: 2.6 mmHg [0.3; 5.0], P = 0.030) and MFR (P = 0.026). We found a significant inverse association between MFR and PCWP with a coefficient of −2.3 mmHg (−4.3; −0.3) in PCWP per integer change of MFR. The MFR ranged from 1.18 to 3.68 in our study, which corresponds to a difference in PCWP of approximately 6 mmHg between patients with the lowest compared to the highest MFR. During exercise, only 2 variables emerged as borderline independent factors associated with PCWP: myocardial flow reserve (coefficient: −4.4 [−9.6; 0.8], P = 0.091) and beta‐blockers use (coefficient: 6.1 [−0.1; 12.4], P = 0.053).

Conclusions

In patients with type 2 diabetes without known HFpEF but risk factors for cardiovascular disease, myocardial blood flow rate was independently associated with PCWP at rest across the range from normal to abnormal left heart filling pressures. A clinically significant difference of 6 mmHg in PWCP was attributable to differences in MFR in patients with the lowest compared with the highest MFR values. This suggests that strategies than attenuate microvascular dysfunction could slow progression of increased left ventricular left heart filling pressures in patients at increased risk.

Introduction

Patients with type 2 diabetes (T2D) have a higher prevalence of diastolic dysfunction and heart failure with preserved ejection fraction (HFpEF) than persons without diabetes.¹ The high prevalence of HFpEF among patients with T2D suggests that long-term dysglyceamia directly or indirectly may promote higher cardiac filling pressures. Transgressing from having normal to abnormal left heart filling pressure increases risk of heart failure symptoms as well as mortality.² Hence, identifying possible factors predisposing to higher ventricular filling pressures is of interest with regard to prevention of heart failure and cardiovascular death. So far, governing factors leading to elevated filling pressures in diabetes have not been fully explored, although many plausible factors have been suggested.³ In general, the gold standard for assessment of filling pressures demands invasive measurement of intracardiac pressures, which might explain some of the paucity of knowledge regarding these factors. In addition, investigating patients with invasive cardiopulmonary stress testing may further reveal early stages of diastolic dysfunction, which are unmeasurable during rest.

Studies investigating patients with both diabetes and HFpEF have reported that coronary microvascular dysfunction (CMD) is associated with left ventricular filling pressures,⁴ and CMD has been shown to associate with the degree of diastolic dysfunction and risk of heart failure hospitalization.² However, patients with diabetes generally have a higher comorbid burden than their non-diabetic peers.⁵ This makes inferences on the most pivotal factors in the transition from normal to abnormal filling pressures difficult, if only patients with abnormal filling pressures are examined. Using patients with T2D with risk factors for increased cardiac filling pressure, we tested the hypothesis that reduced coronary microvascular function as assessed by myocardial flow reserve is associated with higher cardiac filling pressures in high-risk type 2 DM patients using right heart catheterization during rest and exercise.

Methods

Patients enrolled in the SIMPLE trial were invited to participate in this pre-defined, haemodynamic sub-study. The SIMPLE trial enrolled 92 patients who all underwent baseline characterization, including 82Rb-PET/CT. Among the participants from the primary study, 38 patients were randomly selected to participate in this haemodynamic substudy. All patients selected consented for the haemodynamic sub-study. The design of the SIMPLE study has been published previously.⁶ This study was registered at ClinicalTrials.gov (NCT03151343). The study protocol (v7, 02/08/2018) has been approved by the Ethics Committee of the Capital Region (2016-775), Danish Data Protection Board, and the Danish Medicines Agency. EudraCT Number: 2016-003743-10. The study complied with the principles of the Declaration of Helsinki.

In short, patients were enrolled if they had stable T2D and were at high risk of cardiovascular events, defined by one or more of the following; increased urine albumin to creatinine ratio (UACR) ≥ 30 mg/g, plasma N-terminal pro-BNP (NT-proBNP) ≥ 70 pg/mL, or confirmed history of myocardial infarction (MI) > 2 months prior to baseline, heart failure (HF) according to Framingham Heart Failure Criteria, discharged from hospital with a documented diagnosis of unstable angina ≤12 months prior to baseline, evidence of coronary artery disease (CAD) by coronary angiography in one or more major coronary arteries, or at least one of the following: a positive non-invasive stress test or a positive stress echocardiography showing regional systolic wall motion abnormalities; a positive radionuclide test showing stress-induced ischaemia; history of ischaemic or haemorrhagic stroke >2 months prior to informed consent; presence of peripheral artery disease such as previous limb angioplasty; stenting or bypass surgery; previous limb or foot amputation due to circulatory insufficiency; angiographic evidence of significant (>50%) peripheral artery stenosis in at least one limb; evidence from a non-invasive measurement of significant (>50% or as reported as haemodynamically significant) peripheral artery stenosis in at least one limb; ankle brachial index of <0.9. Exclusion criteria included eGFR ≤30 mL/min/1.73 m²), severe liver insufficiency (Child–Pugh class C), clinically significant heart valve disease, MI ≤ 30 days, PCI ≤ 4 weeks prior to baseline, LVEF ≤ 40% by baseline echocardiography, suspicion of cardiac amyloidosis, or hypertrophic cardiomyopathy. The complete listing of inclusion and exclusion criteria have previously been published.⁶

Haemodynamic examination

Right heart catheterization was performed using a standard 7.5-F triple lumen Swan-Ganz catheter (Edwards Lifesciences, Irvine, CA). The catheter was introduced under local anaesthesia into the internal jugular vein using the Seldinger technique and advanced to the pulmonary artery verified by signature pressure curves. The following haemodynamic data were collected; right atrial pressure (CVP), systolic/diastolic/mean pulmonary artery pressure (SPAP/DPAP/MPAP), pulmonary capillary wedge pressure (PCWP), cardiac output using thermodilution technique (CO), non-invasive systolic blood pressure (SBP), non-invasive diastolic blood pressure (DBP), and heart rate (HR). Invasive criteria for HFpEF was defined according to ESC HF guidelines (PCWPrest ≥ 15 mmHg and PCWPexercise ≥ 25 mmHg).⁷

At rest, invasive pressure readings were obtained during end-expiration. During exercise, pressure curves were averaged over 10 s. Cardiac output was measured using thermodilution as the average of 3 measurements with <10% variance and was indexed to body surface area as cardiac index (CI).

Haemodynamic variables were measured at rest and during supine ergometer exercise (ebike; GE Healthcare, Fairfield, CT). Measurements were obtained at rest, submaximal exercise (25 watt), and at peak exercise. Measurements were obtained after 2 min at a given workload. After measurements at rest, workload was increased every 2 min with increments of 25 watt until maximal effort was achieved. Maximal effort/peak exercise was judged by participants and physicians when participants were not able to maintain 60 revolutions per minute on the ergometer at a given workload. All measurements were carried out during the same time of day. Participants were told to consume their regular diet and medication.

Myocardial perfusion

Myocardial blood flow rate was measured using cardiac ⁸²Rb-PET/CT, which allows for flow quantification in absolute terms. All measurements were performed on a Siemens Biograph mCT/PET 128-slice scanner (Siemens Medical Solutions). Measurements were taken at rest and during adenosine-induced stress. Low-dose non-contrast CT was acquired for attenuation correction. During the scan at rest, approximately 1,100 MBq of ⁸²Rb chloride from a CardioGen-82Sr-82/Rb-82 generator (Bracco Diagnostics, Inc.) was intravenously infused with a constant flow rate of 50 mL/min. List-mode 3-dimensional data acquisition was started with the tracer infusion and continued for 7 min. Static images were reconstructed with a 2.5-min delay to allow ⁸²Rb to clear from the blood pool. Maximal hyperaemia was induced with adenosine, infused at 140 μg/kg per minute for 6 min. After 2.5 min of adenosine infusion, intravenous ⁸²Rb infusion and list-mode acquisition were performed, using the same protocol as for rest. Myocardial blood perfusion quantification (millilitres per minute per gram) was performed using Cedars-Sinai QGS + QPS 2015.6 software (Cedars-Sinai Medical Center), which is based on a single-compartment model for ⁸²Rb tracer kinetics. For further details, see design paper.⁶

Derived variables

Body surface area (BSA) was calculated using the Dubois formula. Mean blood pressure was calculated as [(2 × diastolic blood pressure) + systolic blood pressure]/3. Systemic vascular resistance (SVR) was calculated as 80 × (MAP − CVP)/CO. Pulmonary vascular resistance (PVR) in Wood units was calculated as (MPAP − PCWP)/CO. Cardiac index (CI) was calculated as CO/BSA. Stroke volume index was calculated as CI/heart rate (HR). Workload corrected pulmonary capillary wedge pressure (PCWL) was calculated as PCWP/Watt achieved during exercise/kg body weight. Myocardial flow reserve (MFR) was calculated as the ratio between myocardial blood flow rate during adenosine infusion/myocardial blood flow rate at rest. The HF₂PEF score was calculated based on age, BMI, history of hypertension, history of atrial fibrillation, estimated pulmonary artery systolic pressure >35 mmHg, and E/e′ > 9.⁸

Results

Thirty-seven patients were enrolled in this study. Patients were elderly, overweight (Table 1), and had several risk factors for CV disease in accordance with the inclusion criteria. Glycated haemoglobin (HbA1c) was 59 ± 8.8 mmol/mol. MFR ranged from 1.18 to 3.68 among participants.

Table 1 Baseline characteristics

	n = 37
Age (years)	64.0 ± 9.6
Males	27 (73.0%)
BMI (kg/m2)	31.4 ± 6.0
BSA (m2)	2.1 ± 0.2
Systolic blood pressure (mmHg)	132.4 ± 15.0
Diastolic blood pressure (mmHg)	72.2 ± 9.1
Heart rate (b.p.m.)	73.1 ± 13.1
NYHA class
1	18 (58.1%)
2	12 (38.7%)
3	1 (3.2%)
Diabetes duration (years)	12.8 ± 7.6
History of hypertension	15 (40.5%)
History of atrial fibrillation	9 (24.3%)
History of COPD	2 (5.4%)
History of CAD	18 (48.6%)
History of stroke	4 (10.8%)
No. of CV risk factorsa
1	11 (30%)
2	18 (49%)
3	8 (22%)
HF2PEF score	3 [2; 5]
LVEF (%)	57.2 ± 7.5
IVS (cm)	1.3 ± 0.3
LVPWd (cm)	1.3 ± 0.2
LV mass (g)	206 ± 67
LVEDV (mL/m2)	84.0 ± 15.8
Left atrial volume index (mL/m2)	35.7 ± 9.4
E/A ratio	1.2 ± 0.5
E/e′	9.5 ± 3.1
TAPSE	2.1 ± 0.4
Use of insulin	20 (54.1%)
Use of GLP-1ra	6 (16.2%)
Use of DPP-4 inhibitors	7 (18.9%)
Use of ACE inhibitors	30 (81.1%)
Use of beta-blockers	18 (48.6%)
Use of MRA	1 (2.7%)
Use of loop diuretics	6 (16.2%)
Use of statins	34 (91.9%)
Haemoglobin (mmol/L)	8.5 ± 0.7
HbA1c (mmol/mol)	59.0 ± 8.8
eGFR (mL/min/1.73 m2)	81.6 ± 19.0
Urate (mmol/L)	0.3 ± 0.1
Total cholesterol (mmol/L)	3.6 ± 0.9
LDL cholesterol (mmol/L)	1.7 ± 0.5
HDL cholesterol (mmol/L)	1.0 ± 0.2
NT-proBNP (pg/mL)	128 [74; 303]
UACR	50.9 ± 152.4
MFR: rest	1.2 ± 0.4
MFR: stress	2.5 ± 0.8
Myocardial flow reserve	2.3 ± 0.6
PCWP (mmHg)	12 ± 4
CVP (mmHg)	7 ± 3
CI (L/min/m2)	3.5 ± 0.5

Of the 37 patients included in the study, 21 (57%) patients met invasive criteria for HFpEF. Of these, 14 patients had elevated filling pressure at rest (PCWPrest ≥ 15 mmHg), whereas the remaining seven patients only displayed elevated filling pressures during exercise, but not at rest (PCWPrest < 15 mmHg and PCWPexercise ≥ 25 mmHg). Patients without HFpEF had a median [IQR] HF₂PEF score of 3 [2; 4] compared with 3 [2; 6] in patients with HFpEF (P = 0.30).

The association of baseline characteristics with pulmonary capillary wedge pressure at rest

In univariable analyses, PCWP was associated with BMI (P = 0.04), systolic blood pressure (P = 0.04), history of hypertension (P = 0.01), and MFR (P = 0.02) (Table 2).

Table 2 The association of independent baseline variables and PCWP at rest

Rest	Univariable	Multivariable (r2 = 0.42, P = 0.001)
Coefficients (95% CI)	P value	Coefficients (95% CI)	P value
Age (years)	0.09 (−0.06; 0.23)	0.22
Male	−1.0 (−4.2; 2.1)	0.52
BMI (kg/m2)	0.23 (0.01; 0.44)	0.04
Systolic blood pressure (mmHg)	0.09 (0.00; 0.19)	0.04
Diastolic blood pressure (mmHg)	0.00 (−0.16; 0.16)	0.96
Heart rate (b.p.m.)	0.06 (−0.05; 0.16)	0.29
Diabetes duration (years)	0.0 (−0.2; 0.2)	1.0
History of hypertension	3.4 (0.7; 6.0)	0.014	2.6 (0.3; 5.0)	0.030
History of atrial fibrillation	4.2 (1.2; 7.1)	0.007	2.4 (−0.4; 5.2)	0.087
History of COPD	2.4 (−3.8; 8.5)	0.44
History of CAD	0.9 (−1.9; 3.7)	0.53
History of stroke	−2.8 (−7.2; 1.6)	0.20
No. of CV risk factorsa	0.2 (−1.9; 2.2)	0.86
LVEF (%)	0.02 (−0.17; 0.21)	0.81
IVS (cm)	−2.6 (−9.4; 4.2)	0.42
LVPW (cm)	−4.1 (−13.6; 5.4)	0.39
LV mass (g)	0.003 (−0.02; 0.02)	0.79
LVEDV (mL/m2)	−0.0 (−0.1; 0.1)	0.45
Left atrial volume index (mL/m2)	−0.0 (−0.2; 0.10)	0.50
E/A ratio	−2.4 (−5.6; 0.8)	0.14
E/e′	−0.0 (−0.6; 0.5)	0.87
TAPSE (cm)	−1.4 (−4.6; 1.8)	0.40
logUACR	0.001 (−1.33; 1.35)	0.99
Total cholesterol (mmol/L)	−0.8 (−2.4; 0.7)	0.30
LDL (mmol/L)	−0.4 (−3.2; 2.3)	0.75
HbA1c (mmol/mol)	0.09 (−0.08; 0.26)	0.28
Haemoglobin (mmol/L)	0.45 (−2.6; 1.2)	0.45
eGFR (mL/min/1.73 m2)	−0.0 (−0.1; 0.0)	0.44
Uric acid (mmol/L)	0.37 (−14.0; 21.4)	0.67
MFR: rest	1.2 (−2.0; 4.4)	0.44
MFR: stress	−0.6 (−2.4; 1.3)	0.52
Myocardial flow reserve	−2.7 (−4.9; −0.4)	0.021	−2.3 (−4.3; −0.3)	0.026
ACEi use	1.9 (−1.6; 5.4)	0.28
Beta-blocker use	2.5 (−0.2; 5.2)	0.07
Loop diuretic use	1.4 (−2.3; 5.2)	0.44
Statin use	−1.2 (−6.3; 4.0)	0.65

In multivariable analysis, only 2 out of 39 variables emerged as independent factors associated with resting PCWP: history of hypertension (P = 0.03) and MFR (P = 0.03). History of atrial fibrillation was borderline significant (P = 0.09). The relative importance of independent variables, with regard to the association with PCWP was similar: history of hypertension 33%, MFR 33%, and history of atrial fibrillation 34%. The association between MFR and PCWP (r = −0.39, P = 0.02) is depicted in Figure 1.

[IMAGE OMITTED. SEE PDF]

The association of baseline characteristics with pulmonary capillary wedge pressure at peak exercise

Participants achieved a maximal median [IQR] workload of 75 [55; 95] watt.

In univariable analyses, beta-blocker use (P = 0.04), and history of atrial fibrillation (P < 0.05) were all associated with pulmonary capillary wedge pressure during maximal exercise (Table 3).

Table 3 The association of independent baseline variables and PCWP at peak exercise

Rest	Univariable	Multivariable (r2 = 0.18, P = 0.038)
Coefficients (95% CI)	P value	Coefficients (95% CI)	P value
Age (years)	0.13 (−0.22; 0.48)	0.44
Male	−2.0 (−9.5; 5.4)	0.59
BMI (kg/m2)	−0.1 (−0.7; 0.48)	0.75
Systolic blood pressure (mmHg)	0.12 (−0.11; 0.35)	0.30
Diastolic blood pressure (mmHg)	0.06 (−0.32; 0.45)	0.74
Heart rate (b.p.m.)	0.08 (−0.18; 0.34)	0.56
Diabetes duration (years)	0.1 (−0.4; 0.5)	0.69
History of hypertension	2.8 (−3.8; 9.5)	0.40
History of atrial fibrillation	7.5 (0.1; 14.7)	0.046
History of COPD	7.8 (−6.7; 22.2)	0.28
History of CAD	2.9 (−3.7; 9.4)	0.38
History of stroke	−0.4 (−11.1; 10.3)	0.94
No. of CV risk factorsa	3.0 (−1.6; 7.5)	0.20
LVEF (%)	−0.04 (−0.49; 0.40)	0.84
IVS (cm)	2.1 (−14.1; 18.3)	0.79
LVPW (cm)	−5.3 (−28.3; 17.7)	0.64
LV mass (g)	0.008 (−0.04; 0.06)	0.75
LVEDV (mL/m2)	−0.02 (−0.23; 0.19)	0.85
Left atrial volume index (mL/m2)	−0.1 (−0.5; 0.3)	0.56
E/A ratio	−1.4 (−8.8; 5.9)	0.69
E/e′	−0.4 (−1.6; 0.9)	0.55
TAPSE (cm)	−0.1 (−8.6; 6.7)	0.80
logUACR	−1.7 (−4.8; 1.35)	0.26
Total cholesterol (mmol/L)	−2.8 (−6.3; 0.8)	0.13
LDL (mmol/L)	−3.3 (−9.6; 3.0)	0.29
HbA1c (mmol/mol)	−0.04 (−0.43; 0.35)	0.84
Haemoglobin (mmol/L)	−3.26 (−7.7; 1.2)	0.14
eGFR (mL/min/1.73 m2)	−0.1 (−0.1; 0.2)	0.54
Uric acid (mmol/L)	.17.2 (−58.4; 23.9)	0.40
MFR: rest	3.3 (−4.2; 10.8)	0.38
MFR: stress	−0.1 (−4.4; 4.3)	0.97
Myocardial flow reserve	−4.6 (−10.0; −0.9)	0.095	−4.4 (−9.6; 0.8)	0.091
ACEi use	3.3 (−5.1; 11.7)	0.43
Beta-blocker use	6.6 (0.3; 12.8)	0.041	6.1 (−0.1; 12.4)	0.053
Loop diuretic use	−6.1 (−14.9; 2.6)	0.17
Statin use	−0.2 (−12.4; 12.0)	0.97

In multivariable analysis, two variables emerged as borderline independent factors associated with PCWP during exercise: myocardial flow reserve (P = 0.09) and beta-blockers use (P = 0.05).

Discussion

This study sought to expand knowledge on the relative importance of coronary microvascular dysfunction as assessed by the myocardial flow reserve (MFR) in determining left ventricular filling pressure in patients with T2D and risk factors for cardiovascular disease. An extensive invasive and non-invasive characterization of the included patients was utilized to study this question. We were successful in including patients with a spectrum of left ventricular filling pressures ranging from normal to highly abnormal both at rest and during exercise. This allowed us to discern which baseline characteristics—including MFR—that were associated with PCWP and allude to their significance in the transgression from normal to abnormal left ventricular filling pressures.

Among all included participants, 57% fulfilled invasive diagnostic criteria for HFpEF despite that none had a clinical diagnosis of HFpEF at inclusion. This hints both at the difficulty of diagnosing patients with HFpEF, but also that patients with T2D and CV risk factors are at a high risk of either having or developing HFpEF.

Our expanded phenotypic profiling allowed us to associate invasive measures with a range of baseline characteristics, including MFR. We found that three independent baseline characteristics were associated with PCWP at rest: history of hypertension, MFR, and atrial fibrillation (borderline significance). Hypertension is highly prevalent in patients with HFpEF as consistently reported.^9,10 As measured blood pressure at examination was not an independent variable, this suggests that the long-term effects of hypertension are necessary to induce increased PCWP. Atrial fibrillation is prevalent among patients with increased PCWP, and although only a borderline association was noted in this study, atrial fibrillation has previously been recognized as an HFpEF risk factor, making it an integral part of HFpEF screening schemes.^8,11 As such, this finding is in compliance with previous reports and underscores the importance of including this information in HFpEF screening tools, such as the H₂FPEF and HFA-PEFF diagnostic scores.^8,11

People with diabetes have increased risk of coronary atherosclerosis compared with non-diabetic peers, and both coronary artery disease and microvascular dysfunction are prevalent in patients with HFpEF.^12,13 In this prospective study, we were able to test whether MFR was independently associated with PCWP at rest in patients across a wide spectrum of PCWP measurements; from normal to abnormal. We found a significant inverse association between MFR and PCWP at rest. It is worth noting that the coefficient of 2.3 mmHg in PCWP per unit change of MFR is clinically significant. Among our included patients, MFR ranged from 1.18 to 3.68, which corresponds to a difference in PCWP of approximately 6 mmHg between patients with the lowest MFR compared to those with the highest MFR.

These prospective data add to a prior retrospective report showing that PCWP in patients with exertional dyspnoea (43% were diagnosed with HFpEF) was associated with MFR assessed by coronary reactivity testing.¹⁴ Our data expand on this knowledge using a gold standard ⁸²Rb-PET/CT that assesses global MFR, rather than a segment of MFR as previously reported.¹⁴ In our prospective cohort of patients with diabetes at high risk of CV disease with resting PCWP across a wide range from normal to abnormal, the association with MFR was even stronger than previously reported.¹⁴ Preventing progression of coronary atherosclerosis in patients at risk of HFpEF would seem like a reasonable target in light of these and other data. Statin use in this population would theoretically attenuate the progression of coronary atherosclerosis and lead to fewer symptoms of increased left ventricular filling pressure over time compared to non-statin users, which retrospective data indicate.¹⁵

During peak exercise, the association of baseline characteristics with the resultant PCWP was attenuated. MFR and use of beta-blockers were borderline associated with PCWP during exercise. Beta-blocker use was associated with an increased PCWP of 6 mmHg during peak exercise. A recent study showed that cessation of beta-blocker use in HFpEF lead to higher exercise capacity.¹⁶ Perhaps this was in part due to a lower PCWP at a given workload after beta-blocker cessation; however, other studies have shown conflicting results pertaining to beta-blocker use.¹⁷

Strengths and limitations

This study used multiple invasive and non-invasive measures to assess haemodynamics. This was a study with a moderate number of participants, which is an important context when assessing the results. Due to technical reasons, data collection (e.g., right heart catheterization and ⁸²Rb-PET/CT) did not take place simultaneously, but within a short time span of days, which may have weakened associations. This study explored associations rather than causal relationships, making our conclusions hypothesis-generating. The measure of MFR reflects the status of both the macro- and microvascular cardiac circulation. However, no difference in MFR was noted in patients with and without CAD (data not shown).

Conclusions

In patients with T2D and risk factors for cardiovascular risk, MFR was associated with PCWP across the range from normal to abnormal left heart filling pressures.

Statistical analyses

Numerical values are reported as mean ± SD, median [IQR], or counts (%). Univariable linear regression models using PCWP as the dependent variable and baseline characteristics at rest as explanatory variables were tested. Variables were tested for normality and log-transformed if skewed in distribution, which was the case for UACR. Due to collinearity with PCWP, pulmonary artery pressures, CVP, CO, and NT-proBNP were not included in the models. A stepwise forward selection multivariable linear regression model (cutoff: P < 0.1) was used both at rest and maximal exercise with PCWP as the dependent variable and baseline characteristics at rest as explanatory variables. Dominance analysis was used to obtain the proportion of fit metric that was attributable to each independent variable (STATA package domin¹⁸). A P value of less than 0.05 was considered statistically significant. All analyses were conducted using STATA version 15 (College Station, TX).

Conflict of interest

CK has served on scientific advisory panels and received speaker fees from Boehringer Ingelheim, Merck Shape and Dome, Astra Zeneca, Amgen, Novartis, Novo Nordisk, and Shire. PR has received consultancy and/or speaking fees (to his institution) from Astellas, AstraZeneca, Bayer, Boehringer Ingelheim, Eli Lilly, Gilead, Novo Nordisk, and Sanofi Aventis and research grants from AstraZeneca and Novo Nordisk. SEI has received consultancy and/or speaking fees from Boehringer Ingelheim, AstraZeneca, Novo Nordisk, Merck, and Esperion. EW has received speaker fees from Orion Pharma, Novartis Healthcare, Boehringer-Ingelheim, and Merck. The remaining authors have nothing to disclose.

Funding

This work is supported by the Department of Internal Medicine at Herlev Hospital, the Research Council of Herlev Hospital, The Danish Heart Foundation, grant number 16-R107-A6697, The Hartmann Foundation; The Toyota Foundation, and by a Steno Collaborative Grant 2018.

References

Seferović PM, Petrie MC, Filippatos GS, Anker SD, Rosano G, Bauersachs J, et al. Type 2 diabetes mellitus and heart failure: A position statement from the Heart Failure Association of the European Society of Cardiology. Eur J Heart Fail 2018;20:853–872. doi:10.1002/ejhf.1170

Taqueti VR, Solomon SD, Shah AM, Desai AS, Groarke JD, Osborne MT, et al. Coronary microvascular dysfunction and future risk of heart failure with preserved ejection fraction. Eur Heart J 2018;39:840–849. doi:10.1093/eurheartj/ehx721

Murtaza G, Virk HUH, Khalid M, Lavie CJ, Ventura H, Mukherjee D, et al. Diabetic cardiomyopathy ‐ A comprehensive updated review. Prog Cardiovasc Dis 2019;62:315–326. doi:10.1016/j.pcad.2019.03.003

Yang JH, Obokata M, Reddy YNV, Redfield MM, Lerman A, Borlaug BA. Endothelium‐dependent and independent coronary microvascular dysfunction in patients with heart failure with preserved ejection fraction. Eur J Heart Fail 2020;22:432–441. doi:10.1002/ejhf.1671

Jackson AM, Rørth R, Liu J, Kristensen SL, Anand IS, Claggett BL, et al. PARAGON‐HF committees and investigators. Diabetes and pre‐diabetes in patients with heart failure and preserved ejection fraction. Eur J Heart Fail 2021; doi:10.1002/ejhf.2403

Jürgens M, Schou M, Hasbak P, Kjær A, Wolsk E, Zerahn B, et al. Design of a randomised controlled trial of the effects of empagliflozin on myocardial perfusion, function and metabolism in type 2 diabetes patients at high cardiovascular risk (the SIMPLE trial). BMJ Open 2019;9: [eLocator: e029098].

McDonagh TA, Metra M, Adamo M, Gardner RS, Baumbach A, Böhm M, et al. ESC guidelines for the diagnosis and treatment of acute and chronic heart failure. Eur Heart J 2021;2021:3599–3726. doi:10.1093/eurheartj/ehab368

Reddy YNV, Carter RE, Obokata M, Redfield MM, Borlaug BA. A simple, evidence‐based approach to help guide diagnosis of heart failure with preserved ejection fraction. Circulation 2018;138:861–870. doi:10.1161/CIRCULATIONAHA.118.034646

Anker SD, Butler J, Filippatos G, Ferreira JP, Bocchi E, Böhm M, et al. Empagliflozin in heart failure with a preserved ejection fraction. N Engl J Med 2021;385:1451–1461. doi:10.1056/NEJMoa2107038

Solomon SD, McMurray JJV, Claggett B, de Boer RA, DeMets D, Hernandez AF, et al. Dapagliflozin in heart failure with mildly reduced or preserved ejection fraction. N Engl J Med 2022; doi:10.1056/NEJMoa2206286

Pieske B, Tschöpe C, de Boer RA, Fraser AG, Anker SD, Donal E, et al. How to diagnose heart failure with preserved ejection fraction: The HFA‐PEFF diagnostic algorithm: A consensus recommendation from the Heart Failure Association (HFA) of the European Society of Cardiology (ESC). Eur Heart J 2019;40:3297–3317. doi:10.1093/eurheartj/ehz641

Rush CJ, Berry C, Oldroyd KG, Rocchiccioli JP, Lindsay MM, Touyz RM, et al. Prevalence of coronary artery disease and coronary microvascular dysfunction in patients with heart failure with preserved ejection fraction. JAMA Cardiol 2021;6:1130–1143. doi:10.1001/jamacardio.2021.1825

Shah SJ, Lam CSP, Svedlund S, Saraste A, Hage C, Tan R‐S, et al. Prevalence and correlates of coronary microvascular dysfunction in heart failure with preserved ejection fraction: PROMIS‐HFpEF. Eur Heart J 2018;39:3439–3450. doi:10.1093/eurheartj/ehy531

Ahmad A, Corban MT, Toya T, Verbrugge FH, Sara JD, Lerman LO, et al. Coronary microvascular dysfunction is associated with exertional haemodynamic abnormalities in patients with heart failure with preserved ejection fraction. Eur J Heart Fail 2021;23:765–772. doi:10.1002/ejhf.2010

Marume K, Takashio S, Nagai T, Tsujita K, Saito Y, Yoshikawa T, et al. Effect of statins on mortality in heart failure with preserved ejection fraction without coronary artery disease ‐ report from the JASPER study. Circ J Off J Jpn Circ Soc 2019;83:357–367. doi:10.1253/circj.CJ‐18‐0639

Palau P, Seller J, Domínguez E, Sastre C, Ramón JM, La Espriella R, et al. Effect of β‐blocker withdrawal on functional capacity in heart failure and preserved ejection fraction. J Am Coll Cardiol 2021;78:2042–2056. doi:10.1016/j.jacc.2021.08.073

Conraads VM, Metra M, Kamp O, De Keulenaer GW, Pieske B, Zamorano J, et al. Effects of the long‐term administration of nebivolol on the clinical symptoms, exercise capacity, and left ventricular function of patients with diastolic dysfunction: Results of the ELANDD study. Eur J Heart Fail 2012;14:219–225. doi:10.1093/eurjhf/hfr161

Azen R, Budescu DV. The dominance analysis approach for comparing predictors in multiple regression. Psychol Methods 2003;8:129–148. doi:10.1037/1082‐989x.8.2.129