Insulin resistance and exercise tolerance in heart failure patients: linkage to coronary flow reserve and peripheral vascular function

The study was a local substudy based on patients screened for inclusion to the Smartex-HF trial [16] in which patients with chronic systolic heart failure were randomized to different modalities of exercise training. The patients were recruited from the heart failure outpatient clinic at Bispebjerg University Hospital, Copenhagen, Denmark. Inclusion criteria were a LVEF < 35%, clinically stable (no signs of worsening for at least 6 weeks), minimum 3 months of optimal medical treatment, and if CRT or revascularization was performed, this should be more then 6 months prior to inclusion. All patients had been examined with either coronary angiography or cardiac CT as part of their heart failure examination program and were revascularized according to guidelines. All patients had to have a LAD without significant stenosis (<50%) in order to perform coronary flow measurements. Angiography was not repeated prior to the enrolment in to the study. However, the participants were excluded from the study if there were signs of ischemia or ventricular arrhythmias during the maximal symptom limited bicycle exercise test. The study complied with the Declaration of Helsinki and was approved by the science ethics committee for the Capital Region of Denmark (HC-2008-108). All participants gave informed written consent.

All patients underwent an upright bicycle (Via Sprint 150P, Ergoline, Bitz, Germany) exercise test with breath-by-breath gas exchange measurement (Jaeger, Masterscreen CPX, Cardinal Health, Würzburg, Germany). After 3 minutes of rest on the bicycle the test was initiated using either a protocol starting at 20 watt with a 10 watt increase pr. minute or starting at 40 watts with a 20 watt increase pr. minute, to ensure optimal exercise time based on an expectation of the patients exercise capacity. The patients were encouraged to continue until maximal exhaustion. A leveling off of oxygen uptake despite increasing workload and a respiratory exchange ratio (RER) > 1.05 were used as criteria for maximal oxygen uptake. The mean of the 3 highest consecutive 10-second measurements before exercise terminations were used for determining VO₂peak.

A complete transthoracic echocardiography was performed using Philips IE33 (Philips Medical Systems, Andover, MA, USA) with an S5 probe with the patients in the left supine position. Left ventricle end diastolic (EDV) and end systolic (ESV) volumes and ejection fraction (LVEF) were calculated from apical 2- and 4-chamber views using the biplane Simpson model. Left ventricular mass (LVM) was calculated using the formula LVM = 0.8*(1.04*(LVEDD + PWTd + SWTd)³-(LVEDD)³) + 0.6 g, where LVEDD is left ventricle end diastolic diameter, PWTd is posterior wall thickness in diastoly and SWTd is septum wall thickness in diastoly, and indexed (LVMi) to body surface area (BSA) calculated by Du Bois’ formula (BSA = 0.007184 x weight kg^0.425 x height cm^0.725).

CFR can be measured non-invasively using transthoracic Doppler echocardiography with a high success rate [17] and this technique has been validated against invasive measurements [18] with good results. CFR was measured using a high frequency broadband transducer (S8, Philips). All patients were instructed to abstain from caffeine for 12 hours before the examination and oral use of dipyridamole was paused for 72 hours. With the patient in the left supine position the LAD was located as distal as possible using color Doppler either in an apical modified two chamber view or middistal using a modified short axis view. Coronary flow velocity (CFV) was measured using pulsed wave Doppler with a sample size of 3–4 mm, at rest and during a 2 minute infusion of adenosine at 0.14 mg kg^-1 min^-1. The solution was diluted so that the infusion rate was kept at 10 ml min^-1. Before and during the adenosine infusion care was taken to maintain the position and angle of the probe, so measurements were done on the same segment of the LAD at the same angle. During the infusion the scale of flow velocity was changed in order to obtain optimal curves for offline measurement. CFR was calculated as the ratio between peak diastolic CFV during adenosine infusion and during rest using a mean of 3 consecutive cardiac cycles (10 cycles if the patient had atrial fibrillation). Analyses were done offline by an investigator blinded to the other examinations. Blood pressure and ECG were monitored before and during the adenosine infusion. Interobserver variability of CFR was tested on all subjects and the mean difference in CFR was 0.07 with limits of agreement ±0.21. The coefficient of variation (CV), calculated as the within-subject standard deviation divided by the mean of the observations, was 5.5%. Intraobserver variability was tested on 10 randomly selected examinations with a mean difference of 0.03, limits of agreement ±0.29 and CV 7.5%. Figure 2 shows measurement of CFR

Insulin sensitivity was measured using an isoglycemic hyperinsulemic glucose clamp, with an insulin infusion rate at 40 mU/min/m². Subjects met in the morning after a 12-hour overnight fast and without unusual strenuous activity for 3 days before the clamp. Blood glucose concentrations were measured in arterialized blood drawn from a catheter inserted in a dorsal hand vein. Arterialisation of the blood was achieved by placing the hand in a heating pad. Fasting blood glucose concentration was determined as the average of two blood samples, taken at t = −15 min and 0 min. Insulin was initiated by a bolus injection, and then infused continuously for 120 minutes. Glucose was measured every 5 minutes using a handheld glucose-measuring device (Accu-chek Inform, Roche Diagnostics, Indianapolis, USA). Insulin sensitivity was defined as the steady-state average glucose infusion rate during the final 30 minutes of the clamp (M-value). Metabolic clearance rate of glucose was calculated as the M-value divided by the prevailing blood-glucose concentration in order to facilitate comparisons between individuals at the steady-state condition.

A whole body dual x-ray absorptiometry scan (DEXA) (Lunar DPX-IQ, GE Lunar Corp, Madison, WI) was performed for estimation of body composition (fat mass and fat free mass).

Flow mediated vasodilation measurement was performed using Endo-PAT 2000 (Itamar, Israel) which measures arterial pulsatile volume changes in the fingertip before and after upper arm occlusion. The examination was performed in the morning in a fasting state. After 5 minutes of baseline measurements upper arm occlusion on one arm was sustained for 5 minutes using a blood pressure cuff inflated to a minimum of 200 mmHg, while the other arm served as a control. After the release of the cuff the hyperemic response was recorded and the reactive hyperemia index (RHI) was calculated as a measure of endothelial function using the automatic software taking the relative difference between basic and hyperemic blood flow on the occluded arm and dividing it with the response on the control arm, to correct for any autonomic changes in vascular diameter.

Augmentation index is derived from pulse wave analyses and is a measure of arterial stiffness. A higher augmentation index indicates more arterial stiffness [19]. Data was collected using the Endo-PAT 2000, where pulse waveforms from the digital probes obtained before upper arm occlusion were used to calculate the heart rate so the results are normalized for heart rate at 75 bpm and were calculated using the Endo-PAT 2000 software.

Unless stated otherwise values are expressed as median and inter quartile range for continuous variables and as number (percent) for categorical variables. Continuous variables were compared using Students unpaired t-test and differences in categorical variables were assessed using chi-square test. Linear regression was used to test the relationship between variables and correlations were calculated using Pearsons correlation coefficient. A multivariable linear regression using standardized coefficients (SC) was done to identify independent predictors of VO₂peak. A P value of less than 0.05 was considered statistically significant. All analyses were performed in STATA 11.1 (StataCorp. 2009. Stata Statistical Software: Release 11. College Station, Texas, USA).

Median VO₂peak was 16.2 (14.6-19.4) using total body mass and 23.5 (20.1-27.4), when using fat free mass in the calculation. No ventricular arrhythmias or signs of ischemia were seen during the test. Median test duration was 7.2 minutes (6.0-8.7) and maximum watt achieved was 90 watts (70–120). Median RER was 1.14 (1.07-1.19). VO₂peak (FFM adjusted) was higher in the patients with non-ischemic vs. ischemic heart failure (median 26.2 vs. 22.5, p = 0.05). There were no differences between patients with or without diabetes and atrial fibrillation.

Median M-value was 4.8 mg/min/kg (3.6-5.6) and glucose clearance was 4.2 ml/min/kg (3.3-5.1) when using total body mass and 6.8 mg/min/kg FFM (4.9-8.1) and 6.0 ml/min/kg FFM (4.7-7.2) respectively when using fat free mass. When comparing patients with insulin sensitivity above and below the median value 6.8 (Table 2) the patients with higher insulin sensitivity had significantly higher CFR and tended to have higher VO₂peak (p = 0.09), but otherwise there were no differences.

Median CFR was 1.77 (1.26-2.42), and there were no significant differences in CFR between patients of ischemic and non-ischemic origin (1.57 vs. 1.99, p = 0.48), patients with or without type 2 diabetes (1.59 vs. 1.88, p = 0.31), with or without mitral regurgitation (1.67 vs. 1.88, p = 0.24) or between NYHA group II and III (1.78 vs. 1.42, p = 0.48). There were also no differences in patients with or without previous myocardial infarction, percutaneous coronary intervention (PCI) and coronary artery bypass graft (CABG) operation regardless of whether the LAD was involved or not (all p > 0.05). When comparing high and low CFR divided by the median (Table 2), patients with higher CFR had higher VO₂peak and insulin sensitivity, but did not differ with regards to peripheral vascular factors. The patients with higher CFR tended to have lower EDV, ESV and LVMi, but this was not statistically significant.

Table 3 shows correlations between the different measurements. VO₂peak was correlated to insulin sensitivity (r = 0.43, p = 0.007), CFR (r = 0.48, p = 0.002), augmentation index (r = −0.35, p = 0.04) and age (r = −0.38, p = 0.02) (Figure 3). There was a correlation between CFR and insulin sensitivity (r = 0.43, p = 0.008) and augmentation index (−0.45, p = 0.007). The relationship between insulin sensitivity, CFR and VO₂peak was still present when removing the 8 patients with type 2 diabetes (r = 0.48, p = 0.006 and r = 0.49, p = 0.006 respectively). There were no significant correlations between resting (r = −0.29, p = 0.07) and hyperemic (r = 0.29, p = 0.07) CFV and VO₂peak. Insulin sensitivity was correlated to resting (r = −0.33, p = 0.04) but not to hyperemic (r = 0.20, p = 0.22) CFV. There was no significant correlation between CFR and EDV, ESV and LVMi.

Unlike augmentation index endothelial function measured by Endo-Pat 2000 was not correlated with CFR (r = −0.10, p = 0.56), insulin sensitivity (r = 0.13, p = 0.45) or VO₂peak (−0.24, p = 0.16). Left ventricular ejection fraction was not correlated to CFR (r = 0.18, p = 0.26), VO₂peak (r = 0.23, p = 0.15) or insulin sensitivity (r = 0.14, p = 0.41).

In multivariable linear regression adjusting for age (Table 4), CFR remained independently associated with VO₂peak (SC 1.98, p = 0.05) whereas insulin sensitivity (SC 1.75, p = 0.09) and arterial stiffness (SC −1.17, p = 0.29) were no longer associated with VO₂peak. The same results were achieved when using glucose clearance instead of insulin sensitivity.