Patients referred for MPGS with previous or non-coronary artery disease (CAD) and without history of myocardial infarction were prospectively included from March to September 2011. All patients underwent a complete physical exam before inclusion. Exclusion criteria were documented allergies to dipyridamole, asthma, systolic blood pressure < 90 mmHg, decompensated heart failure or acute angina, significant valvular disease, hypertrophic cardiomyopathy, atrial fibrillation and the possibility to induce stress by effort. All patients were informed of the study design and agreed to the protocol before inclusion. Patients with identified MPGS ischemia were also excluded of the analysis.

All enrolled patients underwent MPGS with intravenous dipyridamole pharmacologic stress using the standardized protocols from the European Association of Nuclear Medicine / European Society of Cardiology guidelines 9 . Caffeinated beverages, foods and medications, and medications containing methylxanthine were avoided for at least 12 hours prior to stress testing. Dipyridamole was given as a continuous infusion intravenously at 0.6 mg/kg over the course of 4 minutes. Arterial pressure was recorded before infusion, every 2 minutes during stress and also at the peak of dipyridamole effects (8 minutes). Stress MPGS was performed for all patients with a weight-adjusted dose of 300–400 MBq of 99mTc-tetrofosmin injected 3 minutes after the completion of the dipyridamole infusion. MPGS was acquired 15 to 30 minutes after a radiotracer injection using a Symbia T6 (Siemens Healthcare, Erlangen, Germany) double-headed gamma camera equipped with low-energy, high-resolution collimators. Data was acquired for 180° with 64 frames of 30 and 20 second durations at stress and at rest, respectively; a 64 × 64 matrix; 8-frame gating; and a 20% window centred on the 140-keV photo peak of Tc-99m. Rest MPGS was performed only if stress MPGS was considered as pathological, with a 2-fold-higher dose of 99mTc-tetrofosmin injected at least 3 hours after the stress testing. All patients underwent low-dose CT using a Symbia T6 system (Siemens Healthcare, Erlangen, Germany) for attenuation correction (130 keV, 30 to 45 mAs).

Echocardiography for all patients was performed at rest and 3 minutes after the completion of the dipyridamole infusion, i.e. at peak of stress, with a Imagic KM 60 (Kontron Medical, Saint-Germain en Laye, France) using a 2.5 MHz transducer. A complete two-dimensional grey scale echocardiography including the three standard apical views (four, three and two chambers) with a frame rate > 75 frame/s was performed for each patient. Left ventricle ejection fraction (LVEF) and volumes were assessed before and after the dipyridamole infusion as diastolic parameters. Myocardial strain was measured using speckle-tracking echocardiography.

The 17-segments model, as defined by the American Society of Echocardiography, was used to examine both echocardiography and MPGS 10 . The apex segment was then excluded for the analysis.

For the echocardiography, digital data of 3 consecutive heart cycles were recorded and transferred to a personal computer with My Lab Desk workstation (Kontron Medical, Saint-Germain en Laye, France) for offline analysis. The endocardial border was defined manually in end systole and automatically tracked frame by frame. Operator assessed optimal evaluation of both quality of tracking and region of interest. Global longitudinal strain (GLS) was obtained by averaging all segmental longitudinal strain curves computed from the conventional apical two-, three- and four-chamber views. Longitudinal strain reserve (LSR) was defined by the difference between peak systolic global longitudinal strain at the peak of vasodilatation with dipyridamole and at rest. The longitudinal strain rate was determined for the left ventricle as the maximal strain rate value (calculated as the temporal derivative of strain) during the ejection phase. The longitudinal strain rate reserve (LSRR) was similarly obtained (Figure 1). Left ventricular ejection fraction (LVEF) was assessed by transthoracic echocardiography using the conventional apical two- and four-chamber views and the modified Simpson’s method.

For MPGS studies, off-line analysis was performed on Syngo MI Applications software (Siemens Healthcare, Erlangen, Germany). The images were assessed visually and by applying automated methods. A single blinded observer interpreted echocardiographic and MPGS studies. Myocardial ischemia was defined by at least one reversible myocardial perfusion defect between stress and rest myocardial perfusion gated-SPECT and was expressed by the number of segments affected.

Data about the occurrence of adverse events were obtained from medical records by direct patients’ interviews or from the referring physician. The primary end point was defined by all-cause mortality. Patients unable to be interviewed up to 6 months at the date of follow-up were considered as lost to follow-up.

Data were expressed as mean +/− SD. Nominal values were expressed as numbers and percentages. Normality was tested by the Kolmogorov-Smirnov test. The association between the mean values of continuous normally distributed variables were compared using unpaired and paired Student’s t test and the Mann–Whitney rank sum test was used when the samples were not normally distributed or had unequal variances. Comparison between multiple groups was performed with a variance analysis (ANOVA). Nominal variables were investigated by the χ² test. Linear regression analysis was used to investigate the relation between LSR-LSRR and variables. Conventional variables correlated with LSR with a p value < 0.05 at first univariate analyses were used to build the final multivariate stepwise model. Receiver operating characteristic (ROC) curves were computed to determine optimal cut-off point for longitudinal strain reserve as well as to calculate area under the curve (AUC) to determine prognostic significance. Multivariate Cox regression model was built to identify echocardiographic parameters associated with all-cause mortality. Survival curve was determined according to the Kaplan-Meier method, and cumulative event rates compared by means of the log-rank test. Differences were considered statistically significant for p-values of < 0.05. All analyses were performed on SPSS software version 20 (SPSS Inc., Chicago, Illinois).

Eighty patients were prospectively included. Four patients (5%) were excluded from the analysis due to a poor resolution of 2D echocardiography as a consequence of poor ultrasonic window (obesity or pulmonary disease) that did not allow speckle-tracking imaging and 1 (<1%) due to refusal of gated-SPECT after echocardiography. Among the 75 other patients included, 12 were excluded for MPGS ischemia with at least one or more reversible defect segment (Figure 2). Male represented 59% of the 63 patients finally enrolled in the study. The mean age was 70 ± 11 years with a median age of 71 ranging from 46 to 90 years old. Twenty-six patients (41%) had a previous history of coronary artery disease and 28 (44%) suffered from diabetes. Diabetes lasted less than five years in 10/28 patients. Thirty-nine patients (62%) reported a NYHA stage 2 and the mean LVEF was 51±14%. Baseline characteristics are presented in Table 1.

Values are presented as n (%) or mean +/− SD.

BMI = body mass index; SBP = systolic blood pressure; DBP = diastolic blood pressure; HR = heart rate; ACE = angiotensin-converting enzyme; ARB = angiotensin receptor blocker; LVEF = left ventricle ejection fraction; LVEDV and LVESV = left ventricle end diastolic and end systolic volume.

Hemodynamic during dipyridamole infusions and echocardiographic examinations remained unchanged for all patients. The decrease of systolic blood pressure with dipyridamole after stress testing was not only insignificant for the whole population (136±22 at rest vs. 130±21 mmHg after dipyridamole infusion, p = 0.14) but the diabetic patients in comparison to the non-diabetics also showed a non-significant variation of systolic blood pressure (−3.8±12.4 vs. -2.9±31.6 mmHg; p = 0.88, respectively). Results are similar regarding the diastolic blood pressure (p = 0.09). The mean heart rate increased from 70±14 beats/min at rest to 80±18 beats/min after the dipyridamole infusion (p < 0.001) showing the pharmacological effect of dipyridamole. No examination had to be stopped for safety reasons.

The effects of dipyridamole on LSR according to baseline characteristics, coronary risk factors, dyspnea and medications are presented in Table 2. In our general population, the mean GLS before dipyridamole infusion was −14.5±4.2% and reached −16.8±4.5% at the maximum effect of vasodilatation. Consequently, the mean LSR was −2.28±2.19%. LSR did not depend on systolic blood pressure (p = 0.99), diastolic blood pressure (p = 0.57) or heart rate (p = 0.85) changes during stress, as LSRR with p-values of 0.89, 0.57 and 0.17, respectively for systolic blood pressure, diastolic blood pressure and heart rate.

LSR = longitudinal strain reserve; BMI = body mass index; ACE = angiotensin-converting enzyme; ARB = angiotensin receptor blocker.

*: Spearman correlation.

†: ANOVA.

By univariate analysis, only age was associated with a decrease of LSR after dipyridamole infusion whereas patients with diabetes, higher Body Mass Index (BMI) and current smoking showed an improvement of LSR (Table 3). Increasing age was significantly correlated to a decrease of LSR (p < 0.0001) as presented in Figure 3. As shown in Table 4, no difference was observed between diabetic and non-diabetic patients for GLS before stress (−13.9±3.7 vs. -15.0±4.5%; p = 0.30) and after the dipyridamole infusion (−17.2±4.2 vs. -16.5±4.8%; p = 0.55) but LSR was higher in the diabetic population (−3.27±1.93 vs. -1.49±2.08%; p = 0.001). Moreover, GLS of diabetic increased significantly by 24% in stress conditions (p = 0.003). Among the 28 patients with diabetes, 22 of them presented also overweight, defined as a BMI > 25 kg/m2 (p < 0.004). GLS before dipyridamole infusion was not different between patients with or without overweight (−14.2±3.6 vs. -15.0±4.9%; p = 0.43) but LSR was significantly higher in patients with overweight (−2.83±1.85 vs. -1.50±2.42%; p = 0.016). After a multivariate analysis, only age (p = 0.001) remained independently associated with a decrease of LSR after the dipyridamole infusion. Conversely, LSR remained significantly improved only in diabetic patients (p = 0.008). Among all echocardiographic parameters at baseline and after stress, including systolic, diastolic, hemodynamic and speckle-tracking parameters, only LSR was modified according to the diabetic status (Table 4). LSR was not correlated to the duration of diabetes (p = 0.80) or HbA1c level (p = 0.21) and was not influenced by dedicated treatments especially insulin therapy (p = 0.46), the presence of retinopathy (p = 0.43) or peripheral vascular disease (p = 0.34).

BMI = body mass index; ACE = angiotensin-converting enzyme; ARB = angiotensin receptor blocker; MPGS = myocardial perfusion gated SPECT; CAD = coronary artery disease.

LVEF = left ventricle ejection fraction; EDV = end diastolic volume; ESV = end systolic volume, TVI LVOT = time-velocity integral in the left ventricular outflow tract, GLS = global longitudinal strain; DT = mitral deceleration time, LSR = longitudinal strain reserve; LSRR = longitudinal strain rate reserve.

LSRR was only associated with aging (p = 0.005) and was not influenced by diabetes (p = 0.57). Moreover, longitudinal strain rate increased by 16% between baseline and peak stress in the diabetic population (p = 0.11).

During a mean follow-up of 15±5 months, 6 (10%) patients reached the primary endpoint. Only one patient was lost to follow-up and was excluded for survival analysis. After multivariate analysis, only the left ventricle end systolic volume at stress (HR: 1.047 [95% C.I: 1.018 – 1.075]; p = 0.001; Table 5), the difference of left ventricle end systolic volumes between stress and rest (HR: 1.065 [95% C.I: 1.019 – 1.113]; p = 0.005) and a positive LSR (HR: 15.493 [95% C.I: 1.419 – 169.182]; p = 0.025) remained independently associated with all-cause mortality. ROC curve analysis in Figure 4 identified positive LSR (cut-off value of 0%) as a predictor of all-cause mortality with a sensitivity of 89% and a specificity of 50%, for an area under the curve of AUC = 0.79 (p = 0.021). The Kaplan-Meier analysis showed better survival in patients with a negative LSR (log-rank p = 0.012). All cause-mortality in the diabetic population was associated with a lower LSR (−0.59±1.17 vs. -3.60±1.77%; p = 0.028) but prognosis was not significantly better in the diabetic population compared to the non-diabetic since they have a better LSR (p = 0.102 vs. p = 0.446).

LVEF = left ventricle ejection fraction; EDV = end diastolic volume; ESV = end systolic volume, TVI LVOT = time-velocity integral in the left ventricular outflow tract, GLS = global longitudinal strain; LSR = longitudinal strain reserve; LSRR = longitudinal strain rate reserve.