Strong cardiovascular prognostic implication of quantitative left atrial contractile function assessed by cardiac magnetic resonance imaging in patients with chronic hypertension

We studied a consecutive series of patients with history of chronic hypertension medically treated for at least 6 months who were referred to undergo cardiac magnetic resonance imaging (CMR) for clinical purposes. Patients were referred either for a) evaluation for myocardial ischemia with stress CMR or b) assessment of regional and global left ventricular function and myocardial mass. Patients with any of the following were excluded: a) any evidence of myocardial infarction by history, medical record, or abnormal cardiac enzymes, b) any significant aortic or mitral valvular dysfunction (moderate or severe dysfunction by qualitative echocardiographic grading), and c) confirmed (by biopsy) myocarditis, infiltrative cardiomyopathy (including cardiac hemochromatosis, amyloidosis, or sarcoidosis), or pericardial disease. Other exclusion criteria included concurrent unstable angina, NYHA class IV heart failure, hemodynamic instability, claustrophobia precluding CMR, and metallic hazards. Patients with patterns of late gadolinium enhancement consistent with infiltrative cardiomyopathy or myocarditis were also excluded. The institutional ethics committee of the Brigham and Women's Hospital (Partners Healthcare system) approved the clinical follow-up activities of the study.

All patients underwent a detailed history immediately before the CMR. Hypertension, hypercholesterolemia, diabetes, and family history of premature CAD were defined by published criteria[4‐7]. Significant smoking was defined by >10 pack-years of tobacco use. History of CAD included documented >70% stenosis on angiography or history of coronary revascularization prior to CMR study.

All patients were studied supine in a 1.5T CMR system (Signa^® CV/i, GE Healthcare, USA) with a 4-element or 8-element phased-array surface coil. CMR study consisted of cine SSFP imaging (TR/TE 3.4/1.2ms, in-plane spatial resolution 1.6 × 2 mm, matrix 192 × 160) of LV function and late gadolinium enhancement imaging (TR/TE 4.8/1.3ms, TI 200-300ms) for myocardial scar. All images were acquired using retrospective ECG gating and breath-holding. Cine and late enhancement imaging were obtained in 8-14 matching short-axis (8 mm thick with 0 mm spacing) and 3 radial long-axis planes. The 3 radial long-axis planes were prescribed at 60 degrees apart. Typical view per segment during a cine SSFP acquisition was 12 yielding a temporal resolution of approximately 45 ms and maintaining a breath-hold of approximately 10-12 seconds for each slice location. A previously described segmented inversion-recovery pulse sequence for late enhancement imaging was used[8] starting at 15 minutes after cumulative 0.15 mmol/Kg dose of gadolinium-DPTA. A single reader categorized late gadolinium enhancement as either typical infarction (involving the subendocardium) or atypical (subepicardial, patchy midwall or diffuse circumferential subendocardial pattern).

Quantitative Analysis of the left atrium is illustrated in Figure 1. To measure the left atrial dimensions, manual tracings were made of the left atrial area and long axis in the radial 2-chamber and 4-chamber views. For each radial view, tracings were performed at three phases: maximal LA volume just before mitral valve opening, minimal LA volume at mitral valve closure, and immediately prior to atrial contraction. At each phase, LA volume is calculated by the previously validated biplane area-length method[9] as follows: LA volume (ml) = 0.85*A_2C*A_4C/L, where A_2C and A_4C are the LA areas on the 2-chamber and 4-chamber views, respectively, and L is the shorter long-axis length of the LA from either the 2-chamber or the 4-chamber views (Figure 1). Consistent with the recommendation of published reports[10], all LA volumes were normalized to the patient's body surface area in subsequent analyses. Total LA emptying volume was calculated as the difference between maximum (LAV_max) and minimum LA volumes (LAV_min). Total LA emptying volume was divided into LA passive emptying volume (V_Passive) and LA contractile volume (V_Contractile), in which V_Passive was calculated as the difference between LAV_max and the LA volume preceding atrial contraction (LAV_ac) and V_Contractile was calculated as the difference between LAV_ac and LAV_min. LA total, passive, and active emptying functions (LAEF_Total, LAEF_Passive, LAEF_Contractile) were calculated according to the following formulas:

In additon, we calculate the proportional contribution of LA contraction during diastole by calculating the following parameters:

While left atrial dimension (in millimeters) was made available to the attending physicians on the day of the CMR, all other quantitative measurements of the left atrium were not reported as part of the routine clinical care.

All images were analyzed with specialized software (CineTool 2.80, General Electric Healthcare). We graded segmental systolic wall motion as normal or abnormal in each study and also graded segmental wall motion using a 4-point scale (1 = normokinesia, 2 = hypokinesia, 3 = akinesia, and 4 = dyskinesia) according to the standard 17-segment ACC/AHA nomenclature[11]. We interpreted late gadolinium myocardial enhancement (LGE) as present or absent. Details of segmental wall motion and LGE grading were as previously reported[12]. We manually traced epicardial and endocardial borders of matching short-axis cine locations at end systole and end-diastole to determine the LV ejection fraction (LVEF), end-diastolic volume index (LVEDVI), end-systolic volume index (LVESVI), and the LV myocardial mass (end-diastole only)[13, 14]. LVEF was measured by standard Simpson's Rule[14]. Presence of ischemia during dobutamine stress was defined by standard criteria of worsening regional wall motion by at least 1 grade, matching on short and long-axis cine, as published in prior reports[15]. Presence of ischemia during adenosine stress perfusion was according to prior publication, defined by existence of perfusion defect without infarction by LGE imaging[16].

Resting 12-lead ECGs were obtained at a median of 1 day (interquartile range: 0-7 days) from CMR. We confirmed that no cardiac event or revascularization occurred between the collection of ECG and the CMR study. ECG interpretation was first performed by computer analysis followed by visual over-reading by a single reader blinded to the CMR results and the clinical outcome. ECG left atrial enlargement was defined by a terminal negative P-wave duration of > 40 ms and depth ≥ 1 mm measured on lead V1. We used the Sokolow-Lyon index to indicate LV hypertrophy on ECG[17].

At least 6 months following the CMR, clinical information was obtained from patient telephone interviews, contacting patients' physicians, and hospital records. A standard questionnaire was used during telephone interview. Survival was obtained from the National Social Security Death Index in patients lost on first contact[18]. Major adverse cardiac events (MACE) included any of the following: 1) all-cause mortality, 2) new acute myocardial infarction, 3) unstable angina requiring hospitalization, and 4) development or progression of heart failure requiring hospitalization. New acute myocardial infarction was defined by significant elevation of serum troponins consistent with myocardial injury. Unstable angina was defined by new chest pain hospitalization without non-cardiac origin of chest pain, and with either angiographic coronary stenosis of ≥70% or ischemia on noninvasive imaging. Heart failure was defined by a need for hospitalization for new or worsening symptoms of heart failure as determined by the patient's cardiologist or primary internist. When a patient experienced >1 MACE, the first event was chosen. When ≥2 MACE occurred simultaneously, the worse event was chosen (death>MI>unstable angina>congestive heart failure).

Baseline demographic differences, classified by the median LAEF_Contractile, were compared by Student's t-test or Fisher's exact test. Kaplan-Meier distributions for MACE, all-cause mortality, and nonfatal events were stratified by the median value of LAEF_Contractile and were compared by log-rank tests. We fitted Cox proportional-hazards models to estimate the likelihood ratio chi-square (LRχ²) and the unadjusted hazard ratios (HR) of all the variables. We also assessed the univariable association of the proportional contribution of LA contraction during diastole (Contractile/Passive and Passive/Total ratios) with MACE and other other events. To determine the set of variables that were most strongly associated with LAEF_Contractile, we performed linear regression using LAEF_Contractile as a continuous dependent variable, with all variables in Table 1 treated as independent variables. The prognostic association of LA size measurements and mechanical function estimates were also determined using similar analyses for all-cause mortality and non-fatal events specifically. We tested the interobserver agreement of LAEF_Contractile by Spearman correlation.

We performed 2 separate multivariable approaches to analyze the predictive value of CMR variables for MACE and all-cause mortality. In the first approach, we sought to determine the strongest set of variables that were associated with MACE, all-cause mortality, and non-fatal events, respectively, in this study cohort. We used a stepwise forward selection strategy and considered all clinical, ECG, or left ventricular variables as listed in Table 2. A p-value of 0.1 was used as criteria for variable inclusion or exclusion. In the second approach, we aimed to determine the prognostic association, if any, of LAEF_Contractile after adjustment to well-known risk predictors. Therefore, we sought to determine if LAEF_Contractile remained a significant predictor after adjustment to patient age, gender, minimal left atrial volume, and left ventricular ejection fraction. The validity of the proportional-hazards assumptions of this final model for MACE was again tested and was valid for all the variables in this model. All analyses were performed with SAS 9.1 (SAS Institute, Cary, N.C.) for Windows.

From a consecutive series of 225 patients with a history of chronic hypertension, 3 (1%) had unsuccessful CMR due to large body habitus or technical problem. Seven patients (3%) had moderate or severe aortic or mitral valvular dysfunction and were excluded from the study. Five patients were lost to follow-up but were reported alive. The remaining 210 (123 male, mean age 52 ± 16 years) formed the study cohort and were followed for a median 19 months (range 6-50 months). Patients were referred to undergo stress CMR for evaluation of ischemia (n = 106) or for assessment of left ventricular function (n = 104). Table 1 illustrates the demographical and ECG features of the entire study cohort and also data stratified by the median value of LAEF_Contractile (33%). Baseline average left ventricular ejection fraction at rest was low normal at 58 ± 13%. During the study period, 53 patients were referred to undergo coronary angiography at a median 34 days (interquartile range 1-12 days) from CMR study. Among them 38 patients (72%) were found to have significant (>70%) luminal stenosis. LAEF_Contractile dichotomized by its median level did not show correlation to the presence of angiographic coronary stenosis, history of percutaneous coronary intervention, or history of cardiac bypass surgery. Patients with LAEF_Contractile below median value demonstrated a larger LA antero-posterior dimension (42 ± 9.1 mm vs. 38 ± 6.2 mm, P < 0.01), lower left ventricular ejection fraction (55 ± 15% vs. 61 ± 10%, P < 0.01), and higher left ventricular end-systolic volume index (79 ± 57 ml/m² vs. 65 ± 41 ml/m², P < 0.05). There was high inter-observer correlation in quantifying LA volume (kappa statistic 0.83) across the three time phases of the cardiac cycle.

At a median follow-up of 19 months (range 6 to 47 months), 48 MACE occurred. Among them were 21 cases of death, 3 acute myocardial infarction, 12 cases of hospitalization for unstable angina, and 12 cases of congestive heart failure requiring hospitalization.

Twenty patients (10%) demonstrated left atrial enlargement on ECG. The presence of left atrial enlargement by ECG was only weakly associated with a higher than median LAV_min (P = 0.045) and did not demonstrate significant correlation with above median LAV_max. Left atrial enlargement by ECG demonstrated poor sensitivity (16%) but excellent specificity (93%) in identifying patients with above median LAV_min. Left atrial enlargement by ECG could not differentiate above or below median level of LAEF_Contractile. While LV mass (g) is slightly higher in patients with above median LAV_min (134 ± 44 vs. 152 ± 55, P = 0.01), it was not significantly different among patients with above or below median level of LAEF_Contractile. Left ventricular ejection fraction was negatively correlated with LAV_min (r = -0.19, P = 0.008) and positively correlated with LAEF_Contractile (r = 0.15, p = 0.04). By multivariable linear regression, a history of atrial fibrillation on ECG, LAV_min, and left ventricular ejection fraction were most strongly predictive of LAEF_Contractile.

Univariable association of all variables with MACE, all-cause mortality, and non-fatal events is listed in Table 2. In the current study cohort, presence of LVH on ECG did not demonstrate significant association with MACE or mortality but was associated with non-fatal cardiac events (HR 4.43, 95% CI 1.54-12.77, P < 0.01). LV mass measurement did not demonstrate significant association with MACE. Decreasing left ventricular ejection fraction, increasing LAV_min or LAV_max, and presence of late gadolinium myocardial enhancement were all associated with non-fatal events but did not achieved significant association with MACE or mortality. LAV_min (in ml/m²), LAEF_Contractile, and LAEF_Total demonstrated strong association with patient mortality, non-fatal events, and all MACE. Figure 2 illustrates the Kaplan-Meier curves demonstrating the time-to-event distributions for MACE, all-cause mortality, and non-fatal events, stratified by the median level of LAEF_Contractile (Figure 2A-C). The prognostic association of LAEF_Contractile with all MACE maintained its significance regardless of the presence or absence of ECG finding of LVH (Figure 2D). For every 10% reduction of LAEF_Contractile, unadjusted hazards to MACE, all-cause mortality, and non-fatal events increased by 1.8, 1.5, and 1.4-folds, respectively. By similar argument, preservation of the proportional contribution of diastolic filling by atrial contractile, as quantified by the Contractile/Passive or Contractile/Total ratio, demonstrated strong association with favorable outcome to MACE (unadjusted HR 0.94 and 0.39, respectively, both P = 0.001), with a preserved Contractile/Total ratio portended to lower likelihood to mortality (HR to mortality 0.21, P = 0.003). On the contrary, LA antero-posterior dimension only demonstrated a weak unadjusted prognostic association with non-fatal events but not with all-cause mortality (Table 2).

In the first multivariable approach, stepwise forward selection strategy demonstrated that LAEF_Contractile was the strongest multivariable predictor in each of the best overall models of MACE, all-cause mortality, and non-fatal events. For MACE prediction, LAEF_Contractile, history of cardiac bypass surgery, and left bundle branch block on ECG were selected to form the best overall model. Figure 3 presents the model LRχ² of the selected multivariate variables in each of the best overall models for MACE, all-cause mortality, and non-fatal events, respectively (Figure 3A-C). In the second multivariable approach, adjusted to the patients' age, gender, and LV ejection fraction, LAEF_Contractile maintained strong and independent prognostic association with MACE (model LRχ² increased from 4.99 to 22.02, P < 0.0001), all-cause mortality (model LRχ² increased from 4.54 to 15.88, P < 0.0001), and non-fatal events (model LRχ² increased from 5.05 to 10.58, P = 0.005).

One hundred and six patients (50%) of the cohort underwent CMR for evaluation of myocardial ischemia. Among them 52 (49%) underwent dobutamine stress cine and 54 (51%) underwent adenosine stress perfusion imaging. Nineteen (17%) of the 106 patients were found to have myocardial ischemia during stress CMR. Twenty-six (25%) of the 106 patients experienced MACE during study follow-up, among them 11 died. Mean LAEF_Contractile was not significantly different between patients with or without myocardial ischemia (33 ± 11 vs. 35 ± 11, P = 0.38). In the subgroup of patients who underwent CMR ischemia evaluation, presence of ischemia did not demonstrate any significant association with MACE, all-cause mortality, and non-fatal events (model LRχ² 0.90, 0.08, and 0,07, respectively). When presence of myocardial ischemia was entered into a model that contains LAEF_Contractile, the significant associations of LAEF_Contractile with MACE and all-cause mortality, respectively, were not altered (hazard ratios of LAEF_Contractile adjusted to ischemia, 1.97 and 2.17; P = 0.0009 and 0.004, respectively).

There are a number of study limitations. This study was a retrospective referral-based cohort study at a single center. All of the subjects studied had been referred for CMR examination, while the clinical indications of such were appropriate, the generalizability to other populations with hypertension may be limited. Only 50% of the current study cohort underwent stress CMR evaluation for myocardial ischemia. While we found that presence of myocardial ischemia did not seem to alter the prognostic significance of LA contractile function with the clinical outcomes, the current report does not have the study design or power to properly address whether myocardial ischemia could have been a confounding variable. The method of LA volume quantitation in this study can only provide estimates of LA volumes since it has not been validated against the axial stacked volumetric imaging. The study is also limited by its sample size and the small number of patients who experienced MACE, therefore some degree of model overfitting in the multivariable analyses may have occurred. For this reason, confirmation of the current study findings and exploration of potential therapeutic implications that can alter LA mechanics are necessary in a larger prospective study. Finally, concurrent load-independent measures of LV diastolic function such as tissue Doppler velocity are not available in the current study, thus prohibiting probing in the mechanism of the significant prognostic findings being reported in the current study. Such association of LA contractile function with imaging markers of diastolic dysfunction and serum markers such as naturetic peptide levels[36], will need to be studied prospectively.