Incremental Value of Left Atrial Function Analysis in the Assessment of Left Ventricular Filling Pressures in Patients with ST-Elevation Myocardial Infarction.

Background: Left atrial (LA) reservoir strain provides interesting information about left ventricular (LV) filling pressure. However, the advantages of atrial reservoir strain in comparison with conventional parameters in patients with myocardial infarction are not clear yet.
Methods: Fifty patients with ST-elevation myocardial infarction (STEMI) prospectively underwent echocardiographic assessment of LV systolic and diastolic function by conventional parameters and two-dimensional speckle tracking longitudinal strain of left atrium. LV filling pressure was estimated by brain natriuretic peptide (BNP) levels.
Results: Systolic and conventional diastolic parameters did not show significant differences between patients with increased and normal BNP values, whereas LA reservoir strain was reduced (33.1 ± 8% vs. 46.5 ± 9.8%; P = 0.001) in patients with higher BNP levels. LA reservoir strain had higher area under curve value (0.880) than the other parameters in identifying patients with elevated BNP and a cut-off value of 40.5% reached sensitivity and specificity values of 93% and 86% and positive and negative predictive values of 92% and 85%. LA reservoir strain reclassified 23 patients with increased BNP values, which were previously estimated to have normal (16 patients) and undeterminable LA pressure (seven patients) by using the recommended algorithm for diastolic function. Conclusions: LA reservoir strain is a useful tool for the evaluation of diastolic function and seems to be more sensitive than conventional parameters in the detection of subtle increase of LV filling pressure in patients with STEMI. It allows physicians to reclassify patients with undeterminable diastolic function according to conventional algorithm. Copyright:

INTRODUCTION

The noninvasive assessment of left ventricular (LV) diastolic function and filling pressures should be routinely performed in patients with acute coronary syndrome, particularly in patients with ST-elevation myocardial infarction (STEMI). Leading to acute change of the hemodynamic conditions, STEMI can be complicated by the onset/worsening of symptoms and/or signs of heart failure, resulting in a worse clinical course and requiring the use of diuretic therapy, to be adjusted according to the degree of LV diastolic dysfunction; thus, in these patients, the repeated measurement of LV diastolic function and filling pressures takes on a value. Several echocardiographic parameters have been introduced so far to estimate LV filling pressures;[1] nevertheless, only a few of these, often limited by confounding factors reducing their accuracy, appear to be truly useful in clinical practice. In recent years, left atrial (LA) function started to be explored through new tools, including two-dimensional (2D) speckle-tracking echocardiography (STE)-derived parameters. STE allows the evaluation of all three phases of LA function, i.e., reservoir, conduit, and booster phases,[2] which have been found repeatedly to be abnormal in the presence of LV diastolic dysfunction and elevated LV filling pressures.[34567] Moreover, increased values of brain natriuretic peptide (BNP) have been linked to higher LA pressure (LAP) and LV filling pressures.[48910] Based on these assumptions, we investigated which ones of the evaluated diastolic parameters, including traditional and LA reservoir, could allow a more accurate identification of patients affected by STEMI, with higher LV filling pressure, as estimated by BNP values.

METHODS

Study population

We prospectively enrolled 50 patients with STEMI, admitted at the Intensive Care Unit of University Hospital of Messina (Italy) from October 2016 to September 2017. Exclusion criteria were heart failure with reduced ejection fraction (EF) (LVEF <40%), significant valvular heart disease (greater than mild mitral or aortic regurgitation or mitral or aortic stenosis), prosthetic valves, cardiomyopathies, atrial fibrillation, left bundle branch block, ventricular paced rhythm, and renal failure. Moreover, patients with poor quality images, not allowing a proper tracing of atrial boundaries for 2D speckle tracking analysis, were not included.

All the patients were treated with primary angioplasty and within 24 h from the onset of symptoms, they underwent transthoracic echocardiography (TTE) for the assessment of LV systolic and diastolic functions, and blood sample for the quantification of BNP values.

All the patients gave their informed consent to participate in this study and the institutional review board approved the study protocol.

Echocardiographic study

All participants underwent 2D TTE with Vivid E95 echocardiography equipment (GE Vingmed Ultrasound AS, Horten, Norway).

Image acquisition was performed at a frame rate of 70–90 frames per second, and three cardiac cycles were stored in cine loop format for subsequent off-line calculation of LV volumes and EF by the Simpson method. Similarly, LA volume was calculated in the apical 4-and 2-chamber views by the method of disks and the averaged value was divided by body surface area to obtain LA volume index (LAVi), according to the current guidelines.[11]

Diastolic function was calculated according to the current guidelines.[1] Mitral peak early (E) and late (A) velocities, and E/A ratio, were obtained by pulsed-wave Doppler performed in the apical 4-chamber view; septal and lateral early diastolic mitral annular velocities (e') were evaluated by pulsed-wave tissue Doppler imaging, and the ratio between mitral peak early velocity and the averaged value of septal and lateral early diastolic mitral annular velocities (E/e') was calculated; the maximum tricuspid regurgitation (TR) velocity was measured in apical 4-chamber view through continuous wave Doppler. According to these parameters and LAVi, diastolic dysfunction was graded using the algorithm proposed in Grade I, II, and III;[1] in patients with E/A ≤ 0.8 + E >50 cm/s or E/A between 0.8 and 2, when only two criteria between average E/e' >14, TR velocity >2.8 m/s and LAVi >34 ml/mq were available and one was positive and one negative, diastolic dysfunction grade was considered undeterminable.

Two-dimensional Speckle tracking echocardiography analysis of left ventricular and left atrial function

Images were obtained from apical 4-, 3-, and 2-chamber views for assessing LV global longitudinal strain (GLS) and 4-chamber view for evaluating LA longitudinal strain. Using a customized commercial speckle-tracking software (EchoPAC version 7.0.0; GE Vingmed Ultrasound AS), LV GLS was calculated placing fiducial landmarks to define the base and apex of LV; the software automatically generated the region of interest and, after the processing, the bull's-eye maps, that allowed the calculation of the averaged value of GLS.

Using the same software, the LA endocardium surface was manually traced in 4-chamber and 2-chamber views by a point-and-click approach, excluding the appendage and pulmonary veins. An epicardial surface tracing was then automatically generated by the system, and the region of interest, divided into six segments, was identified. Segments with low tracking quality were corrected, whereas, when more than one segment showed a suboptimal visualization, the patient was excluded from the analysis. The software generates the longitudinal strain curves for each of these segments, together with a mean curve of all segments, whose maximal positive peak was used to calculate LA reservoir strain value. LA strain was determined as the average value from all segments of the LA in the apical 4-chamber and 2-chamber views.

Statistical analysis

A dedicated software package (IBM SPSS, v. 20) was used to perform statistical analysis. Chi-square and student's t-tests were used for comparison, between groups, of categorical and continuous variables, respectively. Receiver operating-characteristic (ROC) analysis was employed to identify the variables with better sensitivity and specificity for patients with higher levels of BNP. Net reclassification improvement, according to Pencina et al.,[12] was calculated to evaluate incremental value of variables, on top of traditional parameters, in assessing LV diastolic function. P ≤ 0.05 were considered significant.

To test interobserver and intraobserver reproducibility, the data sets of 10 randomly selected patients were reanalyzed by the same investigator (>8 weeks apart) and by a second one blinded to the former's data. Intraclass correlation coefficients (ICC) and the absolute difference divided by the mean of the pair-repeated observations (expressed as a percentage) were calculated.

RESULTS

Fifty patients (39 men; mean age 63.1 years) were included in the study. We classified our patients according to BNP values in two groups, Group A (35 patients) with increased BNP values (>100 pg/ml) and Group B (15 patients) with normal BNP values (<100 pg/ml). Main clinical, laboratory, and echocardiographic results are listed in Table 1. As shown, clinical and standard echocardiographic parameters, including mitral E/A ratio and E wave deceleration time (DTE), E/E' ratio and LV systolic indexes, were comparable between the two groups; conversely, LA reservoir strain was reduced (P = 0.001) in patients with higher BNP levels. These findings have been confirmed by ROC analysis showing that LA reservoir strain had higher area under curve values in identifying patients with elevated BNP compared with conventional parameters for the evaluation of diastolic function [Table 2 and Figure 1]. A cut-off value of LA reservoir strain of 40.5% reached sensitivity and specificity values of 93% and 86% and positive and negative predictive values (PPV and NPV) of 92% and 85%, whereas LAVi had, respectively, 90%, 33%, 41%, and 87% when using the validated cut-off of 34 ml/mq and E/E' ratio had a sensitivity of 22%, a specificity of 100%, a PPV of 100%, and a NPV of 10%.

Table 1

Main laboratory and echocardiographic features of the population

	Group A (n=35)	Group B (n=15)	P
Age (years)	62.3±2.2	62.8±3.1	0.273
Male, n (%)	27 (77)	12 (80)	0.181
Body surface area (m2)	1.81±0.37	1.77±0.35	0.765
Systolic artery pressure (mmHg)	134.2±18.6	137.4±15.6	0.563
Diastolic artery pressure (mmHg)	85.9±7.0	82.3±9.2	0.234
Heart rate (b/m’)	75±2	73±2	0.79
BNP (pg/ml)	336.5±349.8	66.7±22	0.008
Troponin I (μg/l)	52.9±38.1	36.9±35.9	0.304
EDV (ml)	123.3±10.4	115.4±9.7	0.201
ESV (ml)	61.2±4.8	55.6±3.9	0.178
Ejection fraction (%)	50.2±4.8	53.3±2.5	0.113
Mitral E/A velocities ratio	0.95±0.42	0.83±0.14	0.410
E wave deceleration time (DTE, ms)	204.1±33.6	188±29.5	0.380
Pulmonary veins S/D velocities ratio	1.29±0.31	1.48±0.02	0.313
Mitral annular averaged S’ velocity (cm/s)	8.4±1.6	9.6±1.0	0.054
E/E’ velocities ratio	9.1±2.4	7.7±1.8	0.122
Global longitudinal strain (%)	−15.9±3.3	−16.1±3.4	0.874
Left atrial volume indexed (ml/mq)	32.1±11.1	29.4±6.8	0.521
Atrial reservoir strain (%)	33.1±8	46.5±9.8	0.001

BNP=Brain natriuretic peptide, EDV=End-diastolic volume, ESV=End-systolic volume, DTE=Deceleration time E wave, S/D=Systolic/diastolic

Table 2

Main receiver-operating characteristic analysis areas under curve

	AUC
Mitral E/A velocities ratio	0.541
E/E’ velocities ratio	0.617
Left atrial volume indexed	0.570
Left atrial reservoir strain	0.880

AUC=Area under curve

Figure 1

Receiver-operating characteristic curve of left atrial reservoir calculated by two-dimensional speckle tracking longitudinal strain

Incremental value of left atrial function parameters over standard echocardiographic parameters for left ventricular diastolic function analysis

We analyzed our population according to the current recommendations about the LV diastolic function[1] and, in particular, employing the proposed algorithm based on the step-by-step evaluation of mitral E/A and E peak velocity followed by the assessment of average E/E', LAVi, and peak velocity of TR. Based on this, 70% of the study population was estimated to have normal LAP, 22.5% indeterminate LAP, and 7.5% increased LAP. However, 71.4% of patients with normal LAP, as estimated by the algorithm, and 88% of patients with indeterminate LAP showed BNP values >100 pg/ml; as it could be expected, 100% of patients with estimated elevated LAP had high values BNP.

Moreover, we tested each diastolic parameter individually and we found discordant results with respect to BNP values. Average E/E', widely used to noninvasively estimate LV filling pressures, was increased (higher than the cut-off value of 14, as previously suggested[1]) only in 8.5% of patients with augmented BNP values; however, these patients had particularly elevated BNP values (1317 ± 854 pg/ml). In addition, LAVi was normal (<34 ml/mq1) in about 68% of patients with augmented BNP values (17/25). Finally, TR can be evaluated only in 54% of patients, however, peak velocity was normal (<2.8 m/sec1) in almost all patients with high BNP values (about 92%).

Better results were obtained by using LA reservoir strain cut-off value lower than 40.5% in separating patients with increased and normal LAP. By using this parameter, we were able to reclassify 23 patients with increased BNP values, which was previously estimated to have normal (16 patients) and indeterminate LAP (seven patients) by using the recommended algorithm for diastolic function. This result was further confirmed by net reclassification improvement analysis which showed significant improvement in diagnosing patients with increased LAP using LA reservoir strain (net reclassification improvement: 0.68, SE 0.22, P = 0.002) over the recommended algorithm.

Reproducibility

Good intraobserver and interobserver agreement for LA reservoir strain was obtained. In this regard, average difference of 8.6% and ICC of 0.995 (95% IC 0.971–0.998) were observed for intraobserver reproducibility, whereas average difference of 9.4% and ICC of 0.967 (95% IC 0.932–0.994) were found for interobserver reproducibility.

DISCUSSION

In this study, we demonstrated that 2D speckle tracking analysis of LA function provides additional information in the assessment of diastolic function in patients with acute myocardial infarction. Indeed, atrial reservoir strain allowed to reclassify a relevant number of patients, erroneously considered to have normal/indeterminate LAP according to current recommendations on LV diastolic function, as having, conversely, increased LV filling pressures as estimated by elevated BNP values.

The evaluation of atrial function is particularly valuable in patients with STEMI since increased LV filling pressures had been related to mortality and morbidity independently of systolic function.[13] According to the current guidelines,[1] the noninvasive estimation of LV filling pressures should be obtained by a multi-parameter analysis of LAP, considered a mirror of LV pressure, relying on a validated algorithm based on the quantification of mitral E/A velocities ratio, E/e' velocities ratio, LAVi, and the maximum TR velocity. Based on these parameters, diastolic dysfunction is categorized in Grades I, II, and III whereas LAP is defined as normal or elevated; when the results are not concordant, diastolic dysfunction grade and LAP are considered undeterminable.

Although the parameters included in the mentioned algorithm have been validated widely and proved to be concordant with catheterization measurements of LV filling pressures,[14] each of them has significant limitations. LAVi provides only anatomical information and is rarely increased in acute diseases like myocardial infarction;[15] E/E' ratio is characterized by a too wide “gray zone,” the range values between 8 and 14, which often includes a significant percentage of patients, particularly those without severe LV systolic dysfunction; TR peak velocity is often undetectable because of lack of TR or bad quality of images. These weaknesses are smoothed by the proposed multiparametric and step-by-step evaluation of LV diastolic function, but the common finding of discordant results within the suggested parameters frequently leads to the inability to determine the degree of LV diastolic dysfunction and LAP in a large number of patients. From this obvious gap, the evaluation of atrial function through LA 2D STE strain was born. LA 2D STE strain curve is like the curves of the LA volume derivatives and includes a positive peak that corresponds to LA reservoir function during ventricular systole, a plateau during early LV diastole, and a negative peak that identifies the atrial contraction[2] [Figure 2]. LA reservoir function depends mostly on LA relaxation, due to both the LV filling pressure and the amount of fibrosis.[4] Indeed, in the first stages, when LV filling pressure is mildly elevated, LA emptying is partially hindered and the most significant percentage of emptying is obtained by atrial contraction rather than the passive early emptying, as demonstrated by the inversion of E/A ratio; when LV filling pressure further increases, atrial contraction cannot completely empty LA during diastole and an increasing amount of blood remains in the atrium, with the increase of LAP. If the elevated LV filling pressure becomes chronic, LA remodels with an increase of collagen fibers and starts to dilate to cope with the augmented quantity of blood with an impairment of LA distensibility. This progressive worsening also explains why LAVi was not an effective parameter in the evaluation of diastolic dysfunction in our cohort of patients. Indeed, myocardial infarction is an acute event that determines a sudden increase of LV and, thus LAPs, not giving the atrium the time to dilate; thus, estimated within 24 h from the onset of symptoms, LAVi was quite often normal in our population, even in patients with very high values of BNP.

Figure 2

Example of left atrium two-dimensional speckle tracking echocardiography strain in a normal subject. Please note, the three phases of atrial cycle: reservoir (yellow arrow), conduit (green arrow), and atrial contraction (blue arrow)

In our study, we chose to indirectly estimate LV filling pressure by BNP levels. Indeed, it has been widely demonstrated that increased values of BNP are closely related with high LV filling pressure[48910] even after adjusting for the confounding effects of LVEF,[8] indicating that plasma BNP levels are an independent determinant of LV filling pressures. This evidence can be easily explained considering that BNP is produced by atrial and ventricular cardiomyocytes in response to mechanical stress, such as the sudden increase of LV and LAPs, which is one of the major triggers for its synthesis. Moreover, it has been shown that different levels of BNP detect different grades of diastolic dysfunction, as estimated by E/E' ratio. Mak et al.[9] reported that patients identified with elevated LV filling pressure, defined as E/E' >15, had BNP levels of 463 ± 80 pg/mL, whereas those with normal LV pressure (E/E' <8) had BNP concentration of 97 ± 27 pg/mL and those with E/E' between 8 and 15 had a mean BNP level of 122 ± 24 pg/mL. These findings are in line with our results and, again, explain why the algorithm suggested by the current guidelines[1] was less effective in the identification of patients with increased BNP. Indeed, in our cohort, most of patients with a mild increase of BNP had a E/E' ratio between 8 and 14, in the gray zone cited above, and were considered by the algorithm as patients with normal LAP; only 8.5% of patients with augmented BNP values showed a E/E' ratio >14, but these patients had particularly elevated BNP levels and often a grade III diastolic dysfunction, with an augmented LAVi and a peak velocity of TR >2.8 m/s. On the contrary, we found that LA reservoir strain provides a more accurate estimation of LV filling pressures. Indeed, in our population, LA reservoir strain was reduced in patients with a mild increase of BNP and a E/E' ratio between 8 and 14, and it was able to reclassify a significant percentage of subjects with a subtle increase of LV filling pressures, previously considered with normal or undeterminable LAP, showing a better sensitivity than the current algorithm and improving the diagnosis in that subgroup of patients with acute and subclinical onset of diastolic dysfunction that can more significantly benefit from an early treatment. Of course, LA reservoir strain was greatly reduced in 100% of patients with particularly elevated BNP levels and a grade III diastolic dysfunction, too, but it did not add any significant information for the assessment of diastolic dysfunction. These results are concordant with previous findings in patients with heart failure with preserved EF[34567] and in patients with coronary artery disease and reduced EF.[16]

In details, Singh et al.[6] derived and tested LA strain thresholds for the assessment of diastolic dysfunction grades in unselected patients with preserved LVEF and found 3 distinct LA strain thresholds that had good to excellent diagnostic utility. On the other hand, Morris et al.[7] compared the rate of abnormal LA strain and of abnormal LAVI in patients with LV diastolic alterations and echocardiographically estimated elevated LV filling pressures, reporting that the rate of abnormal LA strain was significantly higher than an abnormal LAVi and that adding LA strain to LAVI in the current evaluation of diastolic dysfunction increased significantly the rate of detection of this condition. However, both these studies differ from ours for some important elements. First, they included unselected patients whereas our study includes only STEMI patients. Moreover, both Singh et al. and Morris et al. did not use a reference technique like invasive assessment of LV filling pressures or indirect parameters like BNP to establish the entity of diastolic dysfunction, but they based their statements on echocardiography.[67] However, as already stated, the currently used algorithm is often not able to obtain a reliable estimation of LV filling pressures; thus, we feel that it was mandatory to have a reference parameter and this is why we validated LA strain according to BNP values. Consequently, we were the first using LA strain that was able to reclassify patients with indeterminate diastolic dysfunction and high BNP values that confirmed the high LAP.

Limitations

This study has some limitations. First, it is a single-center study with a small cohort of patients that limited the strength of our findings. However, although our results need to be confirmed in larger populations, they are supported by significant pieces of evidence in literature showing that atrial reservoir strain is a reliable tool for the evaluation of LV filling pressure. Second, the estimation of LV filling pressure was performed by BNP levels and not by catheterization study, the gold standard for this measurement; however, these procedures are quite invasive and, again, several pieces of evidence strongly support our choice. The evaluation of TR peak velocity was available only in 54% of patients because, in many cases, TR was not detectable; however, it is a well-known limitation of this parameter, reported and accepted by the current guidelines for the assessment of diastolic function.[1] Finally, LA strain is calculated using a software for LV strain. However, this is a common limitation of all the studies using LA strain, and in recent years, the application of software for LV to LA has been widely validated and provided excellent results.

CONCLUSIONS

LA reservoir calculated by 2D STE strain provides reliable additional information about LV filling pressure and diastolic function in patients with STEMI. Although the conventional algorithm for the assessment of diastolic function provides a good estimation of LV pressure, LA reservoir strain was more sensitive in the detection of subtle preclinical diastolic dysfunction and was able to correctly reclassify patients previously considered to have normal or indeterminate LAP but, conversely, with increased LV filling pressures as estimated by BNP levels. Thus, LA reservoir strain should be considered a useful tool for the everyday assessment of diastolic function, particularly in patients with acute onset or worsening of diastolic dysfunction and should be used in conjunction with the conventional algorithm to reduce the frequency of the “indeterminate” classification.

Ethical clearance

Our institutional review board approved the study protocol.

Financial support and sponsorship

Nil.

Conflicts of interest

There are no conflicts of interest.