DISCUSSION
This study showed normal ranges of atrial S/SR in two large population-based studies. Furthermore, it demonstrated that reduced PACS and ATS were related to high NT-proBNP levels and these parameters had incremental value in the detection of elevated NT-proBNP in the general population.
Normal Ranges for atrial strain and SR
Several previous studies have reported normal values and ranges for atrial S/SR; however, only a few publications have referred to larger population-based studies with more than 1000 healthy subjects (14, 15). Apart from the inclusion of participants with diabetes in the present study, the exclusion criteria were similar in all the three studies. In accordance with these two large previous studies, our data confirmed higher ACS, ATS, and SRE in the younger population and in females. Nielsen et al. reported an increase in PACS with age, whereas our results showed the same but not a significant tendency. The absence of age dependence in SR S and SR A reflects previous findings (14).
Compared with the population-based studies of Nielsen et al. (15) and Liao et al. (14) (1641 and 2812 healthy participants, respectively), the mean atrial strain values in the present study were generally higher. They were also generally higher than those in the meta-analysis by Pathan et al. :23% vs. 21%, 27% vs. 23%, and 47.5% vs. 39% for PACS, ACS, and ATS, respectively. The meta-analysis showed high variations between the underlying studies, with a range of ATS between 28% and 60%, illustrating the challenges of defining normalcy in thin-walled atria. The high range may be due to difficulties in defining the atrial wall, which must be defined at the longest distance from the probe, leading to a low lateral resolution of the atrial walls. Small changes in the ROI position and the definition of atrial wall thickness (as necessary in EchoPac) can significantly affect the outcomes of atrial S/SR measurements. Large differences between studies and the significance of relatively small differences within each study indicate systematic errors between different laboratories.
For clinicians, the most important values for defining normalcy are the lower limits. Interestingly, compared with the study by Nielsen et al. (15), our study showed that PACS and ATS were similar, whereas the lower limits for ACS were lower, with higher variations than those for PACS. Similarly, the lower limits may be explained by the higher values in the present study, with higher data variability. Comparing SR S, SR E and SR A between the present study and the one by Liao et al. (14), we found similar mean values for SR S and SR A (2.5 vs. 2.6; 2.3 vs. 3.0 and 3.0 vs 2.9 /s, respectively), while SR E was lower (2.3 vs. 3.0, respectively). For the SR, the SD was larger (0.9/s) than that of the study by Liao et al. (0.5/s). The higher SD may reflect a higher variability between the Norwegian and Russian populations, apart from the lower population size and lower imaging or reading quality. Higher values of atrial strain may be due to a more central position of the ROI which might also explain the higher variability of the measurements.
Normal EF
Several studies have confirmed the usefulness of NT-proBNP as a marker of elevated filling pressure, indicating the presence of subclinical heart-failure in asymptomatic patients (16, 23, 24). Other studies have shown a relationship between adverse outcomes and HFpEF and heart failure with reduced EF (HFrEF) (25, 26). The cut-off value for a normal NT-proBNP in the present study has been chosen by averaging the 99 percentile of recently published age- and sex-adjusted normal-ranges of the Tromsø Study (22).
A recent population-based study of 620 individuals with normal EFs investigated LA S/SR and LASI in relation to elevated NT-proBNP levels (27) and showed weak but significant correlations between S/SR parameters and elevated NT-proBNP levels. In accordance with the study by Liu et al., all three groups with normal EF, low EF, and AF showed significantly lower atrial S/SR at elevated NT-proBNP levels, confirming the previous numbers with three times as much participants.
Based on the 2016 recommendations for hearts with preserved EF, E/e, ’ septal e, ’ lateral e´, TR velocity, and LA volume should be used to assess elevated filling pressures. In the normal EF group of the present study, E/e’ and MV DT, in addition to ATS, were independently associated with elevated NT-proBNP levels. However, these differing results cannot be regarded as a contradiction since larger studies on diastolic function have so far not been able to show sufficient test accuracies for single parameters, with the highest AUCs of around 0.7. Given the small effect sizes of each possible indicator, multiple regression analyses in different studies always favor different parameters. Liu et al. showed that LASI is a good marker of high NT-proBNP levels, which was confirmed in the present study. However, LASI was not a better indicator than the other atrial S/SR parameters.
Reduced EF
Participants with reduced EF and normal NT-proBNP are a diverse group, where either systolic or diastolic heart function is within the low-normal range, echocardiographic pathology precedes an increase in NT-proBNP, and participants may present at the stage of compensated heart failure. Thus, this study population cannot be compared with clinical studies, in which extreme values render higher test accuracies. However, because the 2016 recommendations for the assessment of diastolic dysfunction suggest a different approach for the assessment of diastolic filling pressures in hearts with reduced EFs (19), we chose to investigate the subpopulation with reduced EF separately.
Although the group with reduced EF comprised only 173 participants and 12 pathological cases, atrial S/SR revealed similar results to the normal EF group. As in the high EF group, independent conventional indicators for high NT-proBNP such as LV SV and LV ESV were different from the recommended parameters (E/A ratio, MV E, E/e´, TR velocity, and LAVI) from the 2016 guidelines (19). According to the multiple regression, PACS was an “independent” indicator for elevated NT-proBNP, supporting previous reports, that atrial S/SR are valuable indicators for high filling pressures and diastolic dysfunction (6-8). Accordingly, ROC curve analysis revealed a significant effect of adding these parameters to the combination of two conventional parameters, LVSV and LVESV. Interestingly, PACS was the only independent diastolic functional indicator of high NT-proBNP levels, indicating the usefulness of adding atrial S/SR parameters to the assessment of diastolic dysfunction.
AF
Previous studies demonstrated that LA S/SR is a useful predictor of the incidence of recurrent AF (3, 4, 28), LA reverse remodeling after radiofrequency catheter ablation (29), and future embolic events (30) in patients with known AF. However, this population-based study included only a small group of participants with AF, and unexpectedly, the number of participants with elevated NT-proBNP levels was low. In addition, this group was highly inhomogeneous, comprising participants with sinus rhythm, a history of AF, and presenting with AF during echocardiographic examination. The small group size and differences between registrations with sinus rhythm or AF were the most likely causes of the insignificant outcomes of the multiple regression analysis. However, a significant indirect comparison of groups and univariate regression analysis revealed the highest number of possible predictors of increased filling pressures, including atrial S/SR parameters, especially PACS and atrial SR, which is in accordance with previous studies confirming the importance of LA S/SR in assessing cardiac function and outcomes.
Clinical application
Invasive catheter-based studies have shown that the atrial S/SR is a valuable indicator of high filling pressures (5, 31). However, this was a population study based on healthy or asymptomatic participants to ascertain whether echocardiographic screening for elevated filling pressures or heart failure could be improved by adding LA S/SR parameters as previous studies have shown that high clinical values of atrial S/SR and LASI may predict outcomes in patients with previous AF (3, 4, 29, 30).
Assessing elevated filling pressures without an invasive approach remains challenging. Although the European Association of Cardiovascular Imaging (EACVI)- recommendations suggest a limited number of parameters, modern echocardiography provides over 40 Doppler, tissue Doppler and speckle tracking-based parameters that may be used to evaluate LV filling pressures. Pandey et al. introduced a deep-learning AI-based approach that included LA S/SR, 13 echocardiographic measures, and clinical parameters(17). All current approaches indicate that a single key parameter cannot accurately assess diastolic dysfunction, especially in asymptomatic patients. Furthermore, AI-based assessment and a clinician visually assessing the echocardiograms will still both be needed to assess many accessible parameters for optimal reading results, especially in large areas of inconclusive or contradictory measurements as the present study showed that LA strain adds incremental information; therefore, these parameters should be integrated into the assessment of LV filling pressure.
Limitations
This study has several limitations. According to previous meta-analyses and reviews (13, 32), LA strain measurements and variability differed significantly between the study groups. Furthermore, this was a single-reader study with low intra-reader variability. Multiple readers should be involved to test for a screening situation. General population-based studies have limitations in reproducing the test accuracies observed in selected patients with symptoms or a high-risk profile. However, this study was able to answer the question of the incremental value of LA S/SR compared with conventional echocardiographic parameters. Averina et al. (22) demonstrated that NT-proBNP has low sensitivity and specificity for detecting cardiac disease in the general population, reflecting the challenge of evaluating the test accuracy within only low pathological deviations. Furthermore, the chosen NT-proBNP cutoff value plays a crucial role in test accuracy. Cardiac disease can be present at high or low filling pressures; therefore, the question remains whether NT-proBNP is a sufficient indicator of filling pressures in the general population.