Can left atrial diameter measured by computed tomography predict the presence and degree of left ventricular diastolic dysfunction?

Article information

Clin Exp Emerg Med. 2024;.ceem.24.194

Publication date (electronic) : 2024 May 23

doi : https://doi.org/10.15441/ceem.24.194

Gae An Kim, Ji Ung Na, Dong Hyuk Shin, Jang Hee Lee

Department of Emergency Medicine, Kangbuk Samsung Hospital, Sungkyunkwan University School of Medicine, Seoul, Korea

Correspondence to: Jang Hee Lee Department of Emergency Medicine, Samsung Kangbuk Hospital, Sungkyunkwan University School of Medicine, 29 Saemunan-ro, Jongno-gu, Seoul 03181, Korea Email: lemonpianote@gmail.com

Received 2024 January 24; Revised 2024 April 15; Accepted 2024 April 19.

Abstract

INTRODUCTION

Traditionally, heart failure (HF) refers to a clinical syndrome in which symptoms occur due to a decrease in left ventricular (LV) systolic function, represented by ejection fraction (EF) [1]. However, since the 1990s, it has been recognized that HF can occur even without a decrease in EF [2]. Currently, HF is divided into HF with reduced EF (HFrEF; EF, ≤40%), HF with midrange EF (EF, 41%-49%), and HF with preserved EF (HFpEF; EF, ≥50%) [3]. The reason why symptoms of HF appear in HFpEF is because of LV diastolic dysfunction (LVDD). In other words, to diagnose HFpEF, it is essential to identify LVDD [4,5]. Practically, the fundamental elements of HFpEF are clinical signs or symptoms of HF, evidence of preserved or normal EF, and evidence of abnormal LVDD that can be determined by Doppler echocardiography or cardiac catheterization [5]. The prevalence of HFpEF is gradually increasing, and some studies show that it accounts for more than half of all HF cases [6,7]. In addition, due to worsening of HF, the hospitalization and mortality rates in HFpEF are also known to be similar to those in HFrEF and traditional HF [8,9].

During acute exacerbations of HFpEF, patients usually visit the emergency department (ED), but it is not easy to evaluate LVDD using a bedside portable sonography in the ED setting. To evaluate LVDD, it is necessary to measure mitral inflow velocity (E/A ratio), mitral annular motion velocity (E'/A' ratio), and E/E’ using Doppler and tissue Doppler imaging [4], while systolic function can be evaluated to some extent by a relatively simple visual estimation of EF via bedside sonography [10]. Technical skills are required to accurately measure LVDD-related indicators, and, even if the values are accurately measured, background knowledge of echocardiography and cardiac pathophysiology is required for correct interpretation. Unfortunately, according to one study, despite receiving training on measuring LVDD right before the study, the accuracy of emergency physicians in distinguishing the presence and degree of LVDD using echocardiography was about 87% compared to that of cardiologists [11,12]. Additionally, many EDs may not be equipped with echocardiography equipment and often only have bedside sonography devices. For these reasons, ED physicians may not recognize LVDD, even when a patient’s symptoms are caused by quite advanced LVDD, until formal echocardiography results are reported by a cardiologist.

As LVDD progresses, it affects the left atrium (LA) and ultimately increases its size [13–15]. Since the use of computed tomography (CT) in the ED has recently increased dramatically [16], this study was conducted to determine whether the presence and degree of LVDD could be predicted using the CT-measured LA diameter (CTLAD).

METHODS

Data acquisition

To measure LA diameter on CT, the anteroposterior diameter of the LA midline was measured on the axial view of a chest CT scan (Fig. 1). A Brilliance iCT-SP 128 system (Philips Medical Systems) was used during the study period. LA diameter was measured using the length measurement tool built into the picture archiving and communication system viewer program on a 27-inch full high-definition (1920×1080) resolution monitor. The LA diameter was measured by continuously moving the longitudinal image in the mediastinal setting and selecting the image in which the anteroposterior diameter of the LA appeared the longest. One emergency physician with 14 years of clinical experience measured the LA diameter on chest CT images without knowing the corresponding echocardiography results. The presence and degree of LVDD were retrieved from formal reports of echocardiograms, which were assessed by an echocardiographer and confirmed by a cardiologist. According to the results of the cardiologist’s formal report, if LVDD was present, it was classified as LVDD of grade 1 (relaxation abnormality), grade 2 (pseudonormal), or ≥grade 3 (restrictive).

Fig. 1.

Measuring left atrium diameter on computed tomography. The anteroposterior diameter of the left atrium was assessed using the built-in measurement tool (red line) of the picture archiving and communication system.

RESULTS

From January 1, 2020, to December 31, 2021, a total of 1,118 adult patients aged ≥18 years underwent chest CT imaging and echocardiography in the ED (Fig. 2). The study was conducted on a total of 373 patients, excluding 684 patients in whom the interval between chest CT and echocardiography was ≥24 hours and 61 patients for whom there were no details on LA size or LVDD in the formal echocardiography report. There were no cases of thoracic deformity or spinal abnormalities severe enough to prevent LA size measurement on chest CT. The median age of the study subjects was 78 years (IQR, 66 to 84 years), and 188 (50.4%) were female (Table 1). The median time interval between chest CT imaging and echocardiography was 6 hours (IQR, 2 to 15 hours).

Fig. 2.

Flowchart of the study. ED, emergency department; CT, computed tomography.

Table 1.

Basic characteristics of the patients

Correlation analysis between CTLAD and echocardiography-measured LA diameter yielded a Pearson r of 0.834 (P<0.001), confirming a strong correlation between the two variables (Fig. 3). The median value of echocardiography-measured LA diameter was 4.4 cm (IQR, 4.0 to 4.9 cm), and the median value of CTLAD was 4.2 cm (IQR, 3.6 to 4.7 cm) (Table 1). Finally, the median value obtained by subtracting the CTLAD from the echocardiography-measured LA diameter was 2.1 cm (IQR, −0.5 to 5.7 cm).

Fig. 3.

A scatterplot showing the relationship between computed tomography (CT)- and echocardiography-measured left atrium diameter (LAD).

Among the 373 enrolled patients, 181 patients (48.5%) had normal diastolic function. There were 192 patients (51.5%) with LVDD, of which 122 (63.5%) had grade 1, 61 (31.8%) had grade 2, and nine (4.7%) had ≥grade 3. When comparing the CTLAD between each group using the Kruskal-Wallis H-test, there was a statistically significant difference between the four groups (χ²[3], 23.0; P<0.001) (Fig. 4). Post hoc analysis showed no statistical difference between the patient group without LVDD and grade 1 LVDD (P=0.124). Conversely, there were statistical differences between grade 1 LVDD and grade 2 LVDD (P=0.011) and between grade 2 LVDD and ≥grade 3 LVDD (P=0.014).

Fig. 4.

A box plot of computed tomography–measured LA diameter (CTLAD) according to the grade of left ventricle diastolic dysfunction (LVDD). The displayed P-values are the results of a grade comparison performed using Dunn test.

The area under the ROC curve (AUC) value of CTLAD for distinguishing LVDD of ≥grade 1 was 0.588 (95% confidence interval [CI], 0.528 to 0.648; P=0.004) (Fig. 5). The AUC value of CTLAD for distinguishing LVDD of ≥grade 2 was 0.657 (95% CI, 0.594 to 0.720; P<0.001), and the best cutoff value was ≥4.4 cm (sensitivity, 61.4%; specificity, 66.0%; PPV, 29.5%; NPV, 88.1%; OR, 3.1). The AUC value of CTLAD for distinguishing LVDD of ≥grade 3 was 0.834 (95% CI, 0.746 to 0.921; P<0.001), and the best cutoff value was ≥4.5 cm (sensitivity, 88.9%; specificity, 70.1%; PPV, 6.8%; NPV, 99.6%; OR, 18.7).

Fig. 5.

Receiver operating characteristic (ROC) curve analysis of computed tomography–measured left atrium diameter (CTLAD) for distinguishing left ventricular diastolic dysfunction (LVDD). (A) LVDD grade 1. (B) LVDD grade 2. (C) LVDD ≥grade 3. AUC, area under the ROC curve; CTLAD_adj, CTLAD adjusted for left ventricular ejection fraction; AUC_adj, AUC adjusted for left ventricular ejection fraction.

As LVDD progressed, LVEF also tended to decrease, and the Kruskal-Wallis H-test showed statistical significance (χ²[3], 22.4; P<0.001). Therefore, LVEF was added as a covariate and adjusted using a linear model, and the results are as follows (Fig. 5). The AUC value of CTLAD adjusted for LVEF for distinguishing LVDD of ≥grade 1 was 0.595 (95% CI, 0.536 to 0.654; P=0.002). The AUC value of CTLAD adjusted for LVEF for distinguishing LVDD of ≥grade 2 was 0.628 (95% CI, 0.557 to 0.699; P<0.001). Finally, the AUC value of CTLAD adjusted for LVEF for distinguishing LVDD of ≥grade 3 was 0.795 (95% CI, 0.696 to 0.895; P<0.001).

DISCUSSION

This study confirmed that CTLAD can distinguish LVDD of ≥grade 2 (CTLAD ≥4.4 cm) and LVDD of ≥grade 3 (CTLAD ≥4.5 cm). In the stage of grade 1 LVDD, although LV relaxation is impaired, it is not accompanied by an increase in LV end-diastolic pressure (LVEDP), therefore the LA pressure and size do not increase and pulmonary edema rarely occurs. Therefore, grade 1 LVDD, which is relatively common in asymptomatic elderly people, has little clinical significance. However, since LVDD of ≥grade 2 is accompanied by an increase in LVEDP, both LA pressure and size increase, and pulmonary edema may also occur [18]. Therefore, the finding of this study that LVDD of ≥grade 2 can be distinguished by CTLAD may be a clinically meaningful finding.

Although patients with diastolic HF may have symptoms like shortness of breath or chest tightness during exercise, these are not disease-specific symptoms that only appear in diastolic HF. In addition, the current diagnostic guidelines for diastolic HF require Doppler echo to be performed to measure and interpret mitral inflow velocity and mitral annular motion velocity [4,5]. Therefore, even if a patient’s cause of symptoms is quite advanced diastolic HF, LVDD may not be diagnosed without objective tests such as echocardiography. The results of this study showing that LVDD of ≥grade 2 might be distinguished using CTLAD should be of great help to clinicians who are unfamiliar with echocardiography or working in situations where echocardiography is not feasible.

Increased LA size can be evaluated by echocardiography-measured LA diameter or volume index, and, in this way, LVDD can be indirectly diagnosed and classified [13–15]. However, echocardiography is often unavailable (especially during the nighttime and holidays), and many EDs are not even equipped with echocardiography technology, therefore this study focused on CTLAD. CTLAD was 2.1 cm shorter than the echocardiography-measured LA diameter in this study. This is thought to be because echocardiography uses electrocardiograms to measure the largest LA diameter at the time of LV end-systole, but CT does not consider this timing. Although the CTLAD was 2.1 cm shorter than the echocardiography-measured LA diameter, the measurement values of these two methods showed a strong correlation, and it was confirmed that LVDD of ≥grade 2 might be recognizable using CTLAD. As such, CTLAD of ≥4.4 cm can serve as a rough reference value for distinguishing LVDD of ≥grade 2, and CTLAD of ≥4.5 cm can reliably distinguish LVDD of ≥grade 3.

One noteworthy finding among the results of this study is the high NPV of CTLAD. If the CTLAD is <4.4 cm, the likelihood of LVDD of ≥grade 2 is significantly lower when considering the NPV of 88.1%, and, if the CTLAD is <4.5 cm, the NPV is 99.6%, meaning that LVDD of ≥grade 3 can be virtually excluded. We suggest that the LA diameter must be checked after a chest CT scan if a patient has HF-relevant symptoms.

There is one previous study by Lick et al. [19] suggesting CTLAD can predict LVDD in patients who underwent CT and echocardiography within 48 hours, and the median time interval between CT and echocardiography in their study was 15.5 hours. The study concluded that CTLAD of ≥4.0 cm can discriminate LVDD of ≥grade 2 with 68% sensitivity and 74% specificity. In our study, the best CTLAD value for distinguishing ≥grade 2 LVDD was 4.4 cm, whereas, in their study, it was 4.0 cm, which is 0.4 cm shorter. We do not have answers to explain this difference; however, we do believe that limiting the time interval between CT and echocardiography to <24 hours (median, 6 hours) is one of the biggest strengths of this study. As the interval between CT and echocardiography increases, the LVEDP, LA pressure, and LA size at each time point may vary due to the influence of intravenous fluids or the use of HF treatment medications. Therefore, the results of this study, where the time interval between the two tests was only about 6 hours, are considered to be reliable.

This study has several limitations. First, this was a retrospective cross-sectional study including only patients who visited the ED and underwent both CT and echocardiography within 24 hours. Performing two tests intensively at the same time suggests that the patient is likely to be more seriously ill, which may have resulted in selection bias in our investigations. Second, CTLAD is not as accurate as the LA volume index in reflecting the true LA size. The actual appearance of the LA is a complex shape with a curved outer margin, and the volume index is known to be more accurate for measuring the true LA size [20,21]. To measure the LA volume index, the area–length method or Simpson method should be used in the apical four- or two-chamber view at LV end-systole, when the LA size is largest. However, it is difficult to measure the LA volume index in the ED, where visual estimation is mainly performed using portable sonography. Although CTLAD does not reflect the true LA size like the LA volume index does, it is an index that indirectly suggests LA enlargement, and, since CT is available in most EDs, it was used in this study as an alternative modality to measure LA enlargement. Third, there is a possibility that fluids or specific medications administered to patients in the ED may have affected their actual degree of diastolic HF and LA size. A drawback of this retrospective study is that the analysis did not reflect whether and how much fluid was administered or what medications were administered to the patient. However, the short time interval between CT and echocardiography in this study should have minimized the potential effects of fluid or medications.

In conclusion, LVDD of ≥grade 2 might be distinguished using CTLAD. CTLAD of ≥4.4 cm can be used as a rough reference value for distinguishing LVDD of ≥grade 2, and CTLAD of ≥4.5 cm can reliably distinguish LVDD of ≥grade 3. This might be a helpful finding in the treatment of patients complaining of HF-related symptoms in situations where echocardiography is not feasible.

Article information Continued

Notes