- Top of page
- Materials and methods
- Disclosure of Conflict of Interests
Background: Early identification by computed tomography pulmonary angiography (CTPA) of patients with acute pulmonary embolism (PE) who have signs associated with a high embolic burden would be highly desirable. Objectives: To investigate whether an increased obstruction of the pulmonary vasculature is associated with reduced left atrial (LA) and increased right atrial (RA) areas. Methods: We retrospectively analyzed a consecutive series of CTPA studies of 137 patients with acute PE and 38 controls without PE between October 2004 and March 2006. Left and right atrial areas and longitudinal and short axis diameters were measured and correlated with the pulmonary arterial obstruction index (PAOI) divided into tertiles (obstruction of < 12.5%, 12.5%–42.5% and ≥ 42.5%). Results: There was a significant negative age- and gender-adjusted correlation between the PAOI and LA measurements, particularly the LA area (r = −0.259) and the LA short axis diameter (r = −0.331). All RA measurements had positive correlations (RA area, r = 0.279; RA short axis diameter, r = 0.313). The LA/RA area ratio correlated negatively with the PAOI (r = −0.447). All above-mentioned correlations had P < 0.002. All the LA measurements were the largest in the controls and gradually decreased with higher PAOIs. A receiver operating characteristic curve analysis demonstrated that the RV/LV diameter, LA/RA area and LA/RA short axis diameter ratios had comparable discriminative ability for higher PAOI tertiles. Conclusions: The higher the clot load in the pulmonary arteries, the smaller the LA area and the larger the RA area. Atrial area measurements by CTPA may serve as a real-time parameter in assessing the severity of PE upon diagnosis.
- Top of page
- Materials and methods
- Disclosure of Conflict of Interests
There were 47 males and 90 females with PE, who had a mean age of 67 ± 19 years (range 22–96). The control group included 12 males and 26 females, who had a mean age of 68 ± 18 years (range 21–88). The age difference of the two groups was not significant (P = 0.610). The baseline characteristics of the study cohort are presented in Table 1. There were more patients with asthma or COPD as well as hypertension and dyslipidemia in the control group than among the patients with PE.
Table 1. Characteristics of patients with PE and controls
| ||PE (n = 137)||Controls (n = 38)||P value|
|Current smoker||4||3||1||3||> 0.999|
|Congestive heart failure||5||4||2||5||> 0.999|
|Ischemic heart disease||11||8||7||18||0.074|
We found substantial agreement between the two independent observers for the LA long and short axis diameter and the RA long diameter measurements (κ = 0.47–0.52), and an even higher agreement (κ = 0.73–0.84) for the LA and RA areas, as well as for the RA short axis diameter.
Significant gender and age associations were noted for all atrial measurements, but not for the ratios of left to right atria (data not shown). Therefore all subsequent atrial analyses were age and gender adjusted. Table 2 presents the correlation between the PAOI and the atrial and ventricular measurements. There was a significant negative correlation between the PAOI and all LA measurements, particularly those of the LA area and LA short axis diameter, while all RA measurements were positively correlated with the PAOIs. Consequently, LA/RA area ratios yielded the greatest negative correlation with the PAOI among all the atrial measurements. We further divided our sample into tertiles based on PAOIs: PAOI < 5 (obstruction of < 12.5%), 5 ≤ PAOI < 17 (obstruction of 12.5–42.5%), and PAOI ≥ 17 (obstruction of > 42.5%), and compared the various measurements between the three tertiles and the controls. Age- and gender-adjusted estimated marginal means for all measurements in each tertile and controls are displayed in Table 3. All the LA measurements were smallest for the group with the highest PAOIs, and the values gradually increased as the PAOIs became smaller. All the LA measurements were highest in the control group. The RV to LV diameter ratio showed no significant difference between the control group and the two lower PAOI tertiles, while the value for that ratio for the group of patients with large PE was significantly higher. Analyzing the less severely ill patients (e.g. those in the two lower PAOI tertiles) and the controls, while excluding the patients in the highest PAOI tertile, revealed that the left atrial area was the only parameter to demonstrate a significant difference between those groups (P = 0.021), and that there was no significant difference in the RV to LV diameter ratio between those groups.
Table 2. Partial correlation* between PAOI and atrial and ventricular measurements
| ||LA area||LA long. diam.||LA short axis diam.|
| P||0.003||0.096||< 0.001|
| ||RA area||RA long. diam.||RA short axis diam.|
| P||0.001||0.017||< 0.001|
| ||LA/RA area ratio||LA/RA long. diam. ratio||LA/RA short axis diam. ratio|
| P||< 0.001||0.004||< 0.001|
| ||RV diam.||LV diam.||RV/LV ratio|
| P||< 0.001||< 0.001||< 0.001|
Table 3. Estimated marginal means (95% CI) of the different atrial and ventricular measurements in PAOI tertiles*
|Atrial measurements||Controls||PAOI < 5 (obstruction < 12.5%)||5 ≤ PAOI < 17 (12.5% ≤ obstruction < 42.5%)||PAOI ≥ 17 (obstruction ≥ 42.5%)||P value|
|Mean||95% CI||Mean||95% CI||Mean||95% CI||Mean||95% CI|
|LA area (cm2)||24.1||22.2–25.9||21.3||19.7–23.0||20.1||18.4–21.9||17.7||16.0–19.5||< 0.001|
|LA long. diam. (cm)||5.9||5.7–6.2||5.6||5.3–5.8||5.5||5.2–5.7||5.3||5.0–5.5||0.003|
|LA short axis diam. (cm)||4.5||4.2–4.8||4.3||4.1–4.6||4.1||3.8–4.3||3.7||3.4–3.9||< 0.001|
|RA area (cm2)||22.6||20.5–24.8||20.7||18.8–22.5||20.0||18.0–22.0||24.4||22.4–26.3||0.007|
|RA long. diam. (cm)||5.0||4.7–5.3||4.7||4.4–4.9||4.6||4.3–4.8||4.9||4.7–5.2||0.060|
|RA short axis diam. (cm)||5.3||5.0–5.6||5.2||4.9–5.5||4.9||4.7–5.3||5.8||5.5–6.1||0.001|
|LA/RA area ratio||1.10||1.01–1.18||1.06||1.00–1.14||1.08||1.00–1.16||0.76||0.68–0.84||< 0.001|
|LA/RA long. diam. ratio||1.23||1.16–1.31||1.21||1.15–1.28||1.23||1.16–1.30||1.09||1.02–1.15||0.008|
|LA/RA short axis diam. ratio||0.89||0.82–0.95||0.86||0.80–0.91||0.85||0.79–0.91||0.65||0.59–0.70||< 0.001|
|RV diam. (cm)||4.0||3.7–4.2||3.7||3.5–3.9||3.8||3.6–4.0||4.4||4.2–4.6||< 0.001|
|LV diam. (cm)||4.1||3.9–4.4||4.2||4.0–4.4||4.1||3.9–4.3||3.6||3.3–3.8||< 0.001|
|RV/LV diam. ratio||0.98||0.90–1.06||0.89||0.82–0.97||0.94||0.86–1.01||1.31||1.23–1.38||< 0.001|
Table 4 presents the results of the ROC curve analysis on the ability of different cut-offs of the atrial measurements to identify individuals with a large clot burden, represented here as the upper PAOI tertile. The right to left ventricle diameter ratio, left to right atria area ratio and left to right atria short axis diameter ratio had the highest area under the curves, without significant difference between them using DeLong and DeLong’s analysis (P > 0.6 for each), and all three measurements performed significantly better than all other atrial measurements (all P < 0.03).
Table 4. ROC curve analysis of the different atria and ventricular measurements for the higher PAOI tertile (obstruction of ≥ 42.5%)
| ||AUC||95% CI||P value||90% Sensitivity|
|LA area||0.659||0.568–0.751||0.002||< 22.0||36.3|
|LA long. diam.||0.609||0.510–0.708||0.037||< 6.3||14.3|
|LA short axis diam.||0.687||0.599–0.775||< 0.001||< 4.5||37.4|
|RA area||0.669||0.577–0.761||0.001||> 16.0||34.1|
|RA long. diam.||0.587||0.486–0.688||0.098||> 3.7||13.2|
|RA short axis diam.||0.694||0.606–0.781||< 0.001||> 4.5||33.0|
|LA/RA area ratio||0.821||0.747–0.894||< 0.001||< 1.03||50.5|
|LA/RA long. diam. ratio||0.672||0.575–0.768||0.001||< 1.41||17.6|
|LA/RA short axis diam. ratio||0.805||0.730–0.879||< 0.001||< 0.82||48.4|
|RV diam.||0.761||0.677–0.845||< 0.001||> 3.5||35.2|
|LV diam.||0.718||0.626–0.810||< 0.001||< 4.6||29.7|
|RV/LV diam. ratio||0.828||0.746–0.910||< 0.001||> 0.84||34.1|
In addition to the area under the curve, we also report the cut-off that achieves 90% sensitivity and the corresponding specificity as displayed in Table 4. The highest discrimination at this point, with 90% sensitivity, was achieved by the left to right atrial area ratio and the left to right atrial short axis diameter ratio, while RV/LV ratio was less discriminative.
Twenty-three of all our 137 PE patients (17%) and four of the 38 controls (11%) died by the end of the 30 days of follow-up. Of the 23 PE patients who died, the cause of death was probably related to the PE in only eight (35%) of them, while the cause of death was most likely related to the background comorbidities or current acute illness in the other 15. Among those eight PE-related deaths, four were in the higher PAOI tertile, three in the middle tertile and only one in the lower PAOI tertile (P = 0.12). In addition, only 10 of the 137 PE patients (7%) and two of the 38 controls (5%) were admitted to an ICU during the 30 days of follow-up. Three of those 10 PE patients were in the higher PAOI tertile, five were in the middle tertile and two were in the lower PAOI tertile (P = 0.50). The small numbers of PE-related deaths and ICU admissions precluded our ability to analyze the parameters of PE mortality and ICU admissions in relation to the atrial and ventricular measurements.
- Top of page
- Materials and methods
- Disclosure of Conflict of Interests
The main finding of our study is that a higher clot load in the pulmonary arteries is associated with a smaller LA size and a larger RA size, as measured on CTPA studies. Moreover, the LA/RA area and short axis ratios were as capable of identifying severe obstruction as was the RV/LV diameter ratio, which is considered the most accepted parameter for severity assessment in patients with acute PE [4,5,7,8]. When comparing atrial and ventricular measurements with regard to patients with low clot burden (i.e. between the two lower PAOI tertiles and the controls), the left atrial area was the only parameter to demonstrate a significant difference, while RV/LV ratio did not. In addition, at the preselected clinically important point of 90% sensitivity, the atrial ratios demonstrated even better specificities compared with the RV/LV diameter ratio. The obtained interobserver agreement of our planimetric measurements of the atria yielded a higher value (κ = 0.73–0.84) than the unidimensional atrial diameters (κ = 0.47–0.52), which were similar to the interobserver agreement for ventricular diameters previously reported in the literature . The significance of our study is its proposal of an additional parameter that reflects modifications in cardiac morphology in response to pulmonary arterial obstruction in patients with acute PE. Atrial assessment is rapid and easy and may therefore contribute to the establishment of CTPA as a useful tool for both the diagnosis of PE and the identification of those patients with cardiac morphological changes that are associated with a high embolic load who might be at risk of sudden circulatory collapse and, consequently, may be considered for closer monitoring, with possible thrombolytic therapy .
Pathophysiological studies have suggested that there is an increase in pulmonary vascular resistance due to the anatomical obstruction caused by the emboli, release of vasoconstricting agents and reflex hypoxemia during a major PE event . This sudden increase in RV afterload results in elevated right ventricular wall tension, dilatation of the RV, inter-ventricular septum shift towards the LV and decreased LV diastolic volume. Right ventricular contractile dysfunction and acute tricuspid regurgitation cause decreased output from the RV, contributing to underfilling of the LV, with decreased cardiac output, decreased systemic blood pressure and decreased coronary perfusion, which may eventually cause circulatory collapse [19,20]. At present, the evaluation of patients with suspected PE consists of echocardiography and CTPA . Echocardiography is both a rapid and accurate risk-assessment tool that is useful in identifying the PE patients who have a poor prognosis [3,21,22]. Its ability to visualize pulmonary arterial clots that are not very large or centrally located is, however, quite limited . Its diagnostic ability is also operator and patient dependent . Previous echocardiographic studies in PE patients mostly concentrated on the right and left ventricular measurements [21,23,24]. We are aware of only one echocardiography work that demonstrated that an increased ratio of right to left atrial end-systolic area correlated with obstruction of more than 30% of the pulmonary arterial tree . This analysis, however, was conducted on 63 elderly patients and used ventilation/perfusion pulmonary scintigraphy to estimate the PE size: only 11 patients (17%) were found to have obstruction of > 30%. Nevertheless, the findings of that earlier study support those of our current work, in which we used CTPA in a larger and unselected group of PE patients and controls.
A number of CTPA studies have reported the presence of RV dilatation and an inter-ventricular septal shift towards the LV in association with severe PE [4–8]. The reduction of LA size, as seen on CPTA, in conjunction with a massive PE that is possibly due to reduced venous return as a result of the high clot load was, however, reported only in case reports [14,26]. Our current study is the first to show an association between the embolic extent, as expressed by PAOI, and atrial size. Because all the LA measurements were smallest for the group of patients with the highest pulmonary arterial obstruction, and gradually increased with smaller obstructions (the largest LA values were in the control group), increased mechanical obstruction can be considered to result in the reduction of pulmonary venous return to the LA. The respective increase in RA dimensions supports the presence of an additional mechanism (i.e. the so-called interdependence of the right and left cardiac chambers). Accordingly, under pericardial constriction, RV dysfunction causes not only impaired LV diastolic filling, but also enlargement of the RA due to increased RV-RA filling pressures, which leads to the compression and reduction in size of the adjacent LA . Because both atrial walls are thin, while the LV wall is significantly thicker than the RV wall, the effect of this mechanism might be more pronounced between the atria.
Several authors found PAOI to predict mortality in acute PE [5,8,10,13], while others found no similar association between the two [4,6,9,15,27]. These contradictory findings might support the hypothesis that PE outcome is related to both embolus size and the underlying cardiopulmonary reserve [18,20]. Large-scale outcome studies are probably still required in order to clarify this issue.
The performance of cardiac-gated CT angiography for quantitative chamber assessment was recently compared with echocardiography, using oblique multi-planar reformations similar to those obtained by transthoracic echocardiography, and there was a good agreement score . LA size assessment was based on its posterior-anterior diameter as seen on an oblique view alone, which we felt to be an inaccurate measure of its entire volume. We preferred to obtain a planimetric assessment, which we later used for RA measurement as well, because there is no previously published quantitative assessment of the RA on CT. Interestingly, it was recently shown that the LA diameter increases significantly with age, by up to 61% between the ages of 40 and 80 years . Our results support this finding and also demonstrate gender differences in atrial parameters. Consequently, both atria measurements were adjusted to patients’ age and gender during the compilation of our results. Finally, RA dilatation in association with RV enlargement and hypokinesis in patients with PE had been previously reported on echocardiography , but not on CTPA.
One of our study’s limitations is its use of non-gated CT angiography for LA and RA measurements. Lu et al.  recently compared LV and RV measurements obtained from gated and non-gated CTPA in a cohort of 30 patients with acute PE. They found a high correspondence rate between measurements performed according to both protocols, and concluded that quantitative RV evaluation using gated CT in acute PE patients is not justified in the clinical setting. Our purpose was to examine data obtained from routine clinical settings. This was also the reason for obtaining atrial measurements directly from axial slices that are regularly used for PE assessment. Planimetric assessments rather than volumetric measurements comprise another drawback of this preliminary study: automated dedicated software for volumetric assessment of all cardiac chambers should be used in future studies after they are proven to be adequately accurate. Another limitation is the inability to provide sufficient data on the relationship between our findings and the patients’ outcomes, because most causes of death were not associated with the presence of PE alone. Finally, the lack of data on cardiac comorbidities, which may influence atrial dimensions as well as comparison in real time with echocardiography, which could provide functional correlation, are additional drawbacks that are also related to the retrospective nature of the study.
In conclusion, assessment of the association between the degree of embolic obstruction and the LA and RA areas revealed that a higher pulmonary arterial clot load is associated with a smaller LA area and a larger RA area. Our findings suggest that atrial dimensions as seen on CTPA may serve as an additional early parameter that reflects modifications in cardiac morphology in response to the extent of pulmonary arterial obstruction, and thus may contribute to a more comprehensive risk assessment in patients with acute PE.