We retrospectively studied all patients with intermediate- to high-risk PE admitted to the intensive care unit of our department between November 2005 and July 2008. The diagnosis of PE was confirmed by 4- or 64-detector MDCT in all patients. The clinical severity of pulmonary embolism was stratified using the system proposed by the ESC guidelines.3 All patients had intermediate- (positive troponin I and/or RV dilatation/dysfunction on echocardiogram) or high-risk PE (shock, defined by systolic blood pressure <90 mmHg). Medical records of patients who were eligible for the study were reviewed by two of the authors, who were not involved in MDCT reporting. Clinical data was collected prospectively, including demographic, clinical, laboratory and echocardiographic data. We assessed a medium-term endpoint of all-cause mortality, and in-hospital clinical deterioration – a composite of in-hospital death, need for mechanical ventilation, inotrope usage and rethrombolysis – was also assessed. The protocol was reviewed by the institutional board. The population was divided into two groups, classified by survival status at follow-up (alive or deceased).

Decisions on treatment were the responsibility of the attending physician. The indications for thrombolytic therapy were hypotension, severe hypoxemia, evidence of RV dysfunction on transthoracic echocardiography and evidence of severe pulmonary thrombus burden. Thirty-three (84.6%) patients underwent thrombolysis with alteplase (Actilyse®, Boehringer Ingelheim Pharma, Germany) followed by a continuous intravenous infusion of heparin. No patient underwent surgical or catheter-based embolectomy. Mean follow-up was three years and survival status was available for 97.4% of patients.

Bedside standard two-dimensional echocardiography was performed by the on-call cardiologist with a portable Aloka® echocardiograph and was available in 30 patients (77%). Pulmonary artery systolic pressure (PASP) was derived from the right atrioventricular pressure gradient using peak velocity of tricuspid regurgitation with continuous wave Doppler in apical four-chamber view and the modified Bernoulli equation. Right atrial pressure was considered equal to 5 mmHg or 10 mmHg according to whether or not the inferior vena cava collapsed during inspiration, respectively.

All MDCT examinations were performed during patient stay in the emergency department by the on-call radiologist. Images were obtained on either a 4-slice or a 64-slice multidetector CT scanner during inspiratory breath holding when possible or during quiet breathing if the patient was short of breath. For patients on mechanical ventilation, ventilation was manually suspended in deep inspiration. Intravenous injection of contrast medium (80 ml of iodinated contrast material with a concentration of 400 mg/ml, at a rate of 3.5–5 ml/s) was started using a bolus tracking device, when attenuation of blood in the main pulmonary artery reached 100 Hounsfield units (HU) for the 4-slice MDCT and 180 HU for the 64-slice MDCT scanners, respectively.

Caudocranial helical acquisition was obtained with a collimation of 0.625–1.25 mm and a pitch of 0.9–1.5. Acquisition time was 0.5 s per rotation and tube current was modulated between 100 and 700 mA and 100 and 140 kV. The z-axis coverage started 2 cm below the diaphragm and extended up to the upper part of the aortic arch, enabling visualization of the heart and pulmonary arteries up to the subsegmental level. Images were immediately analyzed by the on-call radiologist, who diagnosed PE based on the presence of pulmonary arterial filling defects.

All MDCT scans were retrospectively reviewed by two independent radiologists (with 4 and 13 years of experience, respectively) who were blinded to clinical data, results of echocardiography and outcomes. The readers scored pulmonary arterial thrombus load using the system proposed by Mastora et al.7 The widths of the ventricular chambers were evaluated based on a single axial image obtained at the plane of maximal visualization of the ventricular chambers, which was typically at the level of the mitral valve. Axes were measured in diastole on a single transverse scan perpendicular to the long axis of the heart and were defined as the largest distance between the inner aspect of the interventricular septum and the ventricular free wall.14 The right ventricle was considered dilated if the right ventricular chamber was wider than the left ventricular chamber along the short axis.

Vascular measurements for the diameters of the pulmonary artery, aorta and azygos vein were also obtained used adjusted multiplanar images in the plane perpendicular to the long axis of the vessel being considered and were acquired using electronic calipers. The pulmonary artery was measured near its branching, the aorta at the middle third of its ascending portion and the azygos vein where it faced the right tracheal wall.

All variables were tested for normality using the Kolmogorov-Smirnov test. If normal, values are presented as mean values ± standard deviation. If the distribution was skewed, values are presented as medians, first and third quartiles. Correlations were analyzed with the Pearson correlation. Results are presented as coefficients of correlation (r). The area under the receiver operating characteristic curve (AUROC) was used for discrimination; this provides a measure of overall accuracy that is independent of the decision criterion. The criterion value with the highest accuracy is the value with the minimum false negative and false positive results at the same time. Statistical analyses were performed using SPSS 13.0® software (Chicago, Illinois). AUROC analysis was performed using MedCalc® software version 6.12 (MedCalc, Mariakerke, Belgium).

A total of 39 consecutive patients admitted to our intensive care unit with an MDCT-confirmed diagnosis of PE were studied. All MDCT parameters were assessable in all patients. Baseline characteristics are presented in Table 1. Mean age was 59.1±19.6 years (range 22–85 years), with a majority of females (53.8%). Body mass index was high, with a mean value of 29.2±4.2 kg/m2. No differences were found in mean blood pressure or heart rate between the groups; similarly, no differences were found between groups regarding arterial blood gases, blood glucose, troponin I, estimated glomerular filtration rate or C-reactive protein on admission. D-dimers were higher in patients who died than in surviving patients (16.3±6.8 vs. 6.8±5.2 μg/l, p=0.02).

Clinical data are presented in Table 2. Most patients presented with dyspnea (69.2%), followed by chest pain (51.3%). Almost one third suffered syncope (30.8%). The most common risk factor for PE was recent major surgery (20.5%), closely followed by neoplasia (17.9%) and a prior thromboembolic event (17.9%). A typical S1Q3T3 pattern on the surface ECG was found in 23.1% of patients and in 28.2% right ventricular strain was present. The bedside echocardiogram documented mean PASP of 56.6±17.0 mmHg, with no significant differences between groups. Regarding treatment, more than 80% of patients underwent thrombolysis (82.4 vs. 100%, p=0.307). Five percent of these were high-risk patients according to the ESC guidelines, with hemodynamic instability, while 95% were in the intermediate-risk stratum.

There were five deaths, three of them in-hospital (7.7%). None was due to bleeding complications. Median time to death was six days (range 1–40 days). The in-hospital combined endpoint of clinical deterioration occurred in 10.3% of patients. During the three-year medium-term follow-up, all-cause mortality was 12.8%.

Results of the MDCT variables are shown in Table 3. Reflecting the overall severity of the population, the arterial obstruction index was over 40% in 71.8% of patients, with a mean of 49.4±16.9%. Although this index correlated significantly with age, heart rate, troponin I and RV/LV ratio (Table 4), it was not predictive of medium-term mortality. This high arterial obstruction index was accompanied by a very high mean RV/LV ratio of 1.7±0.5. This variable differed significantly between the groups, with higher values in patients who died during follow-up (2.0±0.4 vs. 1.6±0.5, p=0.046). None of the other radiological variables correlated significantly with mortality. Patients who underwent thrombolysis had similar RV/LV ratios to patients not receiving thrombolysis (1.6±0.7 vs. 1.7±0.5, p=0.407).

On AUROC analysis, the best RV/LV ratio cut-off to predict medium-term mortality in this intermediate- to high-risk PE population was 1.81, with a c-statistic of 0.78 (95% confidence interval [CI], 0.60–0.96). Patients with an RV/LV ratio of more than 1.8 had 11-fold higher mortality than patients with RV/LV ratio <1.8 (3.8% vs. 38.8%, p<0.001). This compares with an established risk marker in PE, troponin I, which, in our population, yielded a c-statistic of 0.66 (95% CI, 0.34–0.97) for medium-term mortality.

Sensitivity and specificity (95% CI) of RV enlargement on MDCT for predicting death during follow-up were 80.0% and 76.5%, respectively. Negative and positive predictive values of RV enlargement on MDCT for medium-term all-cause mortality were 96.3% (95% CI, 81.0–99.9%) and 33.3% (95% CI, 9.9–65.1%), respectively. The RV/LV ratio had good calibration (Hosmer-Lemeshow test, p=0.594).

The authors declare that no experiments were performed on humans or animals for this study.

The authors declare that they have followed the protocols of their work center on the publication of patient data and that all the patients included in the study received sufficient information and gave their written informed consent to participate in the study.

The authors declare that no patient data appear in this article.