This retrospective single-center study included all consecutive normotensive cancer patients with PE diagnosed by multidetector computed tomography (MDCT) during emergency department (ED) stay between January 2010 and December 2011.

The inclusion criteria were age ≥18 years, presence of cancer, suspected PE at ED admission, MDCT pulmonary angiography positive for PE, and mean arterial pressure (MAP) of >65 mmHg at ED admission. Patients with unavailable medical records or incomplete follow-up were excluded.

All patients were treated according to the European Society of Cardiology guidelines.1

Demographic and clinical data, MDCT variables and blood test results at the time of ED assessment were collected by reviewing patients’ medical records. The Shock Index (SI) was calculated as the ratio of heart rate to systolic blood pressure and the sPESI was determined as previously described.3,4

The primary endpoint was all-cause 30-day mortality and the secondary endpoint was all-cause mortality at one year. Both endpoints were assessed by consulting an online nationwide death certificate database.

Demographic and clinical variables collected included the following: age, gender, symptoms at hospital admission (collapse, behavioral changes, dizziness, dyspnea, chest pain, asymmetric edema, cough and hemoptysis), vital signs (blood pressure, heart rate, temperature), Glasgow Coma Scale, peripheral oxygen saturation, presence of additional risk factors for PE (surgery in the previous four weeks, immobilization for more than three days) and comorbidities (cardiopulmonary disease, cerebrovascular disease, peripheral artery disease, atrial fibrillation, hypertension and diabetes). Laboratory variables included arterial blood gas parameters (pH, partial pressure of oxygen [PaO2] and lactate), hemoglobin, serum creatinine, D-dimers and cardiac biomarkers (troponin I and brain natriuretic peptide).

All MDCT examinations were performed during patient stay in the ED by the on-call radiologist. Images were obtained on a 64-slice MDCT scanner during inspiratory breath-hold when possible, or during quiet breathing if the patient was short of breath. Intravenous injection of contrast medium was used to assess intraluminal filling defects. Images were immediately analyzed by the on-call radiologist, who diagnosed PE based on the presence of pulmonary artery filling defects. PE was categorized as subsegmental vs. non-subsegmental according to the most proximal clot observed on the MDCT scan.

Categorical variables were described as frequency and percentage. Continuous variables were tested for normality using the Kolmogorov-Smirnov test. If normal, values were presented as mean ± standard deviation. If the distribution remained skewed after logarithmic transformation of the variables, values were presented as median and interquartile range. The distribution of variables between patients who survived and those who had died at 30-day and one-year follow-up was compared using the chi-square test and Fisher's exact test for categorical variables and the Student's t test or the Mann-Whitney U test for continuous variables, as appropriate according to normality and homoscedasticity. Variables that were statistically different between survivors and non-survivors at 30-day and one-year follow-up in univariate analysis were included in a multivariate model. For those that remained statistically different, the relationship to the primary endpoint was analyzed by a Cox proportional hazards regression model to estimate hazard ratios and 95% confidence intervals (CI). Receiver operating characteristic (ROC) curves were also used to assess the ability of these selected variables to discriminate between survivors and non-survivors at 30-day and one-year follow-up. The suggested cut-offs for measurements to predict 30-day and one-year mortality were selected so that the Youden index (sensitivity+specificity-1) had a maximum value. Finally, overall survival curves and time-to-event analysis stratified by the independent predictors of the primary and secondary endpoints were calculated through Kaplan-Meier estimates and were compared using the log-rank test. Two-tailed p values <0.05 were considered statistically significant. IBM SPSS Statistics version 23 (IBM, Vienna, Austria) was used for the statistical analysis.

Figure 1 presents the flowchart of the patient selection process. Between January 2010 and December 2011, 487 patients admitted to the ED had PE documented by MDCT scan. Of these, 97 (19.92%) were cancer patients. A total of 28 patients were excluded from the analysis (23 with MAP <65 mmHg at admission and five for whom medical records were unavailable).

Figure 1.

Flowchart of the patient selection process. MAP: mean arterial pressure; MDCT: multidetector computed tomography; PE: pulmonary embolism.

(0.16MB).

Table 1 summarizes clinical, laboratory and image data for the 69 patients enrolled in our study. Most patients were male (61%) and the median age was 73 years. The most common presenting symptom was dyspnea (58%), followed by chest pain (39%) and cough (30%). The prevalence of comorbidities, mainly cardiopulmonary disease and hypertension, was high. However, besides cancer, additional risk factors for acute PE were not commonly identified at hospital admission: three patients (4%) had undergone surgery in the previous four weeks and six (9%) had been immobilized for more than three days. Patients’ vital signs (MAP, heart rate, tympanic temperature and peripheral oxygen saturation) were within the normal range. In terms of laboratory parameters, median PaO2 was low, with a low PaO2/fraction of inspired oxygen (FiO2) ratio and high mean arterial pH. Lactate levels measured in the blood gas analysis were higher than the upper limit of normal and, as expected, D-dimers were elevated. Interestingly, troponin and brain natriuretic peptide levels in the study population were within the normal range. It should also be noted that only a few patients (9%) had subsegmental PE.

The sPESI and SI were calculated for all patients included in the study. As the sPESI incorporates positive cancer history as an adverse prognostic factor for 30-day mortality, all individuals had a sPESI score of at least one point. Almost two-thirds of patients were classified as sPESI ≥2, but only a few had a sPESI score of 4 or 5 (five and three individuals, respectively) and none had a sPESI score of 6. Finally, the SI was within the normal range, with a mean value of 0.7.

Tables 2 and 3 summarize the distribution of clinical, laboratory, and imaging features and clinical scores in survivors and non-survivors at 30-day and one-year follow-up, respectively. All-cause mortality was 28% and 55% at 30 days and one year, respectively. There were no significant differences between survivors and non-survivors in baseline characteristics, clinical presentation, comorbidities or presence of additional risk factors for PE beyond cancer. However, MAP was significantly lower in non-survivors than in survivors at 30-day and one-year follow-up: 88±3 mmHg vs. 100±3 mmHg (p<0.01) and 92±2 mmHg vs. 103±3 mmHg (p<0.01), respectively. Lactate levels were also significantly higher in patients who did not survive: at 30-day follow-up, mean log lactate was 0.34±0.06 ng/l in non-survivors vs. 0.18±0.03 ng/l in survivors (p=0.02) (corresponding to a mean lactate value of 2.2±1.2 ng/l vs. 1.5±1.1 ng/l, respectively) and at one-year follow-up, mean log lactate was 0.33±0.06 ng/l in non-survivors vs. 0.18±0.03 ng/l in survivors (p=0.03) (corresponding to a mean lactate value of 2.1±1.1 ng/l vs. 1.5±1.1 ng/l, respectively). There were no other significant differences between the two groups in laboratory parameters, including cardiac biomarker levels, at either 30 days or one year of follow-up.

In terms of clinical scores, the presence of at least one additional adverse factor beyond history of cancer in the sPESI (sPESI ≥2) was not associated with increased mortality at 30-day and one-year follow-up. However, the SI was significantly different between survivors and non-survivors at 30 days (0.7±0.03 vs. 0.9±0.04, p<0.01) and one year (0.6±0.04 vs. 0.8±0.03, p<0.01), although mean heart rate did not differ between the two groups in either follow-up period.

A ROC curve analysis was performed to select the most suitable cut-off of the SI (highest sensitivity while assuring acceptable specificity) to predict the primary endpoint. The SI had a good discriminatory ability to predict 30-day and one-year all-cause mortality, with an area under the curve of 0.76 (95% CI: 0.63-0.89; p<0.01) and 0.70 (95% CI: 0.57-0.82; p<0.01), respectively. The best cut-off of the SI to predict both 30-day (sensitivity 90%, specificity 52%) and one-year mortality (sensitivity 71%, specificity 65%) was 0.7.

Figures 2A and B illustrate the survival analysis for 30-day and one-year follow-up, stratified by sPESI <2 (blue line) and sPESI ≥2 (green line). There was no significant difference in the overall survival distribution between the groups for 30-day (17% vs. 28%, chi-square (1) 0.64, p=0.42) and one-year follow-up (43% vs. 53%, chi-square (1) 1.94, p=0.16). Figures 3A and B show survival analysis stratified by the best cut-off of the SI (0.7). Thirty-day and one-year mortality were significantly lower in patients with SI <0.7 than in those with a higher SI (7% vs. 37%, chi-square (1) 9.63, p<0.01 and 32% vs.76%, chi-square (1) 3.01, p<0.01, respectively). The hazard ratio of SI ≥0.7 was 7.20 (95% CI, 1.66-31.21, p<0.01) for 30-day all-cause mortality and 3.51 (95% CI, 1.70-7.25, p<0.01) for one-year all-cause mortality.

Figure 2.

Time-to-event curves for primary and secondary endpoints, stratified by sPESI (sPESI <2 vs. sPESI ≥2). Hazard ratios are for patients with sPESI ≥2. sPESI: simplified Pulmonary Embolism Severity Index.

(0.14MB).

Figure 3.

Time-to-event curves for primary and secondary endpoints, stratified by SI (SI <0.7 vs. SI ≥0.7). Event rates were based on Kaplan-Meier estimates in time-to-first-event analyses. Hazard ratios are for patients with SI ≥0.7.

(0.13MB).