INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Pulmonary embolism (PE) remains a challenging diagnostic problem, and objective tests are needed to confirm or rule out the diagnosis of this condition, since the accuracy of its clinical recognition is lower than 50% (1). Pulmonary angiography is the gold standard for the diagnosis of PE. Unfortunately, it is expensive and invasive (2).

Because of this, noninvasive investigations are done when PE is suspected (3). The ventilation/perfusion (A/) lung scan is often done in suspected PE, but is diagnostic in only 30 to 50% of patients (4). Some studies (5, 6) have done lung scanning in association with compression venous ultrasonography (US), the latter being a sensitive and highly specific investigation for the diagnosis of proximal deep vein thrombosis (DVT) (7). The diagnosis of proximal DVT in a patient with suspected PE warrants anticoagulant therapy, rendering other investigations unnecessary. Other authors (8) put forward a strategy for diagnosing PE that combines clinical probability, lung scanning, venous US, and D-dimer (DD) measurement. The measurement of DD, a degradation product of crosslinked fibrin, proved to be an excellent tool for ruling out PE (9, 10). These noninvasive diagnostic strategies seemed to be safe, but yielded a final diagnosis in only about 60% of patients. Some years ago, a new radiologic technology, spiral computed tomography (CT), became available. It is a minimally invasive technique. Several studies (11) have shown a high specificity for spiral CT (92 to 97%) in diagnosing PE. The sensitivity of spiral CT ranged from 63 to 100% because of its poor performance in the diagnosis of subsegmental emboli, for which reason a normal angioscan cannot rule out PE. Only one study associated spiral CT with other noninvasive examination modalities for patients with suspected PE, but this had a selection bias, since spiral CT was performed only in patients with indeterminate lung scan findings and normal US.

The aim of our study was to evaluate, in consecutive patients with suspected PE, a noninvasive diagnostic strategy combining spiral CT, a A/ lung scan, plasma DD measurement, and in some cases (those with normal spiral CT findings and a nondiagnostic lung scan and increased DD), venous US.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Patients

Two hundred forty-seven consecutive patients with clinically suspected PE over a 2-yr period at the Hôtel-Dieu Hospital were included in the study.

Procedures

Spiral CT, A/ lung scan, and plasma DD measurement were done systematically within 24 h after admission on patients with clinically suspected PE. Lower-limb US was performed only on patients with a normal CT in association with a nonconclusive lung scan and increased DD concentration.

Pulmonary perfusion scans were performed after an intravenous injection of 100 MBq (3 to 4 mCi) of ^99mTc-labeled human albumin macroaggregates. Six views (anterior, posterior, right and left lateral, and right and left posterior oblique views) were obtained with a large-field-of-view rotating gamma camera. Lung scans were classified as normal, or of very low, low, intermediate, or high probability for PE according to the Prospective Investigation of Pulmonary Emolism Diagnosis interpretation criteria (4). Lung scans of very low, low, and intermediate probability were considered nondiagnostic.

In patients for whom ventilation scanning could not be performed, lung scans were considered as normal when no perfusion defect was disclosed, and as nondiagnostic when one or more perfusion defects were observed.

Spiral CT was performed with an Elscint CT Twin scanner (Elscint, Haifa, Israel). From 140 to 160 ml of a nonionic contrast medium with 25% iodine content (Xenetix 250; Guerbet, Aulnay-Sous-Bois, France), corresponding to an iodine dose of 35 to 40 g, were injected at a rate of 4 ml/s. Twenty seconds after the beginning of the injection, a spiral image acquisition was obtained from the aortic arch to the pulmonary base, with a pitch of 1.5, a slice thickness of 2.7 mm, a 1.5-mm overlap between slices, a scan diameter of 430 mm, and a matrix of 512 × 512 pixels. The scan time varied from 22 to 27 s, which corresponded to a spiral length of 16 to 20 cm. Breathholding was not mandatory during the scan.

Additional 5-mm contiguous conventional CT scans were obtained in order to study the rest of the parenchyma and mediastinum.

Detection of central pulmonary embolism included analysis of the main, lobar, segmental, and subsegmental pulmonary arteries. Acute PE was diagnosed if there was at least one pulmonary artery with one of the vascular signs of PE described by Remy-Jardin and colleagues (14), and consisting of a partial or a complete filling defect, a "railway track," or a mural defect.

The plasma DD level was measured within the first day after admission with a new DD assay (VIDAS D-dimer; bioMérieux, Marcy l'Etoile, France) (16). This is a new, quantitative enzyme-linked immunosorbent assay (ELISA) method that is automated on the VIDAS immunoanalyzer, and which combines sandwich immunoenzymatic methods in two steps, with a final fluorescence detection. The ready-to-use reagent for the assay comprises a pipetting device coated with a mouse monoclonal anti-DD antibody (solid phase) and a strip containing the other reagents, consisting of the conjugate (alkaline phosphatase-labeled mouse monoclonal anti-DD antibody), washing solutions, sample diluent, and substrate. After 200 µl of plasma sample is pipetted onto the reagent strip, the analyzer manages all further steps and the result is given after 35 min. All analyses were performed on single specimens and according to the manufacturer's instructions. All values are expressed in ng/ml of fibrinogen equivalent units (FEU).

Lower-limb B-mode venous compression US consisted of a real-time B-mode examination of the iliac, common and superficial femoral, popliteal, and sural veins. The criterion for diagnosing DVT was noncompressibility of the vein and/or the presence of a thrombus.

The physicians who performed the spiral CT, A/ lung scan, and US studies were not blinded to patients' clinical histories, but they were told only that the patients had suspected acute PE. All examinations performed were reviewed by physicians with considerable experience in their respective fields. There was one expert for each of the three imaging techniques (CT scan, A/ lung scan, and US). The technicians who performed DD assays were completely blinded to patients' clinical histories.

Diagnosis or Exclusion of PE

In patients in whom spiral CT, lung scan, and DD measurement could be done (Figure 1), the diagnosis of PE was confirmed if: ( 1) spiral CT showed a picture of thrombus as previously defined; and (2) US showed a thrombus when spiral CT was normal.

View larger version (21K): [in this window] [in a new window]   Figure 1.   Noninvasive strategy including spiral CT, A/ lung scan, and DD measurement for diagnosis of PE. CT = computed tomography; A/ = ventilation/perfusion; Q = perfusion; DD = D-dimer; HP = high probability; DVT = deep venous thrombosis.

When spiral CT showed no thrombus, the diagnosis of PE was ruled out if any of the following occurred: (1) the lung scan was normal; (2) the DD level was < 500 µg/L; (3) the lung scan was nondiagnostic and a prior lung scan was similar; (4) the lung scan was nondiagnostic and the patient had no prior lung scan, and there was an obvious differential diagnosis on spiral CT; or (5) the lung scan was nondiagnostic and the patient had no prior lung scan, the DD level exceeded 500 ng/ml, there was no obvious differential diagnosis on spiral CT, and venous US was normal.

Disagreement between the results of lung scan and spiral CT (a high-probability lung scan and normal spiral CT, or a normal lung scan and a picture of thrombus on spiral CT) made it necessary to perform pulmonary angiography.

Spiral CT could not be performed in some cases: absolute contraindications to the administration of contrast medium were prior hypersensitivity to contrast medium and renal failure defined by a serum creatinine level exceeding 200 µmol/L. Relative contraindications were old age (> 80 yr) and diabetes mellitus. Thus, only A/ lung scan and DD measurement were obtained for patients in whom spiral CT was contraindicated.

The diagnosis of PE was confirmed if: ( 1) a high-probability lung scan was obtained; or (2) US showed a thrombus.

The diagnosis of PE was ruled out if: (1) the lung scan was normal; or (2) the DD level was < 500 µg/L; or (3) the lung scan was nondiagnostic and a prior lung scan was similar.

When patients were submitted to mechanical ventilation, only spiral CT and measurement of DD were obtained. The diagnosis of PE was confirmed if: (1) spiral CT showed an image of thrombus as defined earlier; or (2) US showed a thrombus when spiral CT was normal.

When spiral CT showed no thrombus, the diagnosis of PE was ruled out if: (1) the DD level was < 500 µg/L; or (2) there was an obvious differential diagnosis on spiral CT.

Patients were followed for a 3-mo period.

The study design was in accordance with the revised Helsinki Declaration of 1983. Informed consent was required of all patients.

3-mo Follow-up

The usefulness of the foregoing strategy for ruling out PE was assessed in terms of occurrence of PE in patients in whom the diagnosis of PE had been eliminated and who therefore did not receive anticoagulant therapy. During the 3-mo follow-up, recorded clinical events were venous thromboembolic events (DVT or PE). Diagnosis of venous thromboembolic events was established with the following criteria: noncompressibility of the vein and/or presence of thrombus in the case of DVT; a high-probability A/ lung scan or spiral CT showing a picture of thrombus as previously defined in the case of PE. If the follow-up examination could not be done in our institution, we telephoned patients or their family physicians and inquired about the occurrence of any clinical event that could be related to PE or DVT. If the patient had died, the cause of death was determined through discussion with the physician in charge at the time of death.

Statistical Analysis

Statistical analysis was done with Statview SE software (Abacus Concepts, Berkeley, CA). All results are expressed as mean ± SEM.

Rates of demographic findings, history of cancer, hereditary thrombophilia, risk factors, thromboembolic events, and mortality on follow-up in patients with and without PE were compared through the chi-square test or Fisher's exact test. A value of p < 0.05 was considered statistically significant.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Study Population

Of the 247 patients enrolled in the study, data from 19 were excluded from analysis for the following reasons: One patient died before investigation (the diagnosis of massive PE was made at autopsy). In one patient the diagnosis of PE had already been made by pulmonary angiography before admission. One patient presented several months after the onset of symptoms, and the diagnosis of PE could therefore not be excluded even though our protocol showed no PE. The patient was treated for 3 mo with anticoagulant therapy. In the case of 16 patients the study protocol was violated, with the result that it was not possible to reach any conclusion (Table 1). All 16 of these last-named patients were considered as being without PE, and they were not given anticoagulant therapy. None of these patients had a thromboembolic event during the 3-mo follow-up.

We evaluated 228 patients. The prevalence of PE in this population was 42% (96 of 228). Table 2 summarizes the demographic findings and clinical history in this group.

Diagnosis of PE

Spiral CT was done on 201 patients (88%), and was not possible in 27 patients (12%) because there was an absolute or relative contraindication to administration of the contrast medium (renal failure in 12 patients, hypersensitivity to contrast medium in eight patients, old age in three patients, diabetes mellitus in two patients, pregnancy in one patient, and a technical reason in one patient). Opacification of the pulmonary arteries was satisfactory in 190 patients (95%). One complication (renal failure that required hemodialysis for 3 wk) was associated with spiral CT. A/ lung scan was performed in 178 patients (78%). Twenty-nine patients (13%) had only a perfusion lung scan because they presented with tachypnea. The other patients had no lung scan because they were receiving mechanical ventilatory support in 16 cases and for technical reasons in five cases. DD measurement was done in 227 patients (99.5%). US was necessary in 56 patients (23.6%).

Table 3 summarizes the results of the diagnostic strategy.

Diagnosis of PE

Spiral CT was positive in 70 patients (73% of patients with PE). Of these patients, 34 had a high-probability lung scan, 30 had a nondiagnostic lung scan, and six did not have a lung scan. Of the 70 patients with a positive CT, 69 had an increased DD level. One patient had a DD level < 500 ng/ml, and spiral CT showed many segmental and subsegmental thrombi. This patient had no DVT, and the diagnosis of PE was made 7 d after the onset of clinical symptoms.

Lung scan (high probability) alone was diagnostic in four patients (4%) in whom spiral CT could not be performed.

US findings confirmed the diagnosis of PE in 22 patients (23%). Of these patients, eight did not undergo spiral CT, and had a nondiagnostic lung scan. One patient did not have a lung scan and had a normal spiral CT, but the opacification of the pulmonary arteries was insufficient.

Ten other patients had a normal spiral CT and a nondiagnostic lung scan. In two of these patients the opacification of the pulmonary arteries was insufficient; six had a chronic respiratory disease (chronic obstructive pulmonary disease [n = 4] and asthma [n = 2]); two had left cardiac insufficiency.

Three patients had a normal spiral CT and a high-probability lung scan. In one of these patients, the opacification of the pulmonary arteries was insufficient; the other two patients had a chronic respiratory disease (tuberculosis sequelae and diffuse bronchiectasis, respectively). In these latter three patients the diagnosis of a thromboembolic event was made by US, so that pulmonary angiography was not performed.

Exclusion of PE

PE was ruled out by a normal lung scan in 18 of 132 patients (14%). A normal DD level (< 500 ng/ml) eliminated the diagnosis of PE in 41 patients (31%). A similar prior lung scan ruled out PE in 14 (11%) patients, 10 of whom had a normal CT and in four of whom spiral CT was not done. In all of these patients the DD level exceeded 500 ng/ml. Another 23 (18%) patients had a nondiagnostic lung scan associated with an obvious alternative diagnosis on spiral CT that could explain the clinical presentation. In 13 of these 23 patients the diagnosis was infectious pneumonia: the spiral CT showed alveolar opacities with an air bronchogram suggesting an infectious process. The microbiologic diagnosis was ascertained with bronchoscopy samples in six of these patients. All of these six patients had a good outcome with antibiotics and without anticoagulant therapy. In three other patients the diagnosis was cardiogenic pulmonary edema and the outcome was satisfactory with diuretic treatment. Two other patients had significant pleural effusion. One was due to left ventricular failure and rapidly resolved with diuretics. The other was associated with interstitial opacities on CT suggesting viral or atypical pneumonia, and disappeared within a few days, the only treatment being a macrolide antibiotic. An acute chest syndrome was diagnosed in one patient with sickle cell disease. Spiral CT showed pulmonary arterial compression by a mediastinal tumor in one patient. One patient had pulmonary arterial aneurysmal dilatations. The diagnosis of aortic dissection was made with spiral CT in one patient. One further patient with breast cancer presented with pulmonary artery hypertension, probably due to metastatic microemboli.

In 34 (26%) patients the diagnosis of PE was ruled out by a normal spiral CT scan, nondiagnostic lung scan, the DD level < 500 ng/ml, and negative US, in the absence of a prior lung scan or obvious alternative diagnosis.

Discordance Between Spiral CT and Lung Scan Findings

In two patients (0.9%) with discordance between findings on lung scan (high probability) and spiral CT (no thrombus), pulmonary angiography was done and was normal (Table 3). One patient had an exacerbation of asthma that was subsequently confirmed by lung function tests, and the other patient presented with postoperative atelectasis (after Cesarean section).

Follow-Up

Follow-up was obtained in 92% of the patients; 87% were followed in our institution and 5% were only telephoned. Nineteen patients (8%; 15 without PE and four with PE) had no follow-up. Two of the patients were foreigners and 17 moved. The results of follow-up are shown in Table 4.

In the group of patients without PE, two patients had thromboembolic events during the 3-mo follow-up period. One patient had an active lung cancer. He was already receiving oral anticoagulant therapy. The new event was diagnosed with spiral CT showing a segmental picture of thrombus. The second patient presented with PE 1 mo after our investigational protocol. The diagnosis was made with a A/ lung scan that showed a further segmental defect.

Thus, the risk of thromboembolic events during the 3-mo follow-up was 1.7% (95% confidence interval [CI]: 1.5 to 2.3%).

The mortality rate in the group without PE was 5.1% (95% CI: 4.1 to 5.9%). Six patients in this group died during this follow-up period, none of venous thromboembolism. The causes of death were cancer (n = 2), sepsis (n = 1), suicide (n = 1), pulmonary hypertension due to metastatic microemboli diagnosed at autopsy (n = 1), and respiratory distress (n = 1).

Among patients with PE, three patients (3.2%; 95% CI: 2 to 4%) had a recurrence of thromboembolic events. All three patients had cancer and were given thrombolytic therapy. The mortality rate in the PE group was 7.7% (95% CI: 6.7 to 8.7%). Seven patients died. The causes of death were cancer (n = 3), sepsis (n = 2), and thromboembolic events with advanced cancer (n = 2).

The difference between patients with and those without PE in terms of recurrence of thromboembolic events and mortality was nonsignificant (p = 0.44 and p = 0.45, respectively). No patient presented major bleeding episodes while receiving anticoagulant therapy during the diagnostic workup or during the follow-up period.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The present study was designed to assess the efficacy and safety of a noninvasive diagnostic protocol for PE that combined spiral CT, lung scan, plasma DD measurement, and in somes cases venous US. These noninvasive diagnostic modalities yielded a definite diagnosis in 99% of patients with suspected PE in whom the protocol was met. Adherence to the analysis-based decision strategy was high: only 19 patients (7%) did not have their data analyzed because of incomplete investigations, confirming the feasibility of large-scale systematic application of an analysis-based decision strategy at a clinical center. Only 1% of the patients underwent angiography because the noninvasive strategy was inconclusive.

Our diagnostic strategy appears to be highly safe: our protocol had low-grade iatrogeny. Only one complication occurred with spiral CT, involving a patient who developed renal failure that required hemodialysis for 3 wk, and this patient was known to be at risk of developing renal failure. Patients without PE were at low risk (1.7%) for venous thromboembolism during the 3-mo follow-up period. The recurrence rate for PE among patients without PE was equivalent to that found in other studies (1 to 1.9%) (8, 17).

PE was ruled out by a normal lung scan in only 14% of patients, as previous studies had reported (4, 16, 18), and no patient in this group had a thromboembolic event during the follow-up period. A normal DD level ruled out the diagnosis of PE in 31% of our patients. DD was assessed by means of a novel, rapid, automated ELISA. Previous studies confirmed that DD assay can be used as a screening test for PE (16, 19). In our study only one patient had a false-negative DD test, and for this patient the diagnosis of PE was delayed (7 d) after the onset of clinical symptoms. No patient in this group had a thromboembolic event during the follow-up period.

Eleven percent of our patients were considered free of PE when a prior lung scan was similar to the lung scan performed in our study. All of these lung scans were nondiagnostic. The previous lung scan was done in a clinical context without suspicion of PE in all of these patients (for evaluation of cardiac or pulmonary disease or follow-up of previous PE), and so could represent a new baseline for subsequent episodes of suspected PE. No patient in this group had a thromboembolic event during the follow-up period.

In our population, 18% of the patients had an obvious differential diagnosis that explained both the clinical presentation and the nondiagnostic result at lung scan. Spiral CT helped to establish the diagnosis, showing parenchymal abnormalities that specifically suggested infectious pneumonia, which was confirmed by bacteriologic analysis and/or a good outcome; cardiogenic pulmonary edema; aortic dissection; or pulmonary arterial compression by a mediastinal tumor. None of these patients experienced PE during the follow-up period. Two patients died, one of cancer and the other of pulmonary hypertension caused by metastatic microemboli. This ability of spiral CT to help provide an immediate alternative diagnosis has been emphasized in other studies (15, 20). Van Rossum and colleagues (15) found pleural or parenchymal abnormalities at spiral CT that could explain defects observed on lung scan in 57% of 42 patients with nondiagnostic findings on lung scan. Ferreti and associates (20) also ruled out PE through the use of spiral CT in 11% of 164 patients with clinically suspected PE and a nondiagnostic lung scan and who had obvious alternative diagnoses (infectious pneumonia, bronchiectasis, lung carcinoma, pneumothorax, and atelectasis). No patient in this group had any thromboembolic event during the follow-up period.

In 26% of our patients PE was ruled out by a normal spiral CT, nondiagnostic lung scan, DD level > 500 ng/ml, and negative result of venous US. One study (5) showed that negative results of serial noninvasive investigations for proximal DVT and nondiagnostic lung scans in patients with suspected PE were associated with a spontaneous good prognosis. Also, Ferreti and associates (20) found that only 0.4% of patients with a high clinical probability of PE, normal spiral CT, nondiagnostic lung scan, and normal US had PE confirmed by angiography. In the present study, two patients returned with objectively documented symptomatic venous thromboembolism on follow-up. One had a cancer, and both patients had a segmental or subsegmental PE. No patient died of PE. We probably missed subsegmental emboli. Spiral CT has a very high sensitivity for proximal PE, but its major limitation is a poor ability to reveal subsegmental arteries (11).

In our study, PE was diagnosed with spiral CT in 73% of the patients who had PE. Several major articles have concluded that spiral CT depicts thromboemboli in proximal pulmonary arteries with better than 95% specificity (11, 13). In our study spiral CT allowed the avoidance of pulmonary angiography in 30 of the 70 patients with nondiagnostic lung scans. The performance of spiral CT was obviously better than that of lung scan for confirming PE: only 34% of the patients with PE diagnosed with spiral CT had a high-probability lung scan. No patient had a positive spiral CT result and a normal lung scan, but two patients had a high-probability lung scan and normal spiral CT with normal angiography.

US was the main contributor to the diagnosis in 23% of our patients with PE, in accord with the results of another study (8). Over the past decade, some authors (5) have introduced the concept that diagnosis of DVT is an alternative to proving the presence of PE among patients with suspected of PE and inconclusive lung scan results. Sufficient data now show that this approach to the diagnosis of thromboembolic disease is valid (3).

Our findings indicate that a noninvasive strategy including spiral CT, A/ lung scan, DD assay, and in some cases US would permit a diagnosis of thromboembolic disease or its safe exclusion in 99% of patients with suspected acute PE. Only 1% of our patients with suspected PE required pulmonary angiography. As compared with the results of other studies that assessed noninvasive strategies, our results are obviously more satisfactory. Stein and colleagues (6) showed that in the case of a combination of clinically suspected PE, lung scan, and serial impedance plethysmography, pulmonary angiography remained mandatory in 29% of 662 patients. Moreover, their strategy is time-consuming, costly, requires several plethysmographies, and is inappropriate for patients with poor cardiorespiratory status. Perrier and colleagues (8) studied a diagnostic protocol in 308 consecutive patients with suspected PE that included assessment of clinical probability, lung scan, plasma DD measurement, and lower-limb US. This analysis-based decision strategy yielded a definitive noninvasive diagnosis in only 75% of patients with suspected PE. Recently, Perrier and colleagues (21) evaluated a diagnostic strategy in 918 patients with suspected DVT or PE, using DD measurement and US at the beginning of the diagnostic workup. For patients with a DD level exceeding 500 ng/ml and negative findings on US, clinical assessment was combined with lung scan, but pulmonary angiography was required in 11% of their patients with suspected PE. The 3-mo risk of thromboembolism in patients not given anticoagulants was 1.8%. These studies did not include spiral CT in their strategy. Only one study (20), involving 164 patients with suspected PE, prospectively evaluated the usefulness of spiral CT. Spiral CT enabled pulmonary angiography to be avoided in 91% of the patients. Bias in patient selection was a considerable drawback in this study, since only patients with lung scans of intermediate probability and normal venous US results were included and underwent spiral CT. This study did not really assess the contribution of spiral CT to the diagnostic strategy for all patients with suspected PE.

In conclusion, our study shows that a noninvasive strategy associating spiral CT, A/ lung scan, DD measurement, and in some cases lower-limb US allows confirmation or safe exclusion of acute PE in 99% of patients with suspected PE, without the need for pulmonary angiography. Other prospective studies are needed to assess the role of spiral CT rather than A/ lung scan as a first diagnostic step in patients with suspected PE.