INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The diagnosis of pulmonary embolism by noninvasive means such as ventilation/perfusion lung scan (/ scan), venous ultrasound of the lower extremities, D-dimer assay, clinical assessment, or by a combination of these has remained difficult (1). The Prospective Investigation of Pulmonary Embolism Diagnosis (PIOPED) trial found that: (1) a normal / scan excluded pulmonary embolism; (2) the combination of a high probability / scan plus a high clinical suspicion was diagnostic for pulmonary embolism (96% with pulmonary emboli), and (3) a low probability or normal lung scan with a low clinical suspicion made the diagnosis of pulmonary embolism unlikely (4% with pulmonary emboli). However, these combinations occurred in only 20% of patients studied (180 of 887) (2). The routine use of venous ultrasound has been estimated to reduce the need for angiography further but as many as one-third will still require angiography (3). The need for a better noninvasive diagnostic approach has resulted in reevaluation of the D-dimer (D-D) assay in patients with suspected acute pulmonary embolism (4).

D-D are a specific degradation product of cross-linked fibrin. The pairing of the D domains of fibrin monomers occurs only with full cross-linking of the monomers. However, the presence of D-D alone is not useful in establishing the presence of clot formation with pulmonary embolism or deep venous thrombosis (DVT) because of its low specificity. D-D are elevated in disseminated intravascular coagulation, pregnancy, severe infection, liver disease, surgery, trauma, and malignancy (5, 6). Therefore elevated D-D may be present when pulmonary embolism or proximal DVT are absent.

Although a positive D-D is of no value, a negative D-D is potentially useful excluding the presence of pulmonary embolism. D-D can be measured by enzyme-linked immunosorbent assay (ELISA) or latex agglutination assay. A review of the published literature found that in the diagnosis of pulmonary embolism, D-D as measured by ELISA had a negative predictive value (NPV) of 94.2% (range 91 to 100%, 95% confidence interval [CI] 91.2 to 97.2%) (7). Withholding anticoagulation in outpatients with a D-D concentration below 500 µg/L by ELISA was associated with only a 1% risk of thromboembolic events during 3-mo follow-up (8). However, the ELISA has limited use because it is a time-consuming test, it requires skilled personnel, it is designed for batch testing, it is not available in most institutions, and when available it is usually not performed on nights or weekends. A new, fully automated ELISA has shown great promise (9), but has not yet become widely available and needs further testing. In comparison, the latex agglutination assay for D-D is readily available on an individual basis and can be performed in less than 30 min in most clinical laboratories. This assay, using many different commercially available kits, has had a negative predictive value ranging from 67 to 97% (mean 89.3%, 95% CI 83.6 to 94.9%) (7). This variability in previous results has led many to conclude that the latex agglutination assay is not useful in the diagnosis of pulmonary embolism (7, 10, 11). However, the latex agglutination assay for D-D has been evaluated in only a small number of patients in whom pulmonary angiography was used to conclusively diagnose pulmonary embolism (12, 13).

In this study we evaluated D-D measured by five different available latex agglutination assays and by ELISA in the diagnosis of pulmonary embolism. Pulmonary angiography was used as the diagnostic standard in all patients.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Subjects

From August 1, 1994 to June 30, 1995 at the Massachusetts General Hospital, Boston, MA, blood samples were obtained prospectively, at the time of angiography or within 24 h of angiography, from 103 patients undergoing pulmonary arteriography for the suspicion of acute pulmonary embolism. Blood was obtained from all patients from whom consent could be obtained. Patients were excluded who had a history or suspicion of chronic pulmonary embolism, defined as progressive dyspnea over months with signs on physical exam of right ventricular failure without signs or symptoms of left ventricular failure suggesting pulmonary hypertension. Clinical data and history were obtained from chart review. Study design was approved by the Massachusetts General Hospital Subcommittee on Human Research.

Pulmonary Arteriography

Bilateral pulmonary arteriography was performed as previously described (2). Femoral vein puncture using a Seldinger technique to introduce a multisidehole pigtail 7.1 French catheter was used. Using a tip-deflecting wire, the right and left pulmonary arteries were selected. Small amounts of contrast material (8 to 10 ml) were injected by hand to evaluate the patency of the inferior vena cava fluoroscopically. The catheter was directed into the main pulmonary artery. Filming for each of the pulmonary arteries was initially in the anterior- posterior projection and a subsequent ipsilateral posterior oblique projection. Injections utilized 20 to 35 ml/s for 40 to 50 ml of iodinated contrast material (350 to 370 mg of iodine per milliliter, either high-osmolar or low-osmolar contrast agents). Film rates were three per second for a total of 3 s followed by one film per second for 5 s. Filming was magnified, depending on the size of the lungs. A 12:1 grid was used and roentgenographic factors were in the range of 70 to 80 kilovolts and 0.025 to 0.040 s at a 1,000 mA with a large focal spot of 1 mm in diameter. Magnification films were obtained with an air-gap technique. Roentgenographic factors were in the range of 78 to 88 kilovolts, using a small focal spot 0.3 mm in diameter. If emboli were not identified, injections were repeated and additional views were obtained of the areas suspicious for pulmonary embolism.

Blood Sample Analysis

Blood samples were collected from pulmonary arteriography catheters or venipuncture in 3.8% sodium citrate collection tubes. Samples were centrifuged at 2,500 × g for 10 min. Plasma was removed and stored at 70° C and analyzed by technologists blinded to the results of pulmonary arteriogram.

Latex Agglutination Assays

Five latex agglutination assays were evaluated. These included the D-Di test (Diagnostica Stago, Asnieres, France) Fibrinosticon Latex (Organon Teknika Corp., Durham, NC), D-Dimer Assay (Pacific Hemostasis, Huntersville, NC), Accuclot D-Dimer (Sigma Diagnostics, St. Louis, MO) and D-dimer Wellcotest (Murex Diagnostics Limited, Dartford, UK). Only less than or greater than 500 ng/ml was determined. Results were expressed as fibrinogen equivalent units (FEU, ng/ml). FEU is the quantity of fibrinogen initially present that leads to the observed level of D-D. One D-D unit is approximately half of one FEU because two fibrin monomers must cross-link to form one D-D. Results were expressed as positive when > 500 ng/ml FEU or negative when < 500 ng/ml FEU. All analyses were performed by one experienced technologist blinded to the results of the pulmonary arteriogram.

ELISA

All results were compared with a quantitative ELISA, Asserachrom D-Di (Diagnostica Stago). Analysis in the laboratory required approximately 3 h.

Statistical Methods

Data were analyzed using Statview 4.5 (Abacus Concepts, Inc., Berkeley, CA). Continuous variables were compared between groups using analysis of variance and subsequent multiple comparisons by the Scheffe test. Nominal variables were compared between groups with a contingency chi-square test. Confidence intervals for nominal variables were calculated using an exact binomial technique. Significance was set at p < 0.05.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Patient Population

Of 103 patients undergoing pulmonary arteriogram for suspicion of acute pulmonary embolism 34% (35 of 103) had detectable pulmonary emboli. The average age for all patients was 59 yr with a range of 16 to 87; 44% were female. The risk factors for pulmonary embolism are shown in Table 1. There was no difference in this group between those with pulmonary embolism and those without pulmonary embolism in age, sex, or risk factors.

Latex Agglutination Assays

The five latex agglutination assays had a sensitivity of 97 to 100%, a specificity of 19 to 29% and NPV of 94 to 100% (Table 2). Zero to one patient out of 35 patients with pulmonary embolism had a negative D-D by the five latex agglutination assays (Figure 1 shows the results of the D-Di test, Diagnostica Stago). The one patient with a negative D-D and a positive pulmonary arteriogram had an elevated D-D by ELISA. Pulmonary embolism was correctly excluded in 13 to 20 of 68 (19 to 29%) patients without emboli.

View larger version (23K): [in this window] [in a new window]   Figure 1.   Representative latex agglutination assay (D-Di test; Diagnostica Stago, France) reported as positive/negative compared with ELISA (Asserachrom D-Di; Diagnostica Stago) in ng/ml. Open circles: pulmonary arteriogram negative for pulmonary embolism; closed circles: pulmonary arteriogram positive for pulmonary embolism. Dotted line represents < 500 ng/ml cutoff.

ELISA

The ELISA at a cutoff of 500 ng/ml had a sensitivity of 100% and a specificity of 13%. This resulted in a NPV of 100% (Table 2), but pulmonary embolism was correctly excluded in only 9 of 103 (9%) patients.

Subgroup Analysis

The usefulness of the latex agglutination assays for D-D was limited in certain subgroups. In patients with surgery within 3 mo with a known malignancy, and total bilirubin > 34 µmol/L (> 2.0 mg/dl) the occurrence of a negative D-D was low owing to very poor specificity. Less than 4% of patients with a negative pulmonary angiogram and surgery within 3 mo, active malignancy or abnormal liver function had a negative D-D by latex agglutination (Table 3). In patients without surgery within 3 mo, malignancy or total bilirubin of > 34 µmol/L (> 2.0 mg/dl), the NPV was 100% for all five latex agglutination assays. In this group 10 to 12 of 22 (45 to 54%) with a negative pulmonary angiogram also had a negative D-D by latex agglutination assay (Table 3).

In patients with abnormal liver function tests, elevated D-dimers, and negative pulmonary angiograms, the total bilirubin ranged from 2.2 to 7.1 mg/dl. Both lactate dehydrogenase (LDH) and serum glutamic-oxaloacetic transaminase (SGOT) were not helpful in determining which patients would have elevated D-dimers. A few patients had very high LDH (> 400 U/L, n = 4) or SGOT (> 90 U/L, n = 2) but still had negative D-dimers.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

We found that in our population of patients, the latex agglutination assays for D-D had a NPV of 94 to 100% and the ELISA assay for D-D had a NPV of 96% in the diagnosis of acute pulmonary embolism (Table 2). D-D assays were most useful in patients without surgery within 3 mo, without active malignancy, and with normal liver function because of the very poor specificity of the tests in these conditions (Table 3).

A major criticism of previous clinical trials of the latex agglutination D-D in the diagnosis of pulmonary embolism is the lack of use of a "gold standard" (10). The strength of this study is the use of pulmonary arteriograms as the diagnostic criteria. Most of the studies involving the latex agglutination assay for D-D in the diagnosis of acute pulmonary embolism do not use the pulmonary angiogram as the diagnostic criteria for pulmonary embolism (Table 4). These studies have used / scan alone (14, 15); / scan and pulmonary arteriogram when needed (11); or a combination of clinical probability, / scan, ultrasonography of leg veins, and when necessary a pulmonary arteriogram (16) using a decision analysis- based strategy (21). The large range in the NPV of the latex agglutination assay for D-D (67 to 99%) may reflect the failure of the other diagnostic techniques to clearly delineate the presence or absence of pulmonary emboli. Two previous reports looked at the latex agglutination assay for D-D in the diagnosis of acute pulmonary embolism using pulmonary angiography (12, 13). These studies also found a high NPV of 96 and 100% for the latex D-D (Table 5).

The PIOPED studies (2) show that in patients with high-probability lung scans, pulmonary embolism is present in 87%; in patients with intermediate lung scans pulmonary embolism is present in only 30%; and in patients with low-probability lung scans, pulmonary embolism is, nevertheless, present in 14%. These data indicate that in studies relying on lung scans and not using pulmonary angiography as the diagnostic endpoint, some patients will be misdiagnosed. A true NPV cannot be determined if the diagnostic endpoint is not accurate.

A possible limitation of this study and some previous studies is that the plasma samples were frozen and saved for later analysis in order that the ELISA and the five latex agglutination tests could all be done. In our study all assays were performed by one technologist who although blinded to the angiogram results does have substantial laboratory experience with the interpretation of latex agglutination assays. Because the latex agglutination assay depends on human observation and exact timing of that observation, the results obtained may be more accurate than those obtained from a less experienced technologist in a clinical laboratory.

The length of time the D-D levels remain elevated after the occurrence of pulmonary embolism may limit the usefulness of D-D in the diagnosis of pulmonary embolism. Though the majority of patients have elevated D-D levels as long as 12 d after diagnosis (17), in a few patients with pulmonary embolism the D-D level returns to normal limits by the seventh day after diagnosis (4). In our study we carefully screened patients for suspicion of acute pulmonary embolism. Of our patients, 87% (88 of 103) presented with chest pain, shortness of breath, hemoptysis, or other signs of acute distress (such as unexplained hypoxemia, acute confusion and hypotension) within 7 d; 79% (81 of 103) experienced symptoms within 3 d.

It is not known how long after the initiation of clot formation the D-D becomes elevated in the systemic circulation. In our study, in one patient with pulmonary embolism and a negative latex agglutination D-D on the day of presentation, the D-D level subsequently rose and remained elevated for 7 d. Bounameaux and coworkers (4) also reported one patient with a D-D level of < 500 ng/ml with a pulmonary embolism who had an elevated level on subsequent measurements. Therefore, the D-D may not be useful in patients presenting immediately after onset of symptoms but this appears to occur in a minority of patients. There are no available data on the usefulness of repeated testing in patients with a negative D-D at presentation.

Previous investigators have reported that the D-D assay may be more useful in outpatients than inpatients (4, 22). In our patients who were hospitalized at the time of the diagnosis of pulmonary embolism, D-D were negative in only 4 to 7% with the different latex agglutination assays. This high rate of positive D-D was related to the 70% incidence of surgery, malignancy, or elevated liver enzymes in inpatients. The presence of comorbid conditions was more important than the distinction between inpatients and outpatients.

We conclude that a negative D-D assay by the rapidly performed and inexpensive latex agglutination assays is a clinically useful tool in excluding the presence of pulmonary embolism in patients with symptoms present for less than 1 wk with normal liver function, no active malignancy, and no surgery within 3 mo. Though a negative D-D should never prevent further investigation of pulmonary embolism if the clinical suspicion for pulmonary embolism is high, the latex D-D assay can be clinically useful in patients with low to intermediate clinical suspicion of pulmonary embolism. The latex agglutination D-D should undergo further prospective clinical trials. Areas that need to be addressed include: effectiveness of the different available latex agglutination assays, usefulness of serial assays, and identification of patients in which the test is not reliable.