INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Ventilation-perfusion (/) scintigraphy is the screening method of choice in patients suspected of having pulmonary embolism (PE) (1, 2). Although it is safe to withhold therapy when perfusion scintigraphy is normal, and treatment is usually started when the / scan result indicates a high probability of PE (e.g., segmental perfusion defects with preserved ventilation), PE can neither be excluded nor accepted when the / scan result is neither normal nor of high probability (3). In these cases, pulmonary angiography is usually recommended for reaching a final diagnosis (5, 7). However, this technique is often not used because of its invasiveness, high cost, potential risk in patients with acute respiratory failure and cor pulmonale, and limited availability (8, 9). Recently, new, noninvasive techniques for the diagnostic workup of PE have been proposed, such as D-dimer analysis, spiral computed tomographic angiography (SCTA), and magnitic resonance angiography, in order to decrease the number of required pulmonary angiograms.

Patients with chronic obstructive pulmonary disease (COPD) are considered to be at an increased risk for developing PE, possibly in relation to inactivity, heart failure, and advanced age (10, 11). Moreover, mortality from PE is thought to be increased in the presence of COPD (12, 13). However, differentiating PE from an exacerbation of COPD resulting from a respiratory infection, which is an important alternative diagnosis, is difficult, since clinical signs and symptoms of the two conditions overlap. Unfortunately, the diagnostic value of / scintigraphy may be decreased in patients with COPD (10, 14), since structural alterations in their pulmonary vasculature may have resulted in changes in / relationships (14, 15). Consequently, nondiagnostic / scan results will be encountered more often in patients with COPD, resulting in more frequent indications for pulmonary angiography.

The value of new, noninvasive techniques, including D-dimer analysis and SCTA, for the diagnosis of PE has not yet been differentiated with regard to their performance in the large subgroup of patients with COPD. D-dimer testing is nonspecific, and the D-dimer level may be increased in a variety of diseases, including infections (16, 17).

SCTA, as well as conventional pulmonary angiography, may also perform differently in patients with COPD. This could be due to the anatomic changes in the pulmonary vasculature in COPD, as well as to the possibility that small (subsegmental) emboli may produce (more severe) symptoms, leading to a suspicion of PE, and are more difficult to identify than larger emboli (18).

The purpose of the present study was to assess the effect of COPD on the clinical probability estimate (CPE), SCTA, D-dimer analysis, and / scintigraphy, as well as on the interobserver variability of pulmonary angiography, for the diagnosis of PE.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Subjects

The data reported in the study are from the Advances in New Technologies Evaluating the Localization of Pulmonary Embolism (ANTELOPE) study (22, 23). All consecutive in- and outpatients with clinically suspected PE and for whom perfusion scintigraphy was requested were considered for study entry. Exclusion criteria were age under 18 yr, pregnancy, and immediate need for thrombolytic therapy. Additionally, patients who had already had objective diagnostic examinations for their symptoms were excluded, as were patients for whom the diagnostic workup for PE could not be started within 24 h. The study was approved by the institutional review boards of all participating centers, and informed consent was obtained from all patients enrolled in the study.

Study Protocol

All patients gave a detailed clinical history and underwent a physical examination. The files of all patients who were thought to have COPD on the basis of anamnestic information were checked to confirm this diagnosis. COPD was considered to exist only when at least one of the following criteria was met: (1) a conclusive abnormal lung function test indicative of COPD; (2) a conclusive statement by a pulmonologist; or (3) a conclusive statement by a treating physician other than a pulmonologist in combination with prescribed COPD-related medication. Neither asthma without the presence of any of the other COPD criteria nor COPD medication alone was considered sufficient for a diagnosis of COPD.

All patients enrolled in the study underwent a diagnostic algorithm, as depicted in Figure 1. A clinical probability estimate scored on a visual analogue scale, a Tinaquant D-dimer assay, / scintigraphy, SCTA, and pulmonary angiography were performed as described earlier (3, 22) (see online data supplement). The complete study protocol was performed within 48 h after the perfusion scan, with a maximum time span of 24 h between the examinations. All examinations were interpreted independently, as described earlier (22, 23). PE was considered to exist when a pulmonary angiogram was abnormal or a / lung scan gave a high-probability result. A normal perfusion scan or a normal pulmonary angiogram excluded PE.

View larger version (28K): [in this window] [in a new window]   Figure 1.   Diagnostic algorithm used in the ANTELOPE study, including the number of patients given each test.

In addition, we assessed the influence of COPD on the implementation and execution of ^81mKr ventilation scintigraphy, SCTA, and pulmonary angiography.

Statistical Analysis

It can be calculated that with approximately 70 patients with COPD and 400 without COPD, differences of 15% to 20% can be identified (type I error = 0.05, two-sided; type II error = 0.2). The comparison of clinical characteristics of patients with and without COPD was done with Student's t test for quantitative data and with Pearson's chi-square test or Fisher's exact test for qualitative data. The Mann-Whitney test was used for comparing the duration of symptoms in patients with and those without COPD. The difference in interobserver assessment of pulmonary angiography findings in patients with and without COPD was assessed for statistical significance with the method of Kramer and Feinstein (28).

Multiple logistic regression analyses were performed to evaluate the influence of clinical parameters other than COPD on the accuracy of the diagnostic tests used in the study. Two-tailed values of p < 0.05 were considered to indicate statistical significance. Statistical analyses were performed with SPSS statistical software (version 8.0; SPSS, Inc., Chicago, IL).

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Subjects

During the study period, 1,162 consecutive patients suspected of having PE were considered for study entry at the six ANTELOPE study clinical centers. One hundred and seventy nine of these patients were excluded for the following reasons: age < 18 yr (n = 16), pregnancy (n = 11), indication for thrombolytic therapy (n = 5), diagnostic tests already performed (n = 43), and expected inability to complete the study protocol within 48 h or inability to give informed consent (n = 104). Of the 983 patients eligible for the ANTELOPE study, 627 (64%) gave consent to participate in the study. A diagnosis according to the study protocol could not be obtained for 110 patients (18%) because of withdrawal of consent (n = 10), medical reasons such as allergy to contrast medium (n = 13), discrepancy between local and central / scan results (n = 44), making of an alternative diagnosis before completion of the study protocol (n = 21), and technical reasons (n = 22). No important differences in demographic characteristics or test results were observed in the 517 patients with a final diagnosis and the group of 110 patients in whom no definitive diagnosis was obtained.

Prevalence of COPD and PE

Of the 627 patients who consented to participate in the study, diagnosis of COPD could be confirmed in 91 (15%). Table 1 gives the clinical characteristics of the patients with and without COPD. Patients with COPD were older (p < 0.001), had had symptoms suggestive of PE for a longer period (p = 0.01), and more often had clinical signs of congestive heart failure at study entry (p < 0.01) than did patients without COPD. Patients without COPD had more often undergone recent surgery, but this difference was not significant (p = 0.24). Both the frequency of previous venous thromboembolic disease and inpatient/outpatient status were similar in the groups with and without COPD.

PE was confirmed or refuted according to the study protocol (PE status known) in 69 of the 91 patients with COPD (76%), as compared with 448 of the 536 patients (84%) without COPD (p = 0.07). The prevalence of PE was similar in the two groups: 29% of the patients with COPD versus 31% of those without COPD proved to have PE (p = 0.71).

CPE

A CPE, as made by the treating physician before objective testing, was available for 413 of the 517 patients (80%) with a final diagnosis. The remaining 104 patients (20%) had to be excluded from the analysis because the estimate was not obtained at all or was not obtained before the / lung scan result was known. The CPE categories for patients with and without COPD are presented in Table 2. The distribution of the CPE categories was comparable among patients with and without COPD (p = 0.33). For all CPE categories, the prevalence of PE was comparable in patients with and without COPD (Table 2).

Compression Ultrasonography

Compression ultrasonography was performed in 479 patients whose PE status was known. The sensitivity of compression ultrasonagraphy for PE was found to be somewhat, although nonsignificantly, greater in patients with COPD (seven of 20; 35%) than in those without COPD (28 of 133; 21%) (difference = 14%; 95% confidence interval [CI]: 8 to 36%; p = 0.22). A false-positive test result for PE was obtained for one of the 45 patients (2%) with COPD and for seven of the 281 patients (2%) without COPD. The positive and negative predictive values of ultrasonography were comparable for patients with and without COPD.

/ Scintigraphy

A perfusion scan result was available for 509 of the 517 patients (98%) with a final diagnosis. The frequency of perfusion defects was 75% (50 of 67 patients) among COPD patients and 51% (224 of 442 patients) among patients without COPD (difference = 24%; 95% CI: 13 to 35%; p < 0.001). These defects were more often segmental or larger in patients with COPD (46 of 50 defects; 92%) than in patients without COPD (181 of 224 defects; 81%) (difference = 11%; 95% CI: 2 to 20%; p = 0.02). The proportions of patients in whom ^81mKr ventilation scintigraphy could not be performed or was inadequate were comparable in the groups with and without COPD (2% versus 1%), respectively. As shown in Table 3, / scan results in patients with COPD were more frequently nondiagnostic (46% versus 21%) (difference = 25%; 95% CI: 13 to 38%; p < 0.001) than in those without COPD, notwithstanding the higher prevalence of segmental or larger perfusion defects among these patients. Interestingly, the prevalence of PE was similar (Table 3) in the two groups when the / scan result was nondiagnostic. Overall, the diagnostic performance of a high-probability / scan result for PE did not differ significantly among patients with and those without COPD (Table 4).

Of the 111 patients (28 with COPD) with a nondiagnostic / scan result and a negative result of compression ultrasonography, 16 patients (14%) had PE on pulmonary angiography (three with COPD).

D-Dimer Testing

D-dimer analysis was performed on 313 of the patients with a final diagnosis (61%). The distribution of D-dimer results and prevalence of PE for patients with and without COPD are given in Table 5. Neither the distribution of D-dimer results nor the diagnostic accuracy of the D-dimer assay were influenced by the presence of COPD (Table 4).

SCTA

SCTA results were available for 259 of the 282 patients (92%) who had an abnormal perfusion scan result and a final diagnosis. The proportions of patients on whom SCTA could not be performed (e.g., known allergy to contrast material) or for whom an inadequate test result was obtained (for patient-related reasons, such as an open foramen ovale or superior vena cava syndrome, motion artifacts, or technical failure of the CT scanner) were similar for patients with and without COPD (13% versus 10%) (difference = 3%; 95% CI: 7 to 12%; p = 0.58). SCTA was more often performed on patients with COPD (47 of 69; 68%) than on patients without COPD (212 of 448; 47%) (difference = 21%; 95% CI: 9 to 33%; p < 0.001), owing to the higher prevalence of nondiagnostic / scan results in patients with COPD (Table 6). Hence, an SCTA result that was negative for PE was statistically significantly more often found in patients with than in those without COPD (77% versus 55%) (difference = 22%; 95% CI: 8 to 36%; p = 0.002). As can be seen in Table 4, the diagnostic performance of SCTA was not influenced by the presence of COPD.

Pulmonary Angiography

Pulmonary angiography was indicated in 66 patients with COPD and in 205 patients without COPD. Adequate pulmonary angiograms were obtained in similar proportions of patients in the two groups, consisting of 68% (45 patients) with COPD and 69% (142 patients) without COPD. The interobserver agreement for the two groups was comparable ( = 0.76 and  = 0.78, respectively; p = 0.86).

Additional Observations

For / scintigraphy, D-dimer testing, and SCTA, the test performance characteristics are given in Table 4. Neither congestive heart failure nor age had an influence on SCTA or D-dimer test performance. However, both age and the presence of congestive heart failure increased the number of nondiagnostic / scan results. A multiple regression analysis was performed with age, congestive heart failure, and COPD as independent variables. After correction for age and congestive heart failure, the differences in diagnostic performance of / scintigraphy in patients with and without COPD persisted.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

As expected, our study of a representative group of consecutive in- and outpatients with suspected PE showed that COPD significantly affects the distribution of / scan results. However, in 36 of the 91 patients (39%) with COPD, the / scan result was still conclusive, notwithstanding the more frequent nondiagnostic / scan result for patients with COPD. Moreover, the prevalence of PE both overall and in relation to the results of / scintigraphy was similar in patients with and without COPD.

The performances of the CPE, D-dimer testing, SCTA, and pulmonary angiography did not differ among patients with and those without COPD. Furthermore, the presence of COPD did not influence the failure to perform any of the diagnostic tests or to obtain adequate test results.

In the present study, 15% of patients with suspected PE were diagnosed as having COPD, which is well within the range found in previous studies (i.e., 14% [12] to 20% [29]). The observed comparability of PE prevalence among patients with and without COPD (29% and 31%, respectively) in our study is in good agreement with the findings of Turkstra and colleagues (39% and 43%, respectively), but differs from the results of the Prospective Investigation of Pulmonary Embolic Disease (PIOPED) study in which a lower prevalence of PE was found in the presence of COPD (22% versus 30%, p = 0.07) (12, 29, 30). In our study, the CPE was based on all available information, inevitably including the patient's COPD status. However, we could not observe a difference in the distribution of the CPEs made by the physician among patients with and without COPD. Furthermore, a disease other than PE, which was considered to be as likely as PE to cause the patient's pulmonary symptoms, was reported with a similar frequency in patients with and without COPD (data not shown).

Our results for the diagnostic value of / scintigraphy in relation to COPD are in agreement with those of two large, well performed, previously reported studies (29, 31). In these studies, patients with COPD also more often had an abnormal perfusion scan result, which, in combination with ventilation scintigraphy, resulted more often in a nondiagnostic / scan result.

Although patients with COPD more often had a nondiagnostic / scan result, pulmonary angiography revealed that the prevalence of PE in these patients did not differ from that in those patients without COPD who had a nondiagnostic / scan result. Therefore, the usefulness of the combination of / scanning with pulmonary angiography is hardly influenced by the presence of COPD. This again is comparable with the findings of earlier studies (29, 32).

Limitations of the present study apply to the criteria used for defining COPD. Since it was our intention to use criteria for the presence of COPD that would be clinically applicable to patients with suspected PE, the diagnosis of COPD was based on the clinical information available at the time of enrollment in the study. Although some misclassification of COPD might have occurred, this did not influence our ability to observe differences in the distribution of / scan results between patients with and without COPD. Our reliance on clinical information prohibited us from assessing the influence of severity of COPD on the performance of the diagnostic tests. However, in the study by Lesser and colleagues, the distribution of / scan results in the 35 patients with moderate or severe COPD did not differ from the distribution in all 108 patients with COPD (32).

Because we included consecutive patients with suspected PE, a final diagnosis, as in clinical practice, could not be obtained for all patients. However, we believe that an 18% rate of failure to obtain a final diagnosis according to our strict protocol is reasonably low and acceptable. Moreover, for ethical reasons, based on previous data, not all diagnostic tests were done on all patients. Therefore, the applicability of the results of our study is limited in patients with a normal / scan.

Although the study was reasonably large, the relatively small number of patients with COPD limited its power to exclude small differences in test characteristics between patients with and without COPD. Fortunately, small differences are likely to be of minimal clinical relevance.

It is unlikely that important biases influenced our results with respect to the effect of COPD on the performance of diagnostic tests. First, the protocol for our study was executed equally well in patients with and without COPD. Second, the proportions of patients for whom an adequate test result was obtained were similar in the groups of patients with and without COPD. Third, the / scans, SCTA scans, and pulmonary angiograms were assessed by readers who were blinded to the patients' COPD status and the results of other examinations.

To summarize, we conclude that the presence of COPD does not influence the performance of the CPE, D-dimer test, SCTA, or pulmonary angiography. Furthermore, although more nondiagnostic / scan results can be expected in the presence of COPD, which decreases the cost-effectiveness of this procedure, / scintigraphy remains an informative, non-invasive screening test in patients with COPD. Normal and high-probability / scan results are obtained in a substantial proportion of patients with COPD, and the prevalence of PE in the presence of a nondiagnostic / scan result is comparable for patients with and without COPD. Several diagnostic algorithms including the CPE, / scintigraphy, D-dimer testing, SCTA, and/or pulmonary angiography have been proposed or evaluated for patients with suspected PE (27, 33). Our observations indicate that these alternative strategies will be equally applicable to patients with and without COPD, although the cost-effectiveness of / scintigraphy may vary.