In Reply to: Re: Case 0599_08 -- Suspected Pulmonary Embolus in a 17-year-old posted by m victoria marx on May 05, 1999 at 15:16:15:
One scenario in which CT has a place in the diagnostic algorithm is in the diagnosis of chronic PE. Bergin et al (Radiology 1997)demonstrated CT to be slightly more accurate than arteriography in the diagnosis of central chronic PE (.79 vs .74), but not as accurate as arteriography for segmental vessel disease (.75 vs arteriography as the gold standard).
Drucker's article is an extremely important one, in that it clearly demonstrates that the sensitivity of CT for PE is too low for it to be useful as a screening study. It's use in that manner places patients at risk: The correct diagnosis and treatment of venous thromboembolic disease (VTD) has an enormous impact upon patient morbidity: In a review of studies comprising over 5,000 patients, Douketis et al (JAMA 1998) found an incidence of fatal PE in only 1.5% of patients who were appropriately anticoagulated after initial diagnosis of PE. Compare that to the 30% mortality attributed to untreated PE. Contrarily, anticoagulation is associated with high morbidity (5-25%) and mortality (1-2%) rates, and should not be begun until the diagnosis is certain. The 1990 PIOPED study (933 pts) demonstrated PE in 88% of patients with high prob scans, in 33% of pts with intermediate prob scans, in 16% (subsequently reduced to 12% with pt f/u) of low prob scans, and 4% of patients with near normal or normal scans. Lesser et al (Chest 1992), however, showed that the VQ scan was of little worth in patients with COPD. Given the need for definite diagnosis, and the inadequacy of VQ scanning in providing that diagnosis, it's not surprising that CT, which can demonstrate PE, was initially met with great fanfare. However, after Ferretti et als' study of 164 pts (Radiology 1997) suggested that CT "allowed accurate diagnosis of acute PE" (despite the fact that PE was missed in 6 (5.4%) pts, one of whom died from PE), other investigators have demonstrated even worse sensitivity and specificity of CT for PE. Drucker's study showed sensitivities of 53 and 60% for two different sets of reviewers. Brink et al (Radiology 1997) showed in a pig model that CT was capable of demonstrating only 30% of segmental PE. Pulmonary arteriography remains the gold standard for the diagnosis of PE, but is widely underutilized, probably because of unfounded concerns for procedural risks. My review of all the major series of pulmonary arteriography (Mills 1980; Perlmutt 1987; Stein 1992; Zuckerman 1996; Hudson 1996, among others) found a procedural mortality rate <1% (0 in the two most recent series, which used nonionic contrast), a major complication rate of 1%, and a minor complication rate of 1-5%.
I strongly disagree with the editor's statement that the diagnostic accuracy of digital subtraction pulmonary arteriography is less than that of CTA. On the contrary, Van Beek et al (Radiology 1996) showed less interobserver disagreement (4-11%) on DSA pulmonary arteriograms than on cut film (20-36%). We and Hagspiel independently demonstrated (prospective clinical studies) that DSA was superior to cut film in the diagnosis of PE (Radiology, April 1998), and thus much superior to CTA. I believe that all patients who do not have a definite diagnosis of VTD (such as one with a high probability VQ scan and strong clinical suspicion of PE or one in whom US demonstrated LE clot) should get a pulmonary arteriogram.