Re-emergence of an "orphan" test for pulmonary embolism.

O ften clinically suspected, pulmonary embolism (PE) is a potentially fatal disease. Because highly effective but hazardous therapy exists, accurate diagnosis is extremely important. Indeed, much of the research and publications on PE, such as the Prospective Investigation of Pulmonary Embolism Diagnosis study, pertain mainly to methods of diagnosis. Prior to the development of imaging techniques, the diagnosis of PE was based on a triad of dyspnea, hemoptysis, and chest pain, later shown to be nonspecific and unreliable, for the clinical spectrum can range from no symptoms at all to sudden death. The chest radiograph has been equally nonspecific and unreliable in diagnosis of PE. In 1959, turning to the burgeoning field of clinical pulmonary physiology, Eugene Robin et al1 suggested, “a physiologic approach to the diagnosis of pulmonary embolism” based on the Bohr equation. The simple concept utilized was that embolized parts of the lung are ventilated but not perfused so that alveolar CO2 tension in these areas is low. Mixing of gas from perfused and unperfused units leads to an alveolar (measured as end-tidal) Pco2 abnormally lower than arterial Pco2. In practice, an arterialalveolar CO2 gradient of . 5 mm Hg was considered abnormal. In 10 patients with clinically suspected PE studied by Robin et al,1 a gradient of that magnitude was found in 7 patients (and in 5 of 10 patients in a subsequent publication from the same institution).2 The authors stated that this test is useful to diagnose large emboli only. They also indicated that the test is nonspecific and could be abnormal in the presence of rapid, shallow breathing of any cause, or a variety of other lung diseases. In 1962, Colp and Williams3 reported an arterialalveolar (A-a) gradient of . 5 mm Hg in only three of seven patients with clinical diagnosis of PE. These patients were studied more than a week after onset of symptoms, and it was pointed out that with time both clot lysis and collateral circulation would minimize the gradient. In support of this, one of the patients was restudied after another PE had occurred, and his A-a gradient was abnormal. Also, decreased ventilation due to atelectasis, bronchoconstriction, pleurisy, or infarction would diminish the gradient. Hatle and Rokseth4 studied patients whose diagnosis was confirmed by angiogram or at postmortem. They also note lack of an abnormal A-a gradient with small emboli or with passage of time since embolization; however, all but one of their patients with large PE had gradients . 5 mm Hg. These authors note that false-positive results in patients with COPD, pneumonias, congestive heart failure, or nonembolic shock could be corrected by having the patient expire fully and using end-expired rather than end-tidal CO2 measurement in the calculation. (A significant difference between end-tidal and endexpired CO2 is due to uneven ventilation and should not occur in normal subjects or those with PE). Due to all of these problems with the alveolar dead space test proposed by the late Dr. Robin, in 1977 he eschewed it. He wrote an editorial5 titled, “The Emperor May Have No Clothes,” in which he discussed overdiagnosis of PE, overreliance on the then-popular ventilation-perfusion scan (which is merely an imaging method of detecting alveolar dead space), and the need for frequent pulmonary angiography. In this editorial, he did not even mention capnographic measurement of alveolar dead space or CO2 gradient! Therefore, this test, which he fathered, now became an “orphan” test. More recently, in 1986, Burki6 resurrected the orphan but calculated physiologic dead space based on expired CO2 instead of alveolar. They reported a dead space ratio of . 40% in all of 16 patients with abnormal angiograms. Twenty-nine patients with normal perfusion lung scan results had either normal dead space ratio or abnormal spirometry findings (thought to represent some lung disease other than