Laboratory investigation in the diagnosis of pulmonary thromboembolism

Laboratory findings were compared with lung scans in a prospective study of 260 patients undergoing ventilation-perfusion (V/Q) lung scanning for suspected pulmonary thromboembolism. The best discrimination between different lung scan results was obtained from the level of plasma cross-linked fibrin degradation products, every patient with a scan indicating a high probability of thromboembolism having detectable levels. An acute phase response was demonstrated in patients with pulmonary thromboembolism by a raised neutrophil count and elevated levels of plasma fibrinogen and serum C-reactive protein. A normal level of serum C-reactive protein and/or plasma cross-linked fibrin degradation productions in blood taken within 4 days of onset of symptoms virtually excluded the diagnosis of pulmonary thromboembolism. Detection of free plasma DNA was not helpful in discriminating between groups with different lung scan results. Discriminant analysis was used to assess the variables examined and to derive diagnostic models. An accuracy of 78 per cent was obtained with one model for classifying test patients according to the three lung scan classes of low, intermediate and high probability. A second model, for distinguishing patients with a low and a high probability of pulmonary thromboembolism on the basis of lung scans, and a third for predicting those with a low probability on lung scan, were accurate in 94.6 per cent and 83.5 per cent of patients respectively. Discriminant models could be used in the diagnosis of pulmonary thromboembolism, especially when diagnostic imaging is not available.