LAUSR

www.thelancet.com Published online May 23, 2017 http://dx.doi.org/10.1016/S0140-6736(17)31291-6 1 In 1964, Sasahara and colleagues described the use of selective pulmonary angiography for the diagnosis of pulmonary embolism. 25 years later, the Prospective Investigation of Pulmonary Embolism Diagnosis established criteria for diagnosing acute pulmonary embolism with the use of a non-invasive technique, ventilation-perfusion lung scan. The decade that followed marked the increased use of computed tomographic pulmonary angiography (CTPA), which has since been undergoing continuous technological improvement and is now established as the gold standard in pulmonary embolism diagnosis. However, even the most sophisticated and accurate test can yield useless or even misleading findings if done without a proper clinical indication. This is particularly crucial in the setting of suspected acute pulmonary embolism, in which individual symptoms and clinical signs are, as a rule, non-specific. To address this challenge, two complementary early diagnostic steps were developed and interfaced with CTPA in multistep algorithms: first, clinical prediction rules, which provide a semi-quantitative assessment of the pretest probability of pulmonary embolism based on the combination of predisposing conditions, symptoms, and clinical findings; and second, circulating D-dimer concentrations, a highly sensitive biomarker of ongoing fibrinolysis and thus, indirectly, thrombosis. The successful validation and excellent performance of strategies combining these elements with multidetector CTPA apparently closed the chapter on pulmonary embolism diagnosis 10 years ago. In fact, with the exception of some technical improvements, no major changes in recommended diagnostic tests and algorithms can be found in the 2014 update of the European Society of Cardiology guidelines on pulmonary embolism compared with the 2008 version. It seems that the good could not just get even better. So, are diagnostic tests for suspected pulmonary embolism condemned to stagnation, being a victim of their own success? Tom van der Hulle and colleagues’ YEARS study, published in The Lancet, challenge this notion by showing that clinically meaningful adjustments are still possible. By combining three elements of the Wells’ rule (ie, clinical signs of deep vein thrombosis, haemoptysis, and whether pulmonary embolism is the most likely diagnosis) with D-dimer testing in a single step, and by prospectively defining cutoff values for exclusion of pulmonary embolism without proceeding to an imaging test, van der Hulle and colleagues show that CTPA could be avoided in an additional 13% (95% CI 10–15) of the intentionto-diagnose patient population compared with standard algorithms. This reduction could be achieved, according to the worst case scenario, at an only 0·78% (23 of 2951 patients, 95% CI 0·49–1·2) failure rate with regards to the 3-month incidence of recurrent symptomatic or fatal venous thromboembolism. In the per-protocol patient population, the failure rate was even lower, 0·51% (15 of 2943 patients, 95% CI 0·31–0·84). The results further suggest that the YEARS approach might do slightly better even when compared with a strategy based on age-adjusted D-dimer concentrations. The results of YEARS study are more than another technical refinement of existing algorithms: they address a true medical need in contemporary practice. Indeed, an increase in the use of CTPA has consistently been observed in population studies describing trends in the management of the disease in the past three decades. In one of these studies, the incidence of diagnosed pulmonary embolism increased by as much as 81% (from 62 to 112 cases per 100 000) following the introduction of CTPA (1998–2006) in contrast to Diagnosis of pulmonary embolism: progress after many YEARS