Age-adjusted D-dimer Cut-off to Rule Out Pulmonary Embolism in the Emergency Department : A Real Life Impact Study

BACKGROUND

PE diagnostic strategies Pulmonary embolism (PE) is a frequently suspected diagnosis in the emergency room (ER) in patients presenting with shortness of breath and/or chest pain without any obvious cause identified. Historically, PE diagnosis required a pulmonary angiography, a costly and invasive procedure with associated morbidity and even mortality. Over the last 25 years, evolving diagnostic strategies have been developed and widely validated in prospective multicentre management outcome studies, with the aim of confirming or excluding the diagnosis of PE safely enough to avoid realising pulmonary angiography. Nowadays, PE diagnosis therefore relies on diagnostic strategies, including sequential evaluation of clinical probability, measurement of plasma D-dimer levels and most often CT pulmonary angiogram (CTPA), rather than a standalone test.

The initial and essential step is the assessment of pre-test clinical probability, either by gestalt or by validated clinical prediction rules (Wells rule of Geneva score). This initial evaluation separates patients into different groups of PE prevalence, and thus directly influences the negative and positive predictive values of the diagnostic tests used in these patients.

D-dimer measurement Plasma D-dimer measurement has been extensively evaluated for the exclusion of PE in outpatients. The diagnostic usefulness of D-dimers lies in their high sensitivity and hence their capacity to exclude PE when below a certain cut-off ("negative D-dimers") without further investigations. Indeed, in patients with "non-high" clinical probability (low and intermediate groups in a three category score or unlikely group in a dichotomic score), a highly sensitive negative D-dimer test safely excludes PE without additional investigations. Sensitive D-dimer tests include those performed by the ELISA technique (median sensitivity 99%; Vidas®, Stratus®, AxSYM®) and by quantitative latex methods (median sensitivity 96%; STA Liatest®, Tinaquant®). In patients with high clinical probability or likely PE, the negative predictive value of even a highly sensitive D-dimer test will be insufficient to exclude PE. D-dimer measurement is thus not used in these patients.

Performance of D-dimers in elderly patients The specificity of the ELISA and quantitative latex D-dimer tests for venous thromboembolism (VTE) is limited, ranging from 35 to 40%. Indeed, D-dimers increase in various clinical situations, including cancer, infectious or inflammatory states, pregnancy or during post-operative periods, but also with age, leading to reduced specificity of the test in elderly patients. In other words, the probability of having a negative test result is reduced, so that the number of patients needed to test (NNT) to exclude one PE without further investigations is higher. Indeed, whereas PE can be ruled out in presence of non-high clinical probability and negative D-dimers in one out of 3 outpatients presenting to the emergency room with suspected PE, it can be excluded in only one out of 20 patients > 80 years. As current diagnostic strategies for PE include imaging (most often CTPA) in patients with positive D-dimers, lack of specificity of D-dimers in the elderly leads to a high proportion of these patients undergoing CTPA.

Age-adjusted D-dimer cut-off The question of a higher D-dimer cut-off in elderly patients was raised a decade ago, but studies confirming the potential security of such a strategy by retrospectively applying age-adjusted cut-offs to large prospective cohorts of consecutive patients with suspected VTE were published between 2010 and 2012 and confirmed the safety of using age-adjusted cut-offs on an overall population of several thousands of patients.

A progressive age-adjusted D-dimer cut-off (agex10 microg/L in patients > 50 years) was retrospectively derived and validated in a sample of 1712 patients with suspected PE. The retrospective validation study showed that the age-adjusted cut-off could increase by around 20% the number of patients in whom the D-dimer test was considered negative, without increasing the proportion of false-negative results when compared to the standard cut-off (<500 microg/L). The increase in the diagnostic yield of D-dimers was particularly pronounced in patients over 80 years, as the age-adjusted cut-off allowed an increase in the proportion of "negative" D-dimers from 9% to 21%, without any false negative result.

This progressive age-adjusted D-dimer cut-off has been prospectively validated in the ADJUST-PE study, a large multicentre multinational (19 centres in Belgium, France, the Netherlands, and Switzerland) management outcome study published at the beginning of this year. All consecutive patients who presented to the emergency department with clinically suspected PE were assessed by a sequential diagnostic work-up using clinical probability assessment (by one of the two following scores: simplified, revised Geneva score or the two-level Wells score for PE), highly sensitive D-dimer measurement (ELISA or immuno-turbidimetric assays), and CTPA. Patients with a D-dimer level below their age-adjusted cut-off did not undergo further investigations and were thus left without anticoagulant treatment and followed-up for a period of 3 months.

The ADJUST-PE study included a total of 3346 patients with suspected PE, from whom 22 were excluded (D-dimers not performed in 21 and 1 consent withdrawal), leaving 3324 patients included for the analysis. The subgroup of particular interest was of course patients having D-dimer levels between 500 microg/L and their age-adjusted cut-off (n=337). None of these patients was lost to follow-up and 6 patients received therapeutic anticoagulation for another indication than VTE. Of the remaining 331, 7 died and 7 had suspected VTE. Only one of these 14 events was adjudicated as confirmed VTE (nonfatal PE). The so-called "failure-rate" of the age-adjusted cut-off was thus very low at 1 / 331 (0.3%; 95% CI 0.1%-1.7%), so that this strategy can be considered as very safe.

In the ADJUSt-PE study, the overall number of patients ≥ 75 years was 766, of whom 673 had non-high clinical probability (87.9%). Among these 673 patients, the proportion of patients with D-dimer levels <500 microg/L was 43/673 (6.4%; 95% CI 4.8%-8.5%). Additional 157/673 patients (23.3%) had D-dimer levels below their age-adjusted cut-off. This means that among the patients ≥ 75 years with suspected PE and non-high clinical probability, 200/673 (29.7%; 95% CI 26.4%-33.3%) had a "negative" D-dimer result when age-adjusted cut-off was used. After excluding 5 of these patients who received anticoagulation for another indication, none of the remaining 195 patients experienced confirmed VTE during the 3 month follow-up (0.0%; 95% CI 0.0%-1.9%). The diagnostic yield of D-dimers was thus particularly increased by using an age-adjusted cut-off in patients ≥ 75 years, as the number of patients needed to test to exclude one PE diminished from 16 with conventional cut-off to 3.4 with age-adjusted cut-off, a figure that corresponds to the NNT in the general population in previous studies.

Increasing the proportion of patients in whom PE can be ruled out based on clinical probability assessment and D-dimer measurement without further testing is particularly interesting in older patients. Indeed, the higher prevalence of renal failure in this population increases the potential risk of contrast-induced nephropathy related to CTPA or even contra-indicates this test, and ventilation-perfusion lung scan (which can be performed in patients with severe renal failure) provides a high proportion of inconclusive results in older patients. Moreover, ruling out PE based on clinical probability and a simple blood test could contribute to reduce the time spent in the emergency department and the costs related to PE diagnostic work-up. Indeed, a previous study had shown that D-dimer measurement with conventional cut-off was highly cost-saving in patients less than 80 years, but not in patients over 80 years. Using an age-adjusted D-dimer cut-off dramatically increases the proportion of patients in whom PE can be ruled out and is thus highly likely to reduce the costs of PE diagnosis in the emergency department.