The negative predictive value of D‐dimer on the risk of recurrent venous thromboembolism in patients with multiple previous events: A prospective cohort study (the PROLONG PLUS study)

The optimal duration of anticoagulation after recurrent venous thromboembolism (VTE) is poorly established [1, 2]. Recent studies suggested that D-dimer may identify patients at low risk of recurrence after a first VTE [3, 4]. In a pilot, prospective, cohort study we aimed to assess the negative predictive value of D-dimer in patients with recurrent VTE. Patients with negative D-dimer while on treatment stopped anticoagulation and underwent repeated testing after 7, 15, and 30 days; treatment was resumed if D-dimer turned positive and permanently stopped if it remained negative. The study was interrupted after the enrolment of 75 patients. At that time, treating physicians decided treatment resumption in 12.2% of the patients, but the majority of events were distal or superficial vein thromboses. The rate of objectively documented recurrent proximal deep vein thrombosis (DVT) and/or pulmonary embolism (PE) was 2.56% (95% CI 0.13, 15.07%) in the 39 patients with persistently negative D-dimer at 30 days, for an annual incidence of VTE of 5.65 events/100 patient/years. These preliminary findings suggest that negative D-dimer may identify patients with history of previous VTE at low risk of recurrences, but this approach should be tested in larger trials in highly selected patients. Clinical guidelines recommend long-term anticoagulant treatment for patients with a second episode of unprovoked VTE [2]. There is currently no guidance on the duration of treatment for patients with recurrent venous thrombosis in whom one event is unprovoked and the other event is secondary to a transient risk factor, as well as for patients in whom one event is major (e.g., pulmonary embolism and/or proximal deep vein thrombosis) and the other event is minor (e.g., distal deep vein thrombosis or thrombosis of the greater saphenous vein). This uncertainty likely leads to a heterogeneous approach to the clinical management of these patients. The PROLONG study has previously evaluated the predictive role of D-dimer measurement after withholding VKA treatment in patients with a first episode of unprovoked VTE [5]. Patients with a positive D-dimer 1 month after the discontinuation of anticoagulation who were not restarted on VKA treatment had a significantly higher incidence of VTE recurrences than patients with a negative D-dimer (10.9 events per 100 patient/years and 4.4 events per 100 patient/years, respectively). The results of the PROLONG and of other cohort studies suggest that it could be safe to withhold treatment in patients with a first episode of unprovoked VTE and with a negative D-dimer. In a pilot, prospective, multicenter, cohort study we aimed to assess the negative predictive value of D-dimer in patients with recurrent VTE for whom the optimal duration of VKA treatment was considered uncertain by treating physicians. A total of 75 patients with negative D-dimer during VKA treatment were enrolled. Baseline characteristics of the study population are summarized in Table I. D-dimer levels turned positive in 11 patients at Day 7, 13 patients at Day 15, and 11 patients at Day 30. Thus, a total of 35 patients (47.3%) had a positive D-dimer within 1 month from the discontinuation of anticoagulant therapy. Of these patients, 29 actually resumed VKA treatment according to the study protocol, whereas 6 refused to restart on treatment. One patient was lost to follow up after Day 7 visit (1.3%). Of the 39 patients (52.7%) with persistently negative D-dimer levels at 30 days, 37 maintained the treatment stopped according to the study protocol and two autonomously decided to resume treatment. Female gender and the presence of residual venous obstruction were significantly associated with the positivity of D-dimer levels at Day 30 (Table I). Patients who permanently discontinued anticoagulant treatment because of a negative D-dimer at Day 30 continued to monitor their D-dimer levels (n: 37): 6 of them had a positive D-dimer level at Day 90, and two had positive a D-dimer level at Day 150. Median follow-up time was 365 days. The cumulative incidence of the primary end-point of the study based on the intention to treat analysis (objectively documented symptomatic proximal deep vein thrombosis or pulmonary embolism in the group of patients with persistently negative D-dimer levels at Day 30) was 2.56% (95% CI 0.13, 15.07%), with an annual incidence of 5.65 events/100 patient/years. The overall cumulative incidence of recurrent venous thrombosis (there including major and minor events) leading to resumption of anticoagulation in patients with persistently negative D-dimer levels was 12.82%, 95% CI 4.82, 28.23, with an annual incidence of 13.76 events/100 patient/years. All symptomatic recurrent events are described in Table II. Briefly, three recurrent events occurred within the first 30 days (4.05%, 95 CI 1.05, 12.18): two proximal DVT and one superficial vein thrombosis. Two patients were males, mean age was 67.6 years. D-dimer positivity was documented in all three patients: one patient had a positive D-dimer at Day 15, refused to restart VKA therapy and had a recurrent event (superficial vein thrombosis) diagnosed 1 week later; two patients became symptomatic (one on Day 7 and one on Day 15) and recurrent proximal DVT was objectively documented together with D-dimer positivity. Six recurrent venous thromboembolic events occurred after Day 30: one PE, two distal DVT, and three superficial vein thromboses. Four patients were males, mean age was 64.5 years. Of these six events, one occurred in a patient with positive D-dimer at Day 7 who decided to not resume anticoagulation, five occurred in patients with persistently negative D-dimer levels at Day 30. In patients with positive D-dimer levels, the cumulative incidence of recurrent major events after Day 30 was 0% (95% CI 0.00, 13.34%), and the overall cumulative incidence of recurrent venous thrombosis was 3.13% (95% CI 0.16, 18.01), with an annual incidence of 3.15 events/100 patient/years. Risk factors for recurrence at univariate analysis are summarized in Table III. Because none of the risk factors was significantly associated with recurrent VTE, multivariate analysis was not feasible. No major bleeding events and no deaths were documented during the study. The results of this pilot study show that about half of the patients with an initially negative D-dimer during VKA treatment had persistently negative D-dimer levels during subsequent testing within the first 30 days; that 4.0% of the whole study population had a recurrent event in the first month of the study; and that the annual incidence of recurrent, major VTE in patients who permanently stopped treatment was 5.65 events/100 patient/years. This finding suggests that the proposed approach may be potentially valid to identify low risk patients. The decision to stop recruitment before the initially planned sample of 100 patients has been mostly driven by the occurrence of minor clinical events which prompted treating physicians to immediately restart patients on long term anticoagulant treatment, but the clinical significance of most of these events remains uncertain. A systematic review of the literature assessing the role of D-dimer in predicting recurrent events in patients with a first episode of VTE found a negative D-dimer result measured between 3 weeks and 2 months after VKA discontinuation was associated with a 3.5% annual risk for recurrence [6]. In comparison with all previous studies on D-dimer in patients with a single VTE event [3-5, 7], we observed a lower rate of patients with normal D-dimer at 30 days, a higher rate of recurrences in the first month of the study, and a slightly higher annual rate of VTE recurrences when comparing the groups of patients with negative D-dimer at 30 days. Only few studies provided data on the risk of recurrence in patients with multiple episodes of VTE [8-11]. In a prospective cohort study, Hansson et al. found a 1.7 (95% CI 1.16,2.52) relative risk of recurrence in patients with more than one episode of DVT in comparison to patients with a first DVT after multivariate analysis [9]. In a randomized controlled study comparing 6 months of oral anticoagulant therapy with indefinite duration of treatment in patients after a second episode of VTE, the incidence of recurrences was reduced from 20.7 to 2.6% after 4 years of follow-up [10]. It remains quite difficult to understand which patients with two or more previous VTE should receive long-term, life-long anticoagulant therapy. In a recent statement of the International Society of Thrombosis and Hemostasis, Kearon et al. have suggested that a risk of recurrence appreciably higher than 5% at 1 year and 15% at 5 years should usually discourage stopping anticoagulant therapy [12]. The 5.7% annual incidence of recurrent VTE found in this study may suggest that an approach based on serial D-dimer measurements may identify patients who could safely stop treatment. This pilot study has a number of limitations to be taken into account. First, the small study sample is only hypothesis generating, and does not allow any valid conclusion from our observations. In particular, the study is underpowered to detect statistical associations in univariate or multivariate risk assessment models. Second, this pilot study did not have a central adjudication committee, and all events were locally adjudicated. However, all investigators are highly experienced in these studies and we used definitions of recurrent events that were already applied in previous studies as well as in daily clinical practice. Third, a qualitative method for D-dimer testing was used, and some variability in the interpretation of the test could have occurred. However, this approach was previously used and validated in the Prolong and Prolong II studies [5, 7], which found a good reproducibility of this test. Indeed, the use of quantitative D-dimer tests with a higher sensitivity could have improved the selection of a low-risk population. In conclusion, patients with persistently negative D-dimer who stop VKA treatment may have an acceptable annual rate of recurrences, but at the cost of a nonnegligible incidence of recurrent events during the first month after treatment discontinuation. Future larger studies may assess the safety of this approach based on D-dimer testing in more selected patients (e.g., with the exclusion of patients with a second unprovoked event) for whom indefinite anticoagulation may be unnecessary. Patients aged 18 years or older with at least two episodes of objectively documented venous thrombosis who were treated with VKA for at least 6–12 months (according to the pathogenesis of the most recent event) and who had a negative D-dimer value while on treatment were eligible for inclusion in this study. The following VTE events were acceptable for inclusion in the study: pulmonary embolism (PE), proximal deep vein thrombosis (DVT) of the lower limbs, isolated distal DVT of the lower limbs, and superficial vein thrombosis of the greater saphenous vein. At least one of the two or more events had to be major, defined as PE or proximal DVT. Patients were excluded in case of a previous hemodynamically unstable PE; if the last event was an isolated, unprovoked PE; if two or more VTE episodes were unprovoked; in the presence of permanent risk factors such as cancer or major thrombophilia defined as antiphospholipid antibodies syndrome, antithrombin deficiency, homozygous Factor V Leiden or prothrombin mutation, double heterozygosity; in the presence of any other indication for VKA treatment; in case of pregnancy or puerperium; in the presence of severe liver or kidney failure; in case of limited life expectancy or geographical inaccessibility. The protocol was approved by the institutional review boards of all participating centers and all patients provided written informed consent. Eligible patients underwent D-dimer measurement during VKA therapy. D-dimer testing was performed when the patient had an INR between 1.8 and 3.2. Patients with positive D-dimer were continued on treatment. Patients with a negative D-dimer were instructed to stop VKA treatment and their baseline variables were collected. Regardless of the site of their last event, at the time when treatment was withheld, all patients underwent compression ultrasonography of the proximal veins in both legs to assess the presence and to measure the diameter of residual vein thrombosis in the common femoral, superficial femoral, and popliteal veins. Vein diameters were measured during maximal compression and were considered recanalized if the diameter was less than 2.0 mm in a single determination or <3.0 mm in two consecutive determinations. D-dimer measurement was repeated after 7, 15, and 30 days and VKA treatment was resumed if D-dimer levels turned positive. If D-dimer levels remained negative at Day 30, anticoagulation was permanently stopped and all patients underwent a 1-year clinical follow-up. D-dimer levels were also measured at Days 90, 150, and 270 in all patients who maintained anticoagulation withheld, but their values were not used to decide on treatment strategies. Levels of D-dimer were assessed with the use of the Clearview Simplify D-dimer assay (Inverness Medical Professional Diagnostics), which was donated by Instrumentation Laboratory, Milan. During follow-up, patients were seen at each center at intervals of 3–6 months. Patients who resumed VKA treatment were regularly followed at the local anticoagulation clinic. In case recurrent venous thrombosis was suspected, the results of imaging tests were compared with those of the last available previous examination. A recurrent DVT was diagnosed if a previously fully compressible segment (contralateral or ipsilateral) was no longer compressible or if an increase of at least 4 mm in the diameter of the residual thrombus during compression was detected [13]. A recurrent superficial vein thrombosis was diagnosed in the presence of extension, defined as proximal progression of the initial thrombus by at least 2 cm and to within 3 cm or less from the sapheno-femoral junction, or in the presence of a new thrombus located in a different superficial vein and not directly contiguous upstream with the index thrombus, or located in the same superficial vein, but separated from the index thrombus by at least 10 cm [14]. In patients with suspected PE, the diagnosis of recurrence was based on the use of clinical probability, perfusion lung scanning or helical CT, and compression ultrasonography, D-dimer testing, or both if indicated [15]. Major bleeding events were defined as hemorrhages that were either retroperitoneal or intracranial or were associated with a decrease in hemoglobin of at least 2.0 g dL−1 or that required either transfusion of at least 2 U of blood or surgery or an invasive procedure to stop bleeding. All suspected outcome events and deaths were adjudicated locally. All investigators were asked to immediately report to the steering committee of the study (WA, BC, GP, VP) the occurrence of clinical end-points and to regularly update on the number of patients who required anticoagulation resumption. No prespecified stopping rules were defined. Primary end-point was the occurrence of objectively documented recurrent proximal DVT and/or PE at the 1-year follow-up in patients with persistently negative D-dimer levels. Secondary end-points were the occurrence of objectively documented proximal DVT, distal DVT, superficial venous thrombosis at the greater saphenous vein, or PE at the 1-year follow-up in patients with persistently negative D-dimer levels and in the whole study population of patients who initially stopped VKA treatment; the rate of patients with D-dimer levels turning positive after VKA discontinuation and the incidence of recurrent events in this population; the timing of D-dimer elevations; the occurrence of major bleeding in patients who resumed VKA treatment; and mortality. For this pilot study, a sample of convenience of 100 patients with initially negative D-dimer was chosen. After the enrolment of 75% of the planned sample, the steering committee of the study decided to stop the enrolment and to proceed with the analysis of the data because nine patients (12.2%) overall required the resumption of the treatment based on the decision of the treating physicians. Baseline characteristics are reported by means of descriptive statistics. Baseline differences between groups were assessed by the chi square (or Fisher exact test when appropriate) for categorical variables and t test for continuous variables. Hazard ratios and their 95% confidence intervals were calculated using the Cox proportional hazards models. Initially, an adjusted hazard ratio was calculated with age, sex, presence/absence of residual vein obstruction, the type of index event (superficial vein thrombosis or distal deep vein thrombosis with pulmonary embolism, or proximal deep vein thrombosis with or without pulmonary embolism vs. isolated distal isolated deep vein thrombosis or isolated superficial vein thrombosis), presence/absence of risk factors for the index VTE (idiopathic vs. secondary to a transient risk factor). Then we planned to construct a multivariate model introducing only variable with a P < 0.1 at the univariate model. Analysis was based on the principle of intention to treat. A P value of <0.05 was chosen as the cut off for statistical significance. The data were analyzed with the use o SPSS software, version 19.0 (SPSS). Walter Ageno*, Benilde Cosmi , Angelo Ghirarduzzi , Rita Santoro§, Eugenio Bucherini¶, Daniela Poli**, Domenico Prisco , Adriano Alatri , Vittorio Pengo§§, Luca Galli*, Francesco Dentali*, Gualtiero Palareti , * Dipartimento di Medicina Clinica, Università dell'Insubria, Varese, Italy, U.O. Angiologia e Malattie della Coagulazione Golinelli M Golinelli, Università di Bologna, Bologna, Italy, U.O. Angiologia, Ospedale di Reggio Emilia, Reggio Emilia, Italy, § U.O.C Emofilia, Emostasi e Trombosi, Azienda Ospedaliera di Catanzaro, Catanzaro, Italy, ¶ U.O. Semplice di Medicina Vascolare, Ospedale di Faenza, Ravenna, Italy, ** Dipartimento Cuore e Vasi, AOU Careggi, Firenze, Italy, Dipartimento Area Critica Medico-chirurgica, Università di Firenze, Firenze, Italy, Centro Emostasi e Trombosi, A.O. Istituti Ospitalieri di Cremona, Cremona, Italy, §§ U.O. Cardiologia, Università di Padova, Padova, Italy.