Assessment of D-dimer for ruling out peripherally inserted central catheter-associated upper extremity venous thrombosis: a diagnostic accuracy study

Background: With the extensive use of peripherally inserted central catheter (PICC), PICC-associated venous thrombosis (VT) has become one of the most important complications in the hospital. To reduce unnecessary Color Doppler Ultrasound (CDU) or imaging tests, D-dimer values are usually used to exclude VT for patients with low VT risk. There is little evidence for the usefulness of the D-dimer level as an independent diagnostic marker for excluding PICC-associated VT. Aims: To examine the effectiveness of D-dimer values to be used as an independent diagnostic marker for excluding PICC-associated Upper Extremity Venous Thrombosis(UEVT). Methods:Records were reviewed for 281 patients who underwent PICC insertion over the two years period in Xiangya Hospital Central South University.The patients were categorized into the UEVT unlikely group (<2 points) and the UEVT likely group ( ≥ 2 points) according to the Constans Clinical Decision Score( Constans score) post PICC insertion, before extubation. After the score was determined, the patients underwent a D-dimer test and CDU within 7 days after D-dimer test. Results: Among 281 patients, 180 patients (36%, 95%CI:30.6%-41.8%) had negative D-dimer results, 39 of 180 patients had VT despite having a negative D-dimer result, resulting in a failure rate of 21.7% (95% CI: 16.3–28.3%).The NPV of PICC-associated VT in the cancer group (80.0%,95% CI: 73.2–85.4%) was higher than that of the non-cancer group (60.0%,95% CI: 35.7–80.2%). The NPV of PICC-associated DVT(Deep Venous Thrombosis) (84.9%,95% CI: 78.7–89.6%) was lower than that of the PICC-associated SVT(Supercial Venous Thrombosis) (91.0% ,95% CI: 85.4–94.6%). Conclusion: The D-dimer levels maybe should not be used as a diagnostic index to rule out PICC-associated VT to avoid missed diagnosis of PICC-related VT, which may cause adverse consequences.


Introduction
In patients requiring infusion of corrosive drugs or long-term infusions, PICC is a convenient alternative to Central Venous Catheter (CVC). PICCs are easy to place, can be nurse-led, and do not have risks associated with CVC insertion [1]. However, PICCs have a risk of developing upper extremity super cial venous thrombosis (UESVT) and upper extremity deep venous thrombosis (UEDVT).UEDVT commonly affects inpatients and accounts for half of hospital-acquired DVT [2][3][4], which usually leads to treatment disruption, obvious scarring and occlusion of the deep veins of the upper extremity, impaired venous re ux and subsequent establishment of venous access, and increased hospital stay and cost [5,6] and may even cause pulmonary embolism [1,3]. Despite few studies on PICC-associated UESVT, the association between SVT and DVT is known to be variable [7].
The diagnostic criteria for thrombosis are recommended that patients determined to have a low risk for DVT, the negative D-dimer accurately ruled out UEDVT without the need for CDU or other imaging tests and helped avoid unnecessary anticoagulant treatment( Fig.1) [1,[9][10][11][12]. However, the algorithm fails when low scores are assigned to patients with VT or when tests give falsenormal D-dimer levels [13]. The latter may be the result of a small thrombus, reduced brinolysis, or D-dimer testing at a very early or late stage of the disease [14]. As we known,PICC-associated VT is different from the usual VT, as the former presents primarily as a mural thrombus [2][3][4]. This particular type of mural thrombus may affect D-dimer levels, and the question of whether they are also sensitive to rule out PICC-associated UEVT remains unclear. Thus, this study aimed to investigate whether the D-dimer concentration could also be used as an independent diagnostic marker for excluding PICC-associated UEVT.

Method Study Design
This was a retrospective case cohort study. The study included in inpatients between October 1, 2017, and October 1, 2019 on the all wards of Xiangya hospital Central South University in Hunan, China. The patients were categorized into the UEDVT unlikely group (<2 points) and the UEDVT likely group (≥2 points) according to the Constans score post PICC placement, before extubation.
After the score was determined, the patients underwent a D-dimer test and CDU within 7 days after D-dimer test.To examine the effectiveness of D-dimer concentration to be used as an independent diagnostic marker for excluding PICC-associated UEVT.
Inclusion criteria were as follows: 1) patients aged≥18 years; 2) patients with PICC inserted via the upper arms; 3) <2 points on the Constans score 4) patients who underwent CDU and D-dimer values after PICC placement. Exclusion criteria were as follows: 1) PICC-associated lower extremity venous thrombosis (LEVT); 2) a duration of >7 days between D-dimer examination and CDU;and 3)outpatient management.Subjects were excluded from the study for incomplete data. Data, including basic demographic characteristics, PICC, test results, disease course, and medications, were collected using the standard form in the Reasonable Safety Infusion Monitoring System of Xiangya Hospital of Central South University.
The study was approved by the ethics committee of Xiangya Hospital of Central South University (201907733) and was conducted according to the Helsinki Declaration of Ethical Principles for Medical Research involving Human Subjects.
The Constans score A dichotomized Constans score are recommended by the American Society of Hematology 2018 guidelines to assess clinical probability of suspected UEDVT. It was composed of speci c items that included venous material prensent(central venous catheter or pacemaker thread),localized pain,unilateral edema,other dignosis at least as plausible. The patients were categorized into the UEDVT unlikely group (<2 points) and the UEDVT likely group (≥ 2 points).

D-dimer
Blood samples for D-dimer testing were obtained before ultrasonography investigation. Blood was drawn by clean venipuncture from an antecubital vein with a 22-gauge butter y needle and collected into 3 ml plastic tubes containing 0.3 ml 0.106 M trisodium citrate. After blood collection, the blood was gently reversed and mixed for 3-6 times.Whole blood was centrifuged at 2000 ×g for 20 min at 20 °C. The STA Liatest® D-dimer was employed which is an automated and rapid microlatex D-dimer assay. The STA Liatest® had higher sensitivity (median, ≥95%) which meet the US Food and Drug Administration (FDA) requirements could applied to rule out VT [15,16]. Use of the age-adjusted cutoffs wouldn't signi cantly impairing the NPV [17]. But it seems to be less useful with the STA-Liatest® assay [18] and the results obtained with speci c D-dimer assays cannot be extrapolated from one test to another [18].So the results were expressed in mg/L, the cut-off value for DVT exclusion was 0.5mg/L.The blood laboratory of the hospital did not change the detection method or change the detection reagent between October 2017 and October 2019.
CDU CDU was performed within 7 days after D-dimer detection.CDU was carried out with the Philips epic5 instrument , with a highresolution broadband width 9-13 MHz linear array transducer by CDU technician who had adequate experience with CDU. The main criteria for the diagnosis of VT were as follows [19][20][21]: for probe after compression, the lumen cannot be compressed; the blood ow signal in the lumen is lled with defects; solid return can be seen in the lumen sound; disappearance or weakening of the spent response; phase change in the loss of the blood spectrum; and weakening or disappearance of the blood ow of the distal limb by squeezing. Deep veins of the upper limb included brachial vein, axillary vein, subclavian artery, internal jugular vein, and brachial vein [19,20]. Super cial veins of the upper limb included cephalic and cubital median veins and the basilic veins [2,21,22]. Patients with negative CDU did not undergo angiography.

Statistical analysis
The reliability and effectiveness of D-dimer level as an independent biomarker for PICC-associated VT was evaluated according to its sensitivity, speci city, NPV, and positive predictive value and the need for ultrasonography examinations. The patients were divided into the PICC-treated group, PICC-treated cancer group, and PICC-treated non-cancer group. The 95% con dence intervals (CI) were presented. All statistical analyses were performed using SPSS statistical software version 18.0.

Results
In total, 7454 patients underwent PICC placement during the study period. Of them, 3592 patients without CDU screening were excluded. Of the 3862 patients who underwent CDU screening, 769 developed VT and 3093 did not, yielding an incidence rate of 19.9%. After excluding 3212 patients with no D-dimer data within 7 days before CDU, 266 patients with LEVT, and 103 patients with Constans score ≥2 points, 281 patients were included in the nal analysis (Fig. 2).

Discussion
Diagnostic guidelines for DVT recommend using Clinical prediction rules and D-dimer levels to initially screen for the risk of DVT before using CDU [8]. In this study, we investigated whether the D-dimer values could be used as a key diagnostic marker for excluding PICC-associated VT. We evaluated patients with a PICC who had a Constans score of DVT <2 points (DVT unlikely) and D-dimer <0.5mg/L and found that 21.7% of these patients had PICC-associated VT,15.1% of DVT,9.0% of SVT,This is difference to previous studies showing that the D-dimer level can accurately rule out DVT as the NPV of the D-dimer level for DVT ranged from 99.3% to 99.8% [23][24][25].The possible reasons are as follows: First,the study population are different. We evaluated a speci c PICC population, whereas the majority of patients who undergo PICC placement in China are cancer chemotherapy patients. UEDVT in cancer patients are usually associated with or triggered by vascular access devices [3,[26][27][28].The prevalence of VT is increased (up to 20% of cancer patients develop VT) and the NPV is therefore reduced [29,30]. Not only that,Some hematological malignancies also are known to secrete proteolytic factors, and we speculate that some patients with malignancies and DVT have normal Ddimer levels because of accelerated degradation of D-dimer [31]. Overall, the complexity of various factors in cancer patients, ranging from complications to drugs used for the cancer itself, may affect the sensitivity of D-dimer as a biomarker [10]. In this study, non-cancer patients were mainly from the neurology and neurosurgery departments. The age of these patients, thrombotic burden and brinolytic activity, duration of symptoms, previous VT and in ammatory status, and use of anticoagulants may affect the accuracy of D-dimer levels [32][33][34]. This may partly explain why the NPV of D-dimer in the non-cancer group was lower than that in the cancer group (60.0% [95% CI: 35.7-80.2%] vs. 80.0% [95% CI: 73.2-85.4%]).
Second, the type of VT is different. Due to the difference in pathogenesis and clinical processes between UEDVT and LEDVT [35,36], we only evaluated UEVT directly or indirectly caused by PICC. Some studies have shown that approximately 75% of UEDVT cases is associated with indwelling vascular catheters [35,36]. This association is not surprising as catheter insertion leads to endothelial damage, occupies the vein lumen (promoting venous stasis), and is often required in patients with hypercoagulability because of intercurrent illness or malignancy. Thus, placement of these devices satis es the Virchow's triad, leading to increased risk of VT [37]. PICC-associated VT is unique, with most types being attached to wall thrombosis. A study found that D-dimer plasma levels correlated with the thrombotic burden in patients with LEDVT [38] and a normal D-dimer levels in UEVT can be ascribed to the low thrombotic burden associated with mural thrombus. Another study shows that D-dimer levels correlate with clot volume and surface area [39],the mural thrombus may be the result of a small thrombus causes tests give false-normal D-dimer levels [13,14]. This may be one of the important reasons for the lower NPV of D-dimer level for diagnosing PICC-associated VT in this study.
In this study, we discussed PICC-associated UEDVT and PICC-associated UESVT separately.PICC is placed in a much smaller vein than in CVC, and the risk of UEDVT in PICC is 2.5 times higher than that in CVC [3]. A case cohort study reported a 13-fold increased risk of thrombosis in patients receiving PICC [40]. Usually, PICC-associated UEDVT is clinically asymptomatic. In a recent randomized controlled clinical trial of PICC-associated UEDVT using CDU screening, it was found that up to 75% of patients with catheter had UEDVT, but only 4% of image-diagnosed patients with thrombosis developed clinical symptoms [41].Compared with UEDVT, UESVT is more prone to pain, reddening of the skin and swelling of the surrounding tissue and other clinical signs and symptoms because of its super cial location [42].Localized pain and unilateral edema are included in the Constans score [43] ,so Page 5/10 that the score of patients with UESVT is more than 2 points than that of patients with UEDVT.This may partly explain why the NPV of PICC-associated DVT was lower than that of the PICC-associated SVT.
This study also has some limitations. First, the single-center design may limit the generalizability of our ndings. Further, our cutoff time was patient extubation, and follow-up data were not analyzed. As a retrospective study, our data may underestimate the true number of cases of PICC-associated SVT and DVT because our institution does not have a systematic CDU screening protocol. In addition, A duration of >7 days between D-dimer examination and CDU was another limitation. Finally, Patients with negative ultrasound did not undergo angiography, we were not able to establish whether such VT was a diagnostic failure or a intervening complication.

Conclusion
The D-dimer level as an independent biomarker has low NPV for PICC-associated VT in low-risk patients identi ed according to the Constans score. The D-dimer levels maybe should not be used as a diagnostic index to rule out PICC-associated VT to avoid missed diagnosis of PICC-related VT, which may cause adverse consequences.