Upper Extremity DVT versus Lower Extremity DVT: Perspectives from the GARFIELD-VTE Registry

Abstract Upper extremity deep vein thrombosis (UEDVT) is less common than lower extremity DVT (LEDVT) and consequently less well characterized. This study compared clinical characteristics and 1-year outcomes between 438 UEDVT patients and 7,602 LEDVT patients recruited in the GARFIELD-VTE registry. UEDVT patients were significantly more likely to have a central venous catheter than those with LEDVT (11.5% vs. 0.5%; p < 0.0001), and had a higher rate of active cancer (16.2%) or recent hospitalization (19.4%) compared with LEDVT patients (8.7% and 11.2%, respectively). Nearly all patients with UEDVT and LEDVT were initiated on anticoagulant therapy, which was a direct oral anticoagulant in one-third individuals in both groups. At 3, 6, and 12 months, the proportion of UEDVT and LEDVT patients who were receiving anticoagulant therapy was 82.6 and 87.4%, 66.0 and 72.6%, and 45.7 and 54.6%, respectively. In the UEDVT and LEDVT groups, VTE recurrence rate was 4.0 (95% confidence interval [CI], 2.4–6.7) and 5.5 (95% CI, 4.9–6.1) per 100 person-years, respectively; major bleed was noted in 1.3 (95% CI, 0.6–3.2) and 1.6 (95% CI, 1.3–1.9) per 100 person-years and all-cause mortality in 9.7 (95% CI, 7.1–13.4) and 6.7 (95% CI, 6.1–7.3) per 100 person-years, respectively. Hence, risk of recurrence was similar in the two groups whereas all-cause mortality was significantly higher in the UEDVT group than the LEDVT group (p = 0.0338). This latter finding was likely due to the high prevalence of cancer in the UEDVT group.


Abstract
Upper extremity deep vein thrombosis (UEDVT) is less common than lower extremity DVT (LEDVT) and consequently less well characterized. This study compared clinical characteristics and 1-year outcomes between 438 UEDVT patients and 7,602 LEDVT patients recruited in the GARFIELD-VTE registry. UEDVT patients were significantly more likely to have a central venous catheter than those with LEDVT (11.5% vs. 0.5%; p < 0.0001), and had a higher rate of active cancer (16.2%) or recent hospitalization (19.4%) compared with LEDVT patients (8.7% and 11.2%, respectively). Nearly all patients with UEDVT and LEDVT were initiated on anticoagulant therapy, which was a direct oral anticoagulant in one-third individuals in both groups. At 3, 6, and 12 months, the proportion of UEDVT and LEDVT patients who were receiving anticoagulant therapy was 82.6 and 87.4%, 66.0 and 72.6%, and 45.7 and 54.6%, respectively. In the UEDVT and LEDVT groups, VTE recurrence rate was 4.0 (95% confidence interval [CI], 2.4-6.7) and 5.5 (95% CI, 4.9-6.1) per 100 person-years, respectively; major bleed was noted in 1.3 (95% CI, 0.6-3.2) and 1.6 (95% CI, 1.3-1.9) per 100 person-years and all-cause mortality in 9.7 (95% CI, 7.1-13.4) and 6.7 (95% CI, 6.1-7.3) per 100 person-years, respectively. Hence, risk of recurrence was similar in the two groups whereas all-cause mortality was significantly higher in the UEDVT group than the LEDVT group (p ¼ 0.0338). This latter finding was likely due to the high prevalence of cancer in the UEDVT group.

Introduction
Upper extremity deep vein thrombosis (UEDVT), arising in the brachial, axillary, or subclavian veins, accounts for up to one-tenth of the total burden of DVT. 1-10 UEDVT may be unprovoked or caused by a number of provoking factors, most commonly central venous catheter (CVC) placement, 6,[11][12][13][14][15][16][17] cancer, 5,6,13 surgery, 2,8,18 limb immobilization due to plaster cast, 2,8 heritable and acquired thrombophilias, 2 or the thoracic outlet syndrome (TOS)-that is, compression and occlusion of the subclavian vein between the clavicle and first rib. 2, 19,20 Most of the information on UEDVT derives from small case series that have produced somewhat inconsistent findings. 2 The risk of pulmonary embolism (PE) in patients with UEDVT is lower than that with lower extremity DVT (LEDVT), [21][22][23][24] as is the potential for recurrent venous thromboembolism (VTE). 7,21 Mortality rates in patients with UEDVT range from 15 to 50%, 2 likely depending on the prevalence of cancer in the different populations. 4, 25 The risk of developing postthrombotic syndrome was reported variously from 7 to 44%. 2 Because of the low prevalence and limited knowledge of UEDVT, current management recommendations for this entity parallel those for LEDVT. [26][27][28] To gain further insights into UEDVT, the present study investigated and compared baseline characteristics as well as long-term outcomes of treatment between patients with UEDVT and those with LEDVT enrolled in the prospective GARFIELD-VTE registry. The GARFIELD-VTE registry (Clini-calTrials.gov identifier: NCT02155491) is an ongoing, prospective, observational study of 10,685 patients with objectively diagnosed VTE, from 415 sites in 28 countries. 29

Study Design and Participants
The study design has been reported. 29 Patients were recruited from centers representative of the various care settings for each country. They were enrolled consecutively, and unselected. No specific treatments, tests, or procedures were mandated by the study protocol. Decisions to initiate, continue, or change treatment were solely at the discretion of treating physicians.
The registry records treatment patterns for acute VTE episodes (including UEDVT and LEDVT) as well as the rate and nature of VTE recurrence, bleeding complications, and allcause mortality over 36 months of follow-up. Patients were recruited across a range of clinical settings, including vascular medicine, general practice, and internal medicine. Patients aged !18 years with a confirmed diagnosis of VTE within 30 days of entry into the registry were eligible for inclusion. Excluded were patients with superficial vein thrombosis, those who had not completed treatment for a previous VTE, patients participating in another interventional study, and those in whom long-term follow-up was not feasible.
Data were extracted on patients with a confirmed diagnosis of DVT regardless of whether they had a concurrent episode of PE; those with nonlimb DVT (e.g., vena cava DVT) or missing site of DVT were excluded. Eligible subjects were dichotomized into UEDVT and LEDVT groups and compared in terms of treatment strategies immediately and 1, 3, 6, and 12 months postdiagnosis, and their outcomes at 12 months.

Data Collection
Data were captured by electronic case report form (eCRF), submitted to the registry-coordinating center (eClinicalHealth Services, Stirling, United Kingdom) via secure Web sites, and analyzed by the Thrombosis Research Institute, London, United Kingdom. The completeness and accuracy of data collected from medical records are checked by the registry-coordinating center and the source data systematically verified in 5% of all cases. Data on outcomes relevant to the registry are collected through review of clinical records and patient notes. These include: patient demographics, medical history, provoking VTE risk factors (within the previous 3 months), symptoms, nature of VTE (extent and location), and date and method of diagnosis. Routinely performed tests are documented (including hemoglobin, platelet count, international normalized ratio, and creatinine). All patients are followed prospectively for a minimum of 36 months.

Outcomes
Outcomes were recorded in standardized eCRFs. These included recurrent VTE, all-cause mortality, bleeds (any or major as defined by the International Society on Thrombosis and Haemostasis criteria 30 ), cancer (diagnosed at least 30 days after the index VTE), myocardial infarction/acute coronary syndrome, and stroke/transient ischemic attack. This study did not capture event rates according to treatment status. Outcomes were not centrally adjudicated.

Ethics
The registry is conducted in accordance with the Declaration of Helsinki and guidelines from the International Conference on Harmonisation on Good Clinical and Pharmacoepidemiological Practice, and adheres to all applicable national laws and regulations. Independent ethics committee for each participating country and hospital-based institutional review boards approved the registry design. All patients provided written informed consent to participate.

Statistical Analyses
Patients were categorized into those with UEDVT or LEDVT and their clinical characteristics summarized. Continuous variables are presented as means (standard deviation [SD]), medians (interquartile range), and ranges (min-max). Categorical factors are reported as frequency count (percentage). Intergroup differences in the presence of provoking factors were assessed by Fisher's exact test. Event rates (per 100 person-years) and associated standard errors were estimated using a log-linked generalized linear model (Poisson regression). Large-sample 95% confidence intervals (95% CIs) were obtained by inverting the link function. In this analysis, mortality was posed as a competing risk for all other clinical outcomes. Thereby, hazard ratio for death was estimated by Cox model, and subhazard ratios (sHRs) for other outcomes were calculated by Fine-Gray model. 31 Model coefficients were obtained applying maximum likelihood method and assessed by Wald chi-squared test. In all cases, a p-value of < 0.05 was considered statistically significant. The data presented were extracted from the registry in May 2018. All statistical analyses were performed with SAS Enterprise Guide version 7.1 (SAS Institute Inc., Cary, North Carolina, United States).

Demographics and Pathways of Care
Of the total 10,685 eligible patients, 438 (4.1%) were classified as having UEDVT and 7,602 (71.1%) LEDVT; among these individuals, the proportion complicated by PE was 12.8 and 20.1%, respectively (see group assignation decision tree in ►Fig. 1). In patients with UEDVT, thrombus was detected fairly evenly in left and right limbs, whereas in those with LEDVT there was a slightly higher distribution of left-versus right-sided DVT. In UEDVT and LEDVT groups, percent male was 55.5 and 50.9%, respectively; mean (AESD) age was 55.0 AE 17.6 and 58.1 AE 16.8 years and body mass index 27.1 AE 6.5 and 28.3 AE 6.4 kg/m 2 , respectively (►Table 1).
DVT was diagnosed using compression ultrasonography (CUS) in nearly all cases (UEDVT, 89.7%; LEDVT, 96.1%) with a positive D-dimer assay obtained in 17.8% of UEDVT and 26.2% of LEDVT individuals. Computed tomography (CT) venograms were more frequently used for the diagnosis of UEDVT than for LEDVT (13.5 and 4.4%). Contrast venography was rarely used (1.8 and 1.3%). PE, where present, was diagnosed using CT pulmonary angiography in 92.5% of patients.

Treatment
Approximately two-thirds patients with UEDVT and LEDVT were treated as inpatients (overall, 69.9%). No intergroup difference was observed in this context.
At 3, 6, and 12 months, the proportion of patients in UEDVT and LEDVT groups who were receiving anticoagulant therapy was 82.6 and 87.4%, 66.0 and 72.6%, and 45.7 and 54.6%, respectively (►Fig. 2). Among those taking anticoagulant therapy, approximately half of the patients in the UEDVT and LEDVT groups were using DOAC at 3, 6, and 12 months postdiagnosis, whereas among the remainder there was a greater tendency for UEDVT patients to remain on parenteral therapy while LEDVT patients were switched to VKA at these later time points (►Fig. 2).

Discussion
This prospective, observational study investigated clinical characteristics, treatment patterns, and 1-year outcomes in a large cohort of patients with UEDVT and LEDVT using data captured by the global GARFIELD-VTE registry. 29 Nearly all cases of UEDVT and LEDVT were confirmed using CUS, which is convenient 32 and widely regarded as first-line diagnostic imaging modality in both conditions. 20,33,34 In patients with UEDVT, thrombus was distributed evenly in left and right limbs, whereas in those with LEDVT there was a slightly higher left-to-right ratio, which may be attributable to anatomy of the left iliac vein. 35,36 UEDVT was less often associated with PE compared with LEDVT, consistent with previous reports. 4, [21][22][23][24] One-third patients with UEDVT and LEDVT had recent provoking factors, with a somewhat higher proportion of UEDVT individuals recently affected by cancer or hospitalization than those with LEDVT. Overall prescription rates of anticoagulants were similar between the two groups, although UEDVT patients were more likely to receive parenteral therapy alone than LEDVT patients. This observation might be explained by the higher prevalence of cancer in the UEDVT group: guidelines recommend low-molecular weight   heparin (LMWH) or VKA and not DOAC for cancer-associated thrombosis. 28,37 Annualized recurrence rates were similar between patients with UEDVT and LEDVT, whereas mortality during the first 1 year after diagnosis was significantly higher for UEDVT than LEDVT patients-possibly related to their greater burden of cancer. No difference was observed concerning other long-term clinical outcomes such as bleeding risk and arterial thrombotic events. Cancer is a well-known risk factor for UEDVT-although the enhanced risk may be due to widespread use of a CVC, which is often implanted in the subclavian vein, in cancer patients undergoing treatment rather than cancer itself. 6,19,38 Indeed, in this study patients with UEDVT had significantly higher rates of CVC insertion than those with LEDVT. Cancer also confers a high risk of recurrence and cancer patients who experience UEDVT seem to have worse survival than those who have not had an episode of UEDVT. 7,39 The present cohort of UEDVT patients had a higher incidence rate of all-cause mortality than those with LEDVT, although whether this outcome was specifically due to cancer is unknown.
Among UEDVT patients the relative risk of all-cause mortality was much higher in those with than without cancer, suggesting they were cancer-related deaths, whereas the risk of recurrence and bleeding events was slightly higher in cancer patients experiencing UEDVT. The risk of all adverse events was very low in UEDVT patients without cancer.
In this study, the prevalence of UEDVT (4.1%) and various associated risk factors including cancer is comparable to that observed in other registries. Among 37,366 VTE patients enrolled in the Spanish Registro Informatizado de Enfermedad Trombo Embólica, 9 6% had UEDVT and 94% had LEDVT. Active cancer was the most important risk factor accounting for half of UEDVT cases. The population-based Malmö Thrombophilia Study 5 performed in 1,203 VTE patients similarly reported that UEDVT accounted for 5% of the overall sample; over 5 years' follow-up recurrence rate was 5% and death occurred in 24%, with cancer patients accounting for half (47%) of the mortalities. A similar picture was observed in two single-center experiences, one of 200 consecutively encountered patients with UEDVT in whom 36% had documented malignancy and 76% indwelling lines, 22 and another of 94 UEDVT patients among whom 48% had a diagnosis of cancer and 93% CVCs. 15 Several research efforts have looked at ways to optimize treatment for UEDVT. Rathbun et al 40 treated 67 UEDVT patients with either LMWH followed by warfarin or LMWH alone for 3 months and observed no recurrences and a low rate of major bleeding with both regimens. Montiel et al 8 using data from the Swedish national anticoagulation registry (AuriculA) evaluated 55 UEDVT patients who received DOAC and observed one recurrence during treatment (2%) and two recurrences (4%) after cessation of treatment over a 6-month follow-up period. In the present study, nearly all patients with UEDVT and LEDVT were initiated on anticoagulant therapy (either parenteral therapy with or without VKA, or DOAC) for at least 3 months. Although duration of treatment was similar for both groups, the proportion of patients who discontinued anticoagulants during follow-up was higher for UEDVT than LEDVT. No statistically significant between-group difference was observed for recurrence or bleeding, suggesting that treatment strategies should be equivalent for UEDVT and LEDVT, as recommended by international guidelines. 28 This study has a number of limitations as well as strengths. The former include its observational design and lack of control groups. For example, outcomes in patients on different treatments (treatment modality and duration) or who discontinued treatment are unknown. Moreover, outcomes were not centrally adjudicated. On the other hand, this was a large, prospective study, well powered to provide robust insights. Data were recorded on customized eCRFs intended to collect useful information, although with many personnel in different locations registering events there is some potential for inaccuracy. A future possible investigation might look at provoking factors in patients stratified by age. It seems unlikely that younger patients presenting with UEDVT are at any considerable risk of malignancy, hospitalization, or death, or that elderly patients have TOS; strong evidence supporting these contentions may avoid them undergoing unnecessary and stressful tests.
In conclusion, the GARFIELD-VTE registry provides extensive data showing how patients with UEDVT and LEDVT are being treated around the world. The present general overview of GARFIELD-VTE findings provides a descriptive narrative of this clinical setting, confirming that patients with UEDVT and LEDVT generally receive treatment according to international practice guidelines. 28 However, UEDVT was associated with a higher mortality rate than LEDVT. It is not known whether the excess of deaths was due to cancer or rather that UEDVT, like atrial fibrillation, may be a biologic marker for systemic vascular disease.
What is known about this topic?
• Upper extremity deep vein thrombosis (UEDVT) is less common than lower extremity deep vein thrombosis (LEDVT). • Current guidelines recommend that UEDVT be managed similarly to LEDVT.
What does this paper add?
• In this large, global, observational cohort study patterns of diagnosis and treatment of UEDVT and LEDVT were compared. • UEDVT was more often associated with central venous catheters (CVCs), cancer, or recent hospitalization compared with LEDVT. • Most patients with UEDVT and LEDVT were treated with anticoagulants, in half cases for up to 1 year. • Risk of recurrence was similar with UEDVT and LEDVT, whereas all-cause mortality was significantly higher with UEDVT (p ¼ 0.0338).

Note
The GARFIELD-VTE Registry is an independent academic research initiative sponsored by the Thrombosis Research Institute (London, United Kingdom) and supported by an unrestricted research grant from Bayer Pharma AG (Berlin, Germany).