Risk Factors for Central Venous Line-Related Deep Vein Thrombosis and Occlusions in the Israeli Pediatric Oncology Registry.

The use of central venous lines (CVLs) has greatly improved the quality of care in children with cancer, yet these catheters may cause serious mechanical, infectious and thrombotic complications, both deep vein thrombosis (DVT) and catheter occlusion. The aim of this prospective study is to ascertain the incidence of thrombotic complications and their risk factors. A registry was started in June 2006 for all children undergoing CVL insertion for treatment of cancer in the three largest pediatric cancer centers in Israel. After informed consent was signed, a registration form, that included questions regarding demographic-, clinical- and CVL-related data, and family history of thrombosis, was completed. Blood samples for baseline thrombophilia work-up, i.e. protein C, protein S, anti-thrombin, APCR, Factor V Leiden, Prothrombin gene mutation and MTHFR, were collected with separate consent. The following events were reported to the registry: immediate post insertion complications, venous thrombosis confirmed by imaging, occlusion of the CVL, i.e. inability to infuse and/or withdraw blood, requiring medical or surgical intervention, and CVL infections. The maintenance of CVLs and management of CVL occlusion and infection remained in accordance with institutional protocols. Responsible oncologists decided whether a dysfunctional or an infected CVL was to be removed or replaced, and whether radiographic evaluation for thrombotic complications was indicated. Patients were enrolled until December 2007, and data analysis was completed in June 2008. A total of 414 CVLs, i.e. peripherally inserted central catheters (PICCs) (45%), Hickman catheters (25%) and Port-a-Caths (30%), were inserted into 262 children for a total of 71,241 catheter-days. Fourteen events of venous thrombosis occurred in 13 children (4.9%, 95% confidence interval (CI) 2.6% to 8.3%), including 10 events of CVL-related DVT. The occurrence of CVL-related DVT was significantly higher for PICCs, 4.5%, compared to other types of CVLs, 0.9% (p=0.02, odds ratio (OR) 5.4 (95% CI 1.13 to 25.8)). CVL-related DVT was not associated with age at diagnosis, side of insertion (right vs. left), vessel cannulated, type of cancer (acute lymphoblastic leukemia vs. others), ethnic origin or family history of thrombosis. Occlusion of the CVL occurred at least once in 90 children (34%, 95% CI 29% to 40%). Children with family history of thrombosis were more likely to have CVL occlusion, 62.5%, compared to children without family history of thrombosis, 30.4% (P=0.01, OR 3.8 (95% CI 1.3 to 10.8)). Occlusion was reported in 102 CVLs (24%, 95% CI 20% to 28%). The occurrence of occlusion was higher for Port-a-Caths, 42%, and Hickman catheters, 35%, compared to PICCs, 23% (P<0.01, OR 6.64 (95% CI 2.98 to 14.8) and 4.62 (95% CI 1.84 to 11.6), respectively). CVL-related DVT was not associated with occlusion. Until now, thrombophilia screening has been completed in 85 children (32%), 21 of whom had a positive screen (25%, 95% CI 16% to 35%). A positive thrombophilia screen was found more frequently in children of Arabic origin, 43%, compared to children of Jewish origin, 13% (P=0.006), but was not associated with CVL-related DVT or occlusion. Also, in a subgroup analysis of the children with thrombophilia testing (n=85), children with a family history of thrombosis were more likely to have occlusion compared to children without a family history of thrombosis, 100% vs. 37%, respectively (P=0.01, Bonferroni post-oc correction). Our prospective study shows that insertion of PICCs significantly increases the risk for symptomatic CVL-related thrombosis; other risk factors were not found to be significant. The lower rate of PICC occlusions might be explained by their use for shorter time periods. Interestingly, a positive family history of thrombosis rather than a positive thrombophilia screen was found to be a risk factor for CVL occlusion; perhaps the standard thrombophilia screening is not sensitive enough to detect inherited risk of thrombosis associated with CVLs. The long-term effect of CVL occlusion as a predictor for under-diagnosed CVL-related thrombosis will be determined by following our cohort for development of post-thrombotic syndrome.