Development of a Risk Score to Predict Peripherally Inserted Central Catheter Thrombosis in Active Cancer

Abstract Background Peripherally inserted central catheters (PICCs) are commonly used in cancer patients. PICC-associated upper limb venous thromboembolism (PaULVTE, non-deep and deep involving veins proximal to the brachial vein) and distant VTE (i.e. pulmonary embolism [PE], lower limb deep vein thrombosis [DVT]) prevention is important due to the bleeding risk from subsequent anticoagulation. The Khorana risk score (KRS) is used to predict VTE in malignancies, has been externally validated and prospectively utilized in VTE prophylaxis studies in cancer patients. It utilizes routinely available variables: cancer type, prechemotherapy platelet count, hemoglobin, leukocyte count and body mass index (BMI) to stratify patients into three risk groups. However, the KRS was not designed to predict PaULVTE so its utility in this context is unknown. A specific PaULVTE risk model is the Michigan Risk Score (MRS) which assesses concurrent central catheter at time of PICC insertion, leukocyte count at insertion, PICC lumen number, antecedent VTE (DVT or PE) and active cancer (the latter confers a score of 3). The MRS has not been externally validated and is not specifically designed for cancer patients which comprised 6.2% of their derivation cohort. Objectives To measure the discriminative ability for PaULVTE of the KRS and MRS and the corresponding prevalence of PaULVTE in patients with active cancer in various KRS and MRS strata. To modify the KRS and MRS, to improve discriminative accuracy for PaULVTE. Methods We retrospectively reviewed consecutive patients with malignancies, who received chemotherapy through a PICC between April 2017 to July 2018. Exclusion criteria for cancer cases included PICC insertion for other reasons such as supportive care. We did not exclude patients on prophylactic or therapeutic anticoagulation or those with a history of PaULVTE. The outcome measures included objectively confirmed PaULVTE incidence whilst the PICC remained in place or within 30 days post removal, treatment and to assess the prevalence of PaULVTE according to KRS and MRS categories in patients with active cancer. For cancer patients, KRS and MRS were calculated based on clinical and laboratory values at PICC insertion. We also assessed whether combining components of both scores would yield improved discriminative ability. Furthermore, we recorded symptomatic, radiologically confirmed distant VTE (i.e. PE, DVT in the proximal or distal lower limb) if this occurred whilst the PICC was in situ or up to 30 days post removal. Results Among 147 cancer patients, median age 64 years, PICC duration 70 days (range, 2-452), 7% developed PaULVTE (95%CI 3.6-12.2) and 4%(95%CI 2-9) distant VTE. Of KRS<1, 22% had no PaULVTE (95%CI 0-11) compared to 9% (95%CI 5-16) in KRS≥1(n=111), p=0.12. Patients with MRS<5 and MRS≥5 developed (95%CI 2-11) and 21% (95%CI 7-48) PaULVTE respectively, p=0.002 and accounted for 83% and 17% of patients respectively. The c-statistic for KRS was 0.6 (95%CI 0.44-0.75) whilst MRS was 0.63 (95%CI 0.44-0.82). The addition of thrombocytosis, a variable from KRS, to MRS (modified MRS, mMRS) improved discriminative value (mMRS 0.72 [95%CI 0.58-0.85]) more than BMI≥35kg/m 2 (0.67 [95%CI 0.49-0.85)] (Figure 1). mMRS3 (lowest score) occurred in 47% with 1.4% (95%CI 0-8.3) PaULVTE compared to 11.8% (95%CI 6.1-21.2) in mMRS>3, p=0.01 (Figure 2). The addition of lumen number from MRS to KRS did not improve the area under the curve (0.6 [95%CI 0.44-0.82]). Conclusion PaULVTE risk is high in active cancer (7%). The MRS dichotomized at a score of above or less than 5 showed moderate discriminative ability for PaULVTE. With simple modification of the MRS with the addition of thrombocytosis (mMRS), a score of 3 had a low rate of PaULVTE (1.4%) and accounted for 47% of patients. mMRS>3 comprised 53% of the cohort and had a statically higher rate of PaULVTE 11.8%. More patients were accurately classified as low PaULVTE risk using mMRS3 compared to KRS<1 (47% versus 22%) but both have low rates of PaULVTE. Prospective validation of mMRS would be invaluable to determine which patients are at high risk of PaULVTE and may benefit from prophylaxis. Figure 1 Figure 1. Disclosures No relevant conflicts of interest to declare.