Modified Khorana risk score for prediction of venous thromboembolic events in cancer patients receiving chemotherapy: An Australian single institution experience.

Abstract

e20610 Background: Venous thromboembolic events (VTEs) are a common complication in cancer. The Khorana Score (KS) is widely used for the prediction of VTEs in malignancy. The KS is composed of 5 items: cancer entity, platelet count >350/nL, white cell count (WCC) >11/nL, Hb <100 g/L and body mass index ≥35 (BMI). Scores are grouped into 3 categories indicating the VTE-risk (0=low, 1-2= intermediate, 3 or more points= high-risk). Methods: All ambulatory cancer patients at our institution starting chemotherapy from January 2010 to December 2011 were included. We applied the KS and then modified by adding further cancer subtypes and metastatic status. Results: In 658 of 766 chemotherapy patients, all the data were available for calculating the KS, of whom 52 had a VTE. In multivariate analysis, associations between KS and VTE were found (P≤0.05) in pancreas (p<0.001), lung (p=0.002), stomach (p=0.008), gynaecological cancers (p=0.037), and BMI ≥35 (p=0.004), but not found in lymphoma (p=0.14), high platelet count (p=0.6) and high WCC (p=0.8), or low Hb (p=0.53). There was an increased risk for VTE in some cancers not included in the KS: breast (p=0.01), colorectal (CRC)(p<0.001), prostate (p=0.003) and oesophageal cancer (p= 0.041). The original KS score did not significantly predict VTEs. When adding cases of neoadjuvant/adjuvant (n/a) and/or metastatic (met) CRC, breast, and prostate cancer, significant associations were found, as shown in the Table. Conclusions: The original KS showed only a weak association with VTE occurrence. However, the association was improved by including other cancer entities and / or metastatic status. Major differences between our and other cohorts, such as different proportions of cancer entities and general referral patterns, could explain the discrepancies with other studies. [Table: see text]