Large retroperitoneal lymph nodes (RPLN) as a predictor for venous thromboembolism (VTE) in patients (pts) with germ cell tumor (GCT) receiving first-line chemotherapy (chemo).

Abstract

332 Background: VTE causes significant morbidity and mortality in GCT pts. While an existing and validated predictive model identifies VTE risk in chemo pts with any cancer (Khorana model), a predictive model specific to GCT does not exist. Many GCT pts present with bulky RPLN that produce venous stasis in the lower extremities. The objective of this study was to explore the association between large RPLN and VTE in GCT pts receiving chemo and compare large RPLN as a predictor for VTE in GCT pts to the non-GCT specific Khorana model. Methods: Clinical data from our institutional GCT database was complemented by review of radiology, pharmacy and medical records. All GCT pts receiving 1st line chemo between 1-Jan-00 and 31-Dec-10 were included. Large RPLN were defined as ≥5cm in maximal diameter. Factors used in the Khorana model (baseline BMI, hemoglobin, white cell count and platelets) were collected. We compared the predictive accuracy of large RPLN versus Khorana score ≥3 using receiver operator characteristic (ROC) curve statistical analyses. Results: The cohort consisted of 260 GCT pts, median age 31.5 years, predominantly testis primary (235, 90%) and good risk (171, 66%). 17 (7%) developed VTE prior to the start of chemo. 19 (7%) were given prophylactic anticoagulation, none of whom developed VTE. Of the remaining 224 pts, 20 (9%) developed VTE during chemo. In a univariate analysis, large RPLN was strongly associated with VTE (OR 7.74, p<0.001), as were Khorana score ≥3 (OR 9.81, p<0.001) and hospital admission during chemo (OR 3.96, p=0.004). ROC curve analyses demonstrated large RPLN was a significant individual predictor for VTE (AUC 0.588, p=0.03), however, Khorana score ≥3 was a better predictor (AUC 0.664, p=0.02). Adding large RPLN to create a modified Khorana score provided marginal gains (AUC 0.682, p=0.02). Conclusions: Although large RPLN at diagnosis predicts for VTE in GCT pts, the Khorana predictive model is superior. Given the high rate of VTE in GCT pts receiving chemo, we recommend prophylactic anticoagulation for pts at increased risk, including pts with Khorana score ≥3, pts requiring hospital admission or pts with large RPLN.