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

Abstract

4535 Background: VTE causes substantial morbidity and mortality in GCT pts. The Khorana model is a validated predictive model that stratifies VTE risk in cancer pts receiving chemo; Khorana score ≥3 (K3) identifies pts at high risk. Many GCT pts have bulky RPLN that can cause venous stasis in the lower limbs. This study examines the incidence of VTE in GCT pts and assesses large RPLN as a novel predictor of VTE risk. Methods: Retrospective data from the Princess Margaret Hospital GCT database was complemented by review of medical records. GCT pts receiving 1st line chemo between 2000-2010 were included. Pts diagnosed with VTE prior to chemo and pts receiving thromboprophylaxis (TP) were excluded. Pts diagnosed with VTE during or within 3 months of completing chemo were identified. Large RPLN were defined as having maximal diameter ≥5cm. Odds ratios for VTE risk with large RPLN and K3 were calculated. Discriminatory accuracy (DA) of each predictor was calculated using area under the receiver operating characteristic curves (AUROC). An external cohort from London Regional Cancer Program, with similarly collected data, was used to validate results. Results: In the test cohort 21 (10%) of 216 pts developed VTE. Both large RPLN (OR 6.2, p<0.001) and K3 (OR 11.8, p<0.001) were significantly associated with VTE. Positive predictive value (PPV) was lower for large RPLN compared to K3 (21% v 44%, p<0.014) but sensitivity was greater (71% v 40%, p=0.001), while DA showed no difference (AUROC 0.73 v 0.67, p=0.46). In the validation cohort 10 (9%) of 111 pts developed VTE. There was a non significant trend for associations between VTE and both large RPLN (OR 2.54, p=0.16) and K3 (OR 4.5, p=0.13). When compared to K3, sensitivity was greater for large RPLN (60% v 20%, p=0.003), but there was no difference in PPV (14% v 29%, p=0.25) or DA (AUROC 0.61 v 0.57, p=0.72). Conclusions: VTE occurs in 1 in 10 GCT pts receiving curative chemotherapy. Large RPLN is a novel and clinically applicable predictor of VTE risk, although prospective validation would be beneficial. Randomized controlled trials of TP in GCT pts should be considered in pts at high risk of VTE, identified by large RPLN or K3.