lunes, 23 de mayo de 2016

PO-04 - A new genetic risk score for predicting venous thromboembolic events in cancer patients receiving chemotherapy. - PubMed - NCBI

2016 Apr;140 Suppl 1:S177-8. doi: 10.1016/S0049-3848(16)30137-2. Epub 2016 Apr 8.

PO-04 - A new genetic risk score for predicting venous thromboembolic events in cancer patients receiving chemotherapy.

Martín AJ1, Ziyatdinov A2, Rubio VC1, Olmos VP1, Huerta BM1, Rodríguez JC1, Fernández MS1, de Castro EM1, Caro RL1, Fernández JM2.

Author information

Abstract

BACKGROUND:

Venous thromboembolism (VTE) is one of the major causes of cancer-associated mortality. Risk for developing VTE increases when in chemotherapy, mainly in the outpatient setting. Current risk scores for predicting chemotherapy-associated VTE have low/moderate discrimination capacity. These models use clinical parameters. ThromboinCode (TiC) is a new tool for VTE risk prediction using an algorithm that combines a genetic risk score (GRS) with subject's VTE clinical risk parameters (cancer type and cancer disease status "CDS", included).

AIMS:

To evaluate whether TiC predicts better the risk for chemotherapy-associated VTE than Khorana score.

METHODS:

A prospective, observational study including 251 patients with locally advanced or metastatic cancer (colon, stomach, pancreas and lung) receiving systemic outpatient chemotherapy. Patients are followed-up for 6 months. Three predictive models were compared: a) Khorana score; b) Khorana score plus CDS and c) TiC. Genetic variants included in TiC are FVL, PT, F5 rs118203906 and rs118203905, F12 rs1801020, F13 rs5985, SERPINC1 rs121909548, SERPINEA10 rs2232698 and A1 blood group rs8176719, rs7853989, rs8176743, rs8176750. Clinical risk factors in TiC are age, sex, family history of VTE, BMI, smoking, diabetes, type of cancer and CDS. Prediction capacity of each model was assessed in terms of the discrimination (area under the receiver operating characteristic curve, AUC).

RESULTS:

The incidence of VTE at 6 months was 23.11%. Korana score had an AUC of 0.550 (95% CI 0.485-0.613, p=0.2162), sensitivity 64.41, specificity 46.56, positive likelihood ratio of 1.21, negative likelihood ratio of 0.76. Khorana score+CDS prediction model had an AUC of 0.609 (95% CI 0.545-0.670, p=0.008), sensitivity 70.69, specificity 50.53, positive likelihood ratio of 1.25, negative likelihood ratio of 0.67. Discrimination increased significantly with TiC (AUC: 0.70; 95% CI 0.636 - 0.753, p<0.0001), sensitivity 79.31, specificity 55.44, positive likelihood ratio of 1.78, negative likelihood ratio of 0.37. The discrimination with TiC was significantly higher than with the other two predictive models (p<0.05).