Introduction: Cancer patients who develop acute ischemic stroke (AIS) are at high risk for recurrent thromboembolism. No risk stratification model exists to predict recurrent events in this population. The Khorana score is a validated risk score for predicting venous thromboembolism in newly diagnosed cancer patients.
Hypothesis: The Khorana score can effectively classify the risk of recurrent thromboembolism (RTE) in cancer patients with AIS.
Methods: We retrospectively identified all adults with active systemic cancer diagnosed with AIS by MRI at a tertiary-care cancer center from 2005 through 2009. Two neurologists independently reviewed all available medical records through July 31, 2012. The Khorana score at the time of index stroke was calculated for each patient. Points were assigned for specific cancer sites (2 points for very high-risk sites: stomach or pancreas; 1 point for high-risk sites: lung, lymphoma, gynecologic, bladder, testicular, or renal), platelet count ≥350,000/mcL, hemoglobin <10 g/dl, leukocyte count ≥11,000/mcL, and BMI ≥35 kg/m2. The primary outcome was a composite of RTE, defined as recurrent AIS, TIA, MI, systemic embolism, DVT, or PE. Logistic regression was used to evaluate the association between individual components of the Khorana score and RTE. The c-statistic was used to calculate the discriminatory ability of the Khorana score in predicting RTE.
Results: Among 263 study patients, 90 (34%) were diagnosed with RTE, including 36 (14%) with recurrent AIS. The median Khorana score was 2 (IQR 1-2, range 0-5). None of the individual components of the score were independently associated with RTE, although there was a nonsignificant trend for high-risk cancer sites (OR 1.56, 95% CI 0.88-2.77). The rate of RTE was 28% among patients with a score of 0, 36% among patients with a score of 1-2, and 32% among patients with a score of 3-6. The c-statistic was 0.52 (95% CI 0.45-0.58) for predicting RTE and 0.49 (95% CI 0.38-0.59) for predicting recurrent AIS.
Discussion: The Khorana score has poor discriminatory ability for predicting RTE in cancer patients with AIS, probably because these patients represent an especially high-risk group. Future research is needed to identify better methods for predicting RTE in this high-risk population.