Predictive value of plasma (D)-dimer levels for cancer-related stroke: a 3-year retrospective study.

BACKGROUND

Cerebrovascular disease is the second leading cause of central nervous system pathology in cancer patients. Cancer-associated hypercoagulation plays an important role in cancer-related stroke. The present study aims to test whether plasma d-dimer levels could predict comorbid malignancy in patients with ischemic stroke.

METHODS

Five hundred sixteen stroke patients with measured d-dimer levels and who were consecutively admitted to our stroke center from 2009 to 2012 were included. Cancer status was determined by medical chart, and 59 patients were identified to have active cancer. An additional 48 cancer patients with stroke were identified from the hospital database. Several d-dimer cutoff levels were used to predict cancer-related stroke.

RESULTS

Stroke patients with active cancer had significantly higher d-dimer levels than those without cancer (P < .001). The average d-dimer level in stroke patients without cancer was .66 ± 1.83 mg/L, whereas the levels for active cancer patients from the stroke center and hospital database were 5.70 ± 9.63 mg/L and 10.47 ± 12.31 mg/L, respectively. When using d-dimer of .55 mg/L or more and multiple territory infarctions as criteria, the specificity and positive predictive value (PPV) for cancer-related stroke were 99.7% and 92.9%, respectively. When using d-dimer of 5.5 mg/L or more as the cutoff value, the test had a high specificity and PPV regardless the brain magnetic resonance imaging (MRI) findings. Six stroke patients fitting our criteria were confirmed to have occult malignancy after comprehensive cancer survey.

CONCLUSIONS

Extraordinary high d-dimer levels or combining d-dimer and MRI findings may be used as a screening tool to detect malignancy in stroke patients.