Ability of the number of territories involved on DWI-MRI to predict occult systemic malignancy in cryptogenic stroke patients.

ABSTRACT

BACKGROUND: Lesions in multiple arterial territories is one of the typical features of malignancy patients with cryptogenic stroke. Hence, if patients with cryptogenic stroke display such feature, occult cancer could be predicted. The study aimed to analyze the predictive ability of the number of territories involved on DWI-MRI for occult systemic malignancy (OSM) in patients with cryptogenic stroke. METHODS: We enrolled patients with cryptogenic stroke without a diagnosis of malignancy at stroke onset between January 2013 and November 2018. Clinical variables were analyzed between cryptogenic stroke patients with and without OSM through univariate and multiple logistic regression analyses. Points for OSM were generated by ß-coefficients. The sensitivity and specificity of the risk score were assessed by the area under the receiver operating characteristic curve (AUROC). The cutoff value for predicting OSM was determined by the maximum Youden index. RESULTS: Among 108 cyptogenic stroke patients, compared to patients without OSM (n = 96), patients with OSM (n = 12) had a lower nutrition status (P = 0.031), higher plasma D-dimer levels (P < 0.001) and more territories involved on DWI-MRI (P < 0.001). Multiple logistic regression analysis revealed that plasma D-dimer levels (OR, 3.54; 95% Cl, 1.62-7.76; P = 0.002) and the number of territories involved (OR, 4.45; 95% CI, 1.25-15.80; P = 0.021) independently predicted OSM. The predictive score system built upon the number of territories showed good discrimination with an AUROC of 0.84 (95% CI, 0.71-0.96). The cutoff value was 2 with a maximum Youden's index of 0.56, which means that patients with more than one territory involved on DWI-MRI may need extensive screening for OSM. CONCLUSIONS: The number of territories involved on DWI-MRI was a valid predictor for OSM in cryptogenic stroke patients who need to undergo further evaluations .