Development of a Score for Prediction of Occult Malignancy in Stroke Patients (Occult-5 Score)

Abstract

Background and purpose: Malignancy associated acute ischemic stroke (AIS) requires specific diagnostic work-up, treatment and prevention to improve outcome. This study aimed to develop a biomarker-based score for prediction of occult malignancy in AIS patients.

Methods: Single-center cross-sectional study including consecutive AIS patients treated between July 2017 and November 2018. Patients with active malignancy at presentation, or diagnosed within 1 year thereafter and patients free of malignancy, were included and malignancy associated biomarkers were assessed. LASSO analyses of logistic regression were performed to determine biomarkers predictive of active malignancy. Predictors were derived from a predictive model for active malignancy. A comparison between known and unknown (=occult) malignancies when the index stroke occurred was used to eliminate variables not associated with occult malignancy. A predictive score (OCCULT-5 score) for occult malignancy was developed based on the remaining variables.

Results: From 1001 AIS patients, 61 (6%) presented an active malignancy. Thirty-nine (64%) were known and 22 (36%) occult. Five variables were included in the final OCCULT-5 score: age ≥ 77 years, embolic stroke of undetermined source, multi-territorial infarcts, D-dimer levels ≥ 820 µ/gL, and female sex. A score of ≥ 3 predicted an underlying occult malignancy with a sensitivity of 64%, specificity of 73%, positive likelihood ratio of 2.35 and a negative likelihood ratio of 0.50.

Conclusions: The OCCULT-5 score might be useful to identify patients with occult malignancy. It may thus contribute to a more effective and timely treatment and thus lead to a positive impact on overall outcome.

Keywords: Biomarkers; Ischemic stroke; Malignancy-related stroke; Occult malignancy; Predictive score.