RCVS2 score and diagnostic approach for reversible cerebral vasoconstriction syndrome

Objective To develop a method to distinguish reversible cerebral vasoconstriction syndrome (RCVS) from other large/medium-vessel intracranial arteriopathies. Methods We identified consecutive patients from our institutional databases admitted in 2013-2017 with newly diagnosed RCVS (n = 30) or non-RCVS arteriopathy (n = 80). Admission clinical and imaging features were compared. Multivariate logistic regression modeling was used to develop a discriminatory score. Score validity was tested in a separate cohort of patients with RCVS and its closest mimic, primary angiitis of the CNS (PACNS). In addition, key variables were used to develop a bedside approach to distinguish RCVS from non-RCVS arteriopathies. Results The RCVS group had significantly more women, vasoconstrictive triggers, thunderclap headaches, normal brain imaging results, and better outcomes. Beta coefficients from the multivariate regression model yielding the best c-statistic (0.989) were used to develop the RCVS2 score (range -2 to +10; recurrent/single thunderclap headache; carotid artery involvement; vasoconstrictive trigger; sex; subarachnoid hemorrhage). Score 5 had 99% specificity and 90% sensitivity for diagnosing RCVS, and score 2 had 100% specificity and 85% sensitivity for excluding RCVS. Scores 3-4 had 86% specificity and 10% sensitivity for diagnosing RCVS. The score showed similar performance to distinguish RCVS from PACNS in the validation cohort. A clinical approach based on recurrent thunderclap headaches, trigger and normal brain scans, or convexity subarachnoid hemorrhage correctly diagnosed 25 of 37 patients with RCVS2 scores 3-4 across the derivation and validation cohorts. Conclusion RCVS can be accurately distinguished from other intracranial arteriopathies upon admission, using widely available clinical and imaging features.