Angelo Cascio Rizzo, Ghil Schwarz
ISPS25 is feasible and recategorizes many embolic stroke of undetermined source cases, but attribution is largely probabilistic and most patients remain undetermined. Limited extended diagnostics and frequent overlapping PESs reveal gaps in both diagnostic workup and the ISPS25 construct, supporting expansion and further refinement.
BACKGROUND: The Ischemic Stroke Phenotyping System 2025 (ISPS25) expands the minimum diagnostic evaluation and introduces causality grading to improve stroke classification but lacks formal validation. We assessed its real-world applicability in a large embolic stroke of undetermined source (ESUS) cohort and explored additional potential embolic sources (PESs) not captured by the original framework.
METHODS: We conducted a single-center study including consecutive patients with embolic stroke of undetermined source admitted to Niguarda Hospital (Milan, Italy). Two neurologists independently reviewed records, assigned ISPS25 categories (large-artery atherosclerosis, cardioembolism, small-vessel disease, other determined, undetermined) and the highest causality grade (definite/probable/possible); equal-grade categories were classified as multiple causes. Prespecified additional PESs (cardioembolic/large-artery atherosclerosis/other) not included in ISPS25 were collected. We quantified pathogenetic redistribution after ISPS25 application and compared second-pass testing and PES prevalence between ISPS25-reclassified and ISPS25-undetermined patients.
RESULTS: Among 802 patients (median age, 69 years; 44.9% women; National Institutes of Health Stroke Scale score, 5), ISPS25 reassigned pathogenesis in 313 (39.0%): cardioembolic, 186 (23.2%); large-artery atherosclerosis, 68 (8.5%); other determined, 54 (6.7%); and multiple causes, 5 (0.6%). Causality was definite in 24 (7.7%), probable in 112 (35.8%), and possible in 177 (56.5%). Undetermined cases (489 [61.0%]) underwent less second-pass testing than reclassified patients (46.2% versus 70.3%, P<0.001) yet more frequently harbored additional PESs (62.0% versus 42.5%, P<0.001) and overlapping PESs (25.1% versus 10.5%, P<0.001). Considering additional PESs, predominantly cardioembolic, as candidate mechanisms reduced undetermined cases to 23.2%.
CONCLUSIONS: ISPS25 is feasible and recategorizes many embolic stroke of undetermined source cases, but attribution is largely probabilistic and most patients remain undetermined. Limited extended diagnostics and frequent overlapping PESs reveal gaps in both diagnostic workup and the ISPS25 construct, supporting expansion and further refinement.