Collateral status predicts functional outcome in early-treated large-core anterior circulation stroke.

BACKGROUND AND PURPOSE: Endovascular therapy (EVT) is increasingly offered to patients with large-core acute ischemic stroke (AIS), yet outcomes remain highly heterogeneous. Collateral circulation may be a key determinant of infarct evolution and recovery, but its role in early-window large-core stroke is not fully defined.

METHODS

We retrospectively analyzed consecutive adults from a prospective stroke registry who presented within 6 h with anterior-circulation large-vessel occlusion, NIHSS ≥6, and a large ischemic core (MRI core >50 mL or CT perfusion core >70 mL, up to 150 mL). All patients received reperfusion therapy (intravenous thrombolysis, EVT, or both). Collateral status on baseline single-phase CTA was graded using the Tan scale (0-3); no patients had grade 3. The primary outcome was 90-day modified Rankin Scale (mRS); secondary outcome was NIHSS at discharge.

RESULTS

Fifty-four patients met inclusion criteria (Tan 0: n = 24; Tan 1: n = 14; Tan 2: n = 16). Baseline NIHSS, ASPECTS, and core volume were similar across groups. Patients without collaterals (Tan 0) had worse 90-day outcomes (median mRS 4 [IQR 3-6]) compared with those with Tan 1 (2 [IQR 1-3]) or Tan 2 (1 [IQR 1-2]) collaterals (both p < 0.001), whereas Tan 1 and Tan 2 did not differ significantly (p = 0.27). NIHSS at discharge showed a similar gradient. In proportional-odds logistic regression, each one-grade increase in collateral status was associated with lower odds of worse 90-day mRS (adjusted per-grade OR 0.32; 95% CI 0.15-0.68; p = 0.003).

CONCLUSION

In early-treated large-core AIS, even simple CTA-based collateral assessment strongly predicts recovery. Patients with absent collaterals follow a distinctly poorer trajectory, while those with any collateral filling behave more favorably. Incorporating collateral status into routine evaluation may improve prognostic accuracy and support treatment decisions in this challenging subgroup.