Thrombectomy Beyond the Usual Boundaries: Biplane Weekly No. 6

LATE-MT asked whether thrombectomy still helps when patients present well beyond the late windows tested in DAWN (6 to 24 hours) and DEFUSE 3 (6 to 16 hours). In this open-label randomised trial with blinded outcome assessment, 336 patients with an anterior-circulation large-vessel occlusion and imaging-identified salvageable tissue were randomised 24 to 72 hours after onset or last known well, at comprehensive stroke centres in China: 168 to thrombectomy plus standard medical care and 168 to standard care alone. Median time to randomisation was 38.5 hours.

At 90 days the modified Rankin scale distribution favoured thrombectomy (common odds ratio 1.74, 95% CI 1.19 to 2.55). The gain came with a safety cost: symptomatic intracranial haemorrhage occurred in 4.8% of the thrombectomy group and in none of the controls, and 90-day mortality was 6.0% versus 1.8%.

Beyond DAWN's 24-hour limit, the trial offers a trade rather than a free gain: a better 90-day disability distribution against more symptomatic haemorrhage and numerically higher mortality, in imaging-selected patients from a single health system.

MILD-MT addressed patients with a large-vessel occlusion but only a mild deficit. The open-label trial in China randomised 304 patients with NIHSS below 6, an anterior-circulation occlusion, a small core and a large perfusion mismatch within 24 hours of onset or last known well. Three hundred entered the modified intention-to-treat analysis: 144 thrombectomy and 156 medical management alone.

Excellent 90-day outcome (mRS 0 to 1) occurred in 69.4% with thrombectomy and 50.0% with medical management (odds ratio 2.32, 95% CI 1.41 to 3.81; P<0.001). Symptomatic intracranial haemorrhage within 48 hours occurred in three patients in each group. Serious procedure-related complications affected 19 thrombectomy patients (13.2%), and the one death in the trial was in the thrombectomy group.

The absolute difference of about 19 percentage points means roughly one additional patient in five reaching mRS 0 to 1, a large gain for patients who arrive with little deficit, achieved in a population selected on core and perfusion mismatch.

A cumulative meta-analysis in the Journal of Clinical Medicine included nine randomised trials (2,828 participants) of adjunctive intra-arterial fibrinolytic given after successful or near-successful reperfusion; eight contributed to the pooled outcomes. Excellent 90-day outcome (mRS 0 to 1) was more frequent with intra-arterial treatment (risk ratio 1.266, 95% CI 1.135 to 1.412), with no heterogeneity (I² 0%) and a prediction interval that stayed above 1 (1.047 to 1.530). Mortality did not differ significantly (risk ratio 1.036, 0.817 to 1.314), but the estimate moved when CHOICE-2 was added: it was 0.984 (0.801 to 1.209) before, and CHOICE-2 itself recorded mortality of 12.1% in the alteplase arm versus 6.4% in controls. In a reverse-fragility analysis, changing the outcome of 24 patients was enough to produce a nominally significant increase in mortality (risk ratio 1.174, 1.002 to 1.377); the authors state that this does not establish a mortality excess. They rate certainty as moderate for both outcomes.

Two subgroup analyses of randomised trials add context (PEARL prespecified, POST-UK a secondary analysis). In PEARL, patients who had not received intravenous thrombolysis (n=189) reached mRS 0 to 1 in 43.2% versus 22.6% (adjusted risk ratio 1.86, 1.19 to 2.91), with no significant difference in symptomatic haemorrhage (adjusted risk ratio 0.93, 0.28 to 3.13). In POST-UK, patients with cardioembolic stroke (n=209) had more disability-free survival with intra-arterial urokinase (adjusted risk ratio 1.49, 1.08 to 2.06) and no significant difference in death or intracranial haemorrhage. Neither is a selection criterion; both are subgroups of single trials, and the pooled estimate remains the better guide.

The basilar-occlusion trials ATTENTION and BAOCHE enrolled patients with NIHSS of 10 or more and 6 or more respectively, leaving milder deficits to observational data. This eight-centre prospective registry (ChiCTR2300075711) compared thrombectomy plus best medical management with medical management alone in 401 patients with basilar occlusion and NIHSS below 10, using inverse-probability weighting.

Functional independence (mRS 0 to 2 at 90 days) was more frequent with thrombectomy (odds ratio 1.55, 95% CI 1.17 to 2.06), as was mRS 0 to 1 (odds ratio 1.48, 1.09 to 2.03), and early neurological deterioration was less frequent (odds ratio 0.35, 0.20 to 0.61); sensitivity analyses for crossover attenuated the functional advantage. The association varied with baseline severity (interaction P=.02). The functional advantage was confined to NIHSS 6 to 9, a range BAOCHE already admitted. The new ground is NIHSS 0 to 5, below both trials' thresholds: there thrombectomy was associated with less early deterioration and death but not with better 90-day function, an exploratory contrast rather than a treatment threshold.

Two single-centre retrospective series compared embolic agents in middle meningeal artery embolisation for chronic subdural haematoma.

In Chemnitz, 194 procedures in 147 patients (median age 78) used Squid 12 (118 procedures) or Magic Glue (76). Unplanned repeat surgical evacuation followed 15 of 118 Squid 12 procedures (13%) and 3 of 76 Magic Glue procedures (4%): unadjusted risk ratio 0.31 for Magic Glue versus Squid 12 (95% CI 0.10 to 0.96). The difference was not independent in the exploratory multivariable analysis, baseline haematoma thickness differed (18 versus 12 mm for Squid 12 versus Magic Glue), and the authors call the finding hypothesis-generating.

In Little Rock, calibrated Embozene 75 µm microspheres (34 embolisations) were compared with n-BCA or Onyx 18 (32). Embolisation failure occurred in 5 of 34 versus 4 of 32, and unplanned evacuation in 3 of 34 versus 1 of 32. The series is underpowered, agent choice was discretionary, and one fatal posterior-circulation infarct occurred after particle embolisation (1 of 35 versus 0 of 32).

A retrospective cohort from Frankfurt reports 55 patients with cranial dural arteriovenous fistulas (direct carotid-cavernous fistulas excluded) treated between 2014 and October 2019 with transarterial liquid embolisation, transvenous coils, or both, chosen according to angioarchitecture and accessibility. Thirty-nine of the 55 were high grade, and venous or sinus thrombosis was present in 22.

Complete occlusion was achieved in 49 patients and three more had residual Cognard I shunting, giving angiographic success in 52 of 55. There was one recurrence, three clinical complications and one technical complication. The nine presenting haemorrhages were associated with high-risk venous drainage (8 of 22 versus 1 of 33; p=0.0017) and with high-grade lesions (9 of 39 versus 0 of 16; p=0.046). Among 50 patients with patient-reported follow-up (median 53 months), 46 reported improvement; tinnitus improved most, and diplopia persisted in 2 of 6.

In the Vascular Quality Initiative registry (January 2005 to January 2025), 275,668 patients underwent carotid intervention (endarterectomy, transcarotid revascularisation or stenting); 73,582 (26.7%) had diabetes, 29,541 of them insulin-dependent. After multivariable adjustment, insulin dependence was associated with higher 30-day odds of myocardial infarction (odds ratio 1.63), ipsilateral stroke (1.24), death (1.19) and major adverse cardiovascular events (1.19), but not of the composite of ipsilateral stroke or death.

The excess was concentrated in asymptomatic patients: the 30-day odds ratio for major adverse cardiovascular events was 1.35 in asymptomatic patients versus 1.07 in symptomatic ones (interaction P<0.001). Compared with endarterectomy, transcarotid revascularisation and stenting were associated with higher 30-day odds of ipsilateral stroke (odds ratios 2.10 and 1.93), and the interaction with insulin dependence was not significant for any endpoint.