Challenging the Intervention Bias: Moving from Automatic Action to Strategic Restraint in INR: Biplane Weekly No. 4

This post hoc analysis asked whether CT perfusion mismatch could identify patients with medium vessel occlusion who benefit from thrombectomy in ESCAPE-MeVO, a trial in which thrombectomy was not superior to usual care overall. Interpretable baseline perfusion was available for 309 of 530 randomised patients (146 thrombectomy, 163 usual care). Median mismatch volume was 37.2 mL and median mismatch ratio 6.1.

Neither measure was associated with 90-day ordinal mRS: adjusted OR 1.03 (95% CI 0.99–1.07) per 10 mL of mismatch volume and 1.03 (0.97–1.09) per unit of ratio. Neither modified the treatment effect on ordinal mRS or on mRS 0–1 and 0–2. Interactions with volume reached nominal significance at thresholds of 30 mL (P=.04) and 40 mL (P=.03) but not at 10, 20 or 50 mL, and were uncorrected for multiple testing; the authors call them hypothesis-generating. The nominal interaction comes from the low-mismatch group: below 30 mL, the adjusted common OR was 2.08 (1.07–4.04) in favour of usual care, against 0.82 (0.49–1.37) at 30 mL or more. A higher ratio was associated with a smaller 24-hour infarct volume.

The analysis is exploratory, and 220 of 530 patients (41.5%) were excluded for missing or uninterpretable perfusion imaging, with baseline differences between those analysed and those not. It sits beside three neutral randomised trials of medium vessel thrombectomy, ESCAPE-MeVO itself, DISTAL and DISCOUNT, and the positive ORIENTAL-MeVO, which enrolled patients with more severe deficits. Within ESCAPE-MeVO, mismatch thresholds derived in large vessel occlusion did not sort medium vessel occlusion patients by benefit.

A nationwide registry from 116 Chinese sites (2023–2025) compared thrombectomy plus standard medical treatment with standard medical treatment alone in 2,524 patients with distal medium vessel occlusion presenting within 24 hours of last known well; 1,564 were treated with thrombectomy and 960 were not.

With inverse probability of treatment weighting, freedom from disability (mRS 0–1) at 90 days was more frequent with thrombectomy: adjusted risk ratio 1.15 (95% CI 1.02–1.29), an adjusted risk difference of 4.36 percentage points (1.73–6.99). Symptomatic intracranial haemorrhage within 48 hours was also more frequent: adjusted risk ratio 4.03 (2.46–6.59). Ninety-day mortality did not differ (adjusted risk ratio 0.97, 0.81–1.16).

These are observational data, in which confounding by indication cannot be excluded, and the lower confidence limit for the functional benefit is close to 1. The pooled randomised trials Lumen covered on 12 October found no thrombectomy benefit in distal medium vessel occlusion; this larger real-world cohort finds a small one, paid for in symptomatic haemorrhage.

This meta-analysis pooled 12 randomised trials comparing minimally invasive haematoma evacuation, endoscopic and other approaches, with medical or conventional surgical management in acute spontaneous supratentorial intracerebral haemorrhage, using random-effects models and trial sequential analysis.

All-cause mortality was lower with evacuation: RR 0.77 (95% CI 0.61–0.97), and RR 0.79 (0.64–0.96) at six months. Functional outcome did not differ significantly, whether defined as mRS 0–1 or 0–3. Rebleeding was more frequent with evacuation (RR 2.37, P=0.002), as were cardiac events (RR 5.08, P=0.009).

Both mortality intervals lie close to 1, and trial sequential analysis judged the reductions not yet conclusive: the required information size was not reached (2,118 of 5,372 patients for all-cause mortality; 1,485 of 2,833 at six months). Endoscopic and other techniques were pooled although they are different procedures. Fewer deaths without better function leaves open what disability the additional survivors live with.

A fixed-effect network meta-analysis of 11 randomised trials compared medical therapy, balloon angioplasty, drug-coated balloon, bare-metal stent and drug-eluting stent in symptomatic intracranial atherosclerosis with 70–99% stenosis. The 30-day network contained 10 studies and 2,334 participants; the long-term network, mostly at one year, contained 8 studies and 1,712 patients. 82.5% of participants were Asian.

The one significant finding was early: for stroke or death within 30 days, medical therapy was safer than bare-metal stenting (RR 0.47, 95% CI 0.31–0.72). No other pairwise difference was significant at 30 days, and none was significant in the long-term network.

Certainty was moderate for the safety network and low or very low for most long-term comparisons. The network had no closed loops, so inconsistency could not be tested, and the drug-coated balloon node rests on a single trial. The result is consistent with the early hazard of bare-metal stenting seen in the trials it pools, but the network does not rank the devices for long-term benefit.

In 313 patients with chronic subdural haematoma from the 12-centre MESH Registry (2016–2024), 172 (55.0%) were treated with middle meningeal artery embolisation alone and 141 (45.0%) with embolisation adjunctive to surgical evacuation. The question was whether more liquid embolic is better.

No dose metric was significantly associated with the 180-day composite of retreatment, residual or recurrent haematoma over 10 mm, or death. Absolute volume gave an OR of 0.80 per interquartile-range increment (95% CI 0.56–1.16, P=0.24), where an OR below 1 would favour larger volumes, and normalising to vessel diameter, baseline thickness or estimated haematoma volume did not change this. Median volume was 0.60 mL. Centre explained 44.5% of the variation in volume; haematoma burden explained 0.3%. Procedural complications were 6.4% and not dose-related.

This is a retrospective registry without a control arm, so it cannot define a minimum effective dose. It does argue against treating injected volume as a procedural target.

A meta-analysis of 9 observational studies, 282 patients with Loeys-Dietz syndrome, found intracranial aneurysms in 64, a pooled prevalence of 23% (95% CI 18–28%). Aneurysms were more frequent in women (OR 2.49, 1.51–4.10). Mean size was 3.23 mm, 78.1% were saccular and 85% lay in the anterior circulation or posterior communicating segment.

No rupture occurred among 62 aneurysms followed for 2.0 to 8.9 years across studies. Treatment data came only from 11 case reports covering 16 treated aneurysms, all occluded without major complications attributed to the connective tissue disorder.

Most aneurysms were small, and none ruptured over follow-up of 2.0–8.9 years across studies, but the evidence is insufficient to define their natural history. With treatment data drawn only from case reports, the evidence describes how often aneurysms occur in Loeys-Dietz syndrome, not how they behave.