What this is

The MR CLEAN Registry analysis of 539 patients with M2 segment MCA occlusions finds that bridging therapy (IV alteplase before EVT) was associated with better 90-day functional outcomes than EVT alone — but only one of nine reported endpoints clears conventional statistical significance, and the benefit is concentrated in the mothership subgroup. Below: where the claim holds, where it strains, and what an M2 program should actually do on Monday morning.

1. Shadow Audit

What the article is not saying.

The headline result — “bridging therapy may yield superior functional outcomes” — rests on a single ordinal mRS shift with a confidence interval that just barely clears unity (aOR 1.52 [1.04–2.21]; P=0.03). Every dichotomized secondary endpoint is non-significant: the paper itself reports “dichotomized functional outcomes and mortality were numerically in favor of IVT+EVT” with P values of 0.19 (mRS 0–1), 0.15 (mRS 0–2), 0.09 (mRS 0–3), and 0.51 (90-day mortality). One test does all the work for a positive conclusion.

The shadow audit also has to flag the structural exclusion: “We must consider that patients who responded to IVT and did not require EVT were excluded from the analysis.” That exclusion removes the very patients for whom IVT works best — early recanalizers who never reach the angio suite — and biases the comparison toward IVT-refractory cases. The transfer subgroup is explicitly noted to be enriched for these patients: “excluding those who responded to IVT and did not need EVT left primarily IVT-refractory cases, which may have skewed the results.”

What’s missing from the comparison: no head-to-head against TNK bridging, no stratification by M2 sub-territory (superior vs inferior division; dominant vs codominant), and no E-value or sensitivity analysis for residual confounding despite the authors’ own concession that “the potential for residual bias remains.”

2. Inversion Engine

What would have to be true for the opposite conclusion.

For EVT alone to be non-inferior or superior for M2 occlusions, three things would need to hold.

First, the IVT response-exclusion bias would have to be negligible. The IRIS patient-level meta-analysis framing — that “delaying EVT for IVT administration may negate any potential benefit, as each hour of delay reduces functional independence by 6%” — would predict that in a population that includes IVT responders, the EVT-alone arm looks better because no patient is delayed for futile IVT. The MR CLEAN analysis never sees that population.

Second, the adjustment would have to fail to capture the contraindication signal. With “anticoagulation use at 1.9% vs 16.9%” and “atrial fibrillation at 19.0% vs 39.0% (P<0.001)” between IVT+EVT and EVT-alone groups, the EVT-alone arm is enriched for patients in whom clinicians already feared hemorrhagic transformation or suspected large cardioembolic thrombi. If those patients would have done worse with IVT too — and the comparison never measures that — then the propensity-weighted result still overstates the IVT benefit.

Third, the recanalization channel would have to matter less than the paper assumes. “Successful recanalization (eTICI 2b-3)” was numerically higher with EVT alone (69.9% vs 64.7%), as were excellent (2c/3) and complete (3) recanalization. If reperfusion is the dominant mechanism of EVT benefit — which the broader EVT literature asserts — then a therapy that reduces recanalization cannot, in expectation, improve functional outcome unless IVT provides an entirely separate microvascular benefit (“early partial or complete recanalization… enhance microvascular reperfusion, reduce infarct growth”). The mechanism is plausible but unfalsified here.

The assumption that breaks first: that ordinal mRS shift with overlapping CIs across every secondary endpoint is reliable signal rather than the primary endpoint happening to land just below alpha in a single regression specification.

3. Second-Order Catalyst

If the result is real — who changes practice first.

In 12 months, stroke systems of care that already default to bridging will be reinforced — particularly in mothership pathways. The AHA/ASA class IIb endorsement for M2 thrombectomy becomes more practice-shaping because the addition of IVT gives the neurointerventionalist a “do everything” rationale that aligns with the broader institutional reflex to give tPA when no contraindication exists. The transfer pathway, where IVT was already given at the spoke, gets a softer reinforcement — the paper found no significant effect in the transfer subgroup (aOR 1.19, P=0.62), with P interaction = 0.02 between mothership and transfer.

Who changes practice first: comprehensive stroke centers with mothership dominance and emergency departments that already treat M2 occlusion as aggressively as M1. The Dutch MR CLEAN network itself will fold this finding into the next protocol iteration of the MR CLEAN Registry analysis pipeline. European systems already practicing drip-and-ship with alteplase will codify the mothership-bridging recommendation.

Who changes practice last: centers that have already shifted to tenecteplase-only protocols and are waiting for BRIDGE-TNK M2 subgroup data to mature. The paper explicitly notes that “the number of patients with M2 occlusions in this trial was small — 18 (6.5%) in the TNK group and 20 (7.4%) in the thrombectomy-alone group — limiting the ability to draw definitive conclusions.” A 2024-era practice that uses TNK won’t update until TNK-specific bridging evidence emerges.

What the field does in 12 months: expect a meta-analysis pooling this with the ESCAPE-MeVO and DISTAL M2 subgroups, post-hoc analyses of completed thrombectomy trials extracting their M2 cohorts, and a surge in registry-based analyses from the same MR CLEAN network examining time-window stratification (early vs late, on-hours vs off-hours) — the paper already hints at this with its off-hours subgroup analysis, which showed bridging was numerically favorable across both windows with no significant interaction.

4. Asymmetric Leverage

Where is the asymmetric payoff?

The largest claim that, if true, gives the most leverage: bridging therapy should be the default for M2 occlusions presenting to mothership-capable centers. If this is real, it doesn’t just refine a clinical decision — it changes the architecture of prehospital triage. A mothership routing strategy (bypass primary stroke center → go directly to thrombectomy center) only makes economic and clinical sense if the time saved is spent on thrombectomy, not on administering tPA in the back of an ambulance or in the referring ED. If bridging is genuinely better, the cost-effectiveness of drip-and-ship collapses for M2 strokes, and mothership routing becomes more compelling, more defensible to payers, and more aligned with EMS protocols. The leverage compounds through system design.

The smallest claim that costs the most if wrong: that the ordinal mRS shift (aOR 1.52) reflects treatment effect rather than residual confounding. The downside is real: we would (1) justify continued alteplase use in M2 patients who could otherwise go directly to angio and save 30+ minutes, (2) entrench mothership-vs-drip-and-ship debates with a false data point, and (3) raise expectations among patients and families that bridging meaningfully changes outcome when, on every dichotomized endpoint, the difference is statistically indistinguishable from zero.

The asymmetry: upside is “tens of thousands of M2 stroke patients per year benefit from preserved bridging practice”; downside is “we lose a generation of patients to a statistical artifact while we wait for the RCT.”

5. Paradigm Destroyer

What does this paper kill?

Honestly, not much. The closest it comes is to push back against the rising “direct EVT, skip the tPA” reflex that the IRIS meta-analysis and the BRIDGE-TNK thrombectomy-alone arm have been encouraging. The paper’s strongest framing — that bridging yields “superior functional outcomes” — does not “destroy” a paradigm because the paradigm it challenges (direct EVT for medium vessel occlusion) is itself still being established.

What IS now untenable, or at least substantially weakened:

The claim that M2 occlusions can be treated identically to M1 occlusions in trial design without stratifying by bridging approach. The P interaction = 0.02 between mothership and transfer subgroups establishes that the bridging benefit is pathway-dependent, which means future M2 trials must pre-specify or stratify by drip-and-ship vs mothership.The implicit assumption in IRIS that bridging is equally marginal across all anterior circulation occlusions. The paper claims a more “pronounced difference” for M2 than IRIS found overall — 38.9% vs 29.7% achieving mRS 0–1 versus IRIS’s 1.7% absolute difference at mRS 0–2. If real, this means M2 occlusions are a distinct biological or therapeutic subgroup where lytics reach territory they cannot reach in larger clot burdens.The neutrality stance on alteplase in M2 patients without EVT contraindication. “Bridging therapy may yield superior functional outcomes” is the closest the field has to a recommendation in favor of alteplase for medium vessel occlusion.Nothing paradigm-destroying at the level of DISTAL-MeVO’s null result for distal EVT. The paper adjusts margins; it does not redraw them.

MVP — Minimum Viable Proof

The cheapest test that would change this paper’s evidentiary weight: a pre-specified, stratified subgroup analysis of M2 patients in BRIDGE-TNK (or any TNK bridging RCT), with at least 200 M2 patients per arm (target n=400). TNK’s faster onset, single bolus, and greater fibrin specificity mean that even a modest attenuation of the alteplase effect observed here would suggest bridging’s benefit is a property of the lytic, not of the procedural delay.

Cost: a few months of biostatistics work on already-collected trial data. No new recruitment, no new imaging, no new IRB. Just extract the M2 subset, run the ordinal mRS shift analysis with the same IPTW strategy the MR CLEAN team used, and report the interaction P.

If the TNK-M2 analysis is positive (aOR >1.5, P<0.05), the paper’s claim generalizes to modern practice and the case for bridging strengthens. If it is null or negative, the alteplase-specific MR CLEAN finding may reflect either the lytic agent’s pharmacology or an era effect (the MR CLEAN Registry enrolled 2014–2018, before modern stroke workflow optimization), and current practice should not be reformed on its basis.

Best Combination

The single move with the highest asymmetric payoff: mandate bridging (alteplase or TNK) for mothership M2 occlusions without IVT contraindication, while explicitly preserving drip-and-ship flexibility for transfer patients until RCT data matures.

This is the conclusion the data actually support — the mothership aOR was 1.79 (P=0.01), while the transfer aOR was 1.19 (P=0.62) — and it is the conclusion that converts the paper’s main result into an actionable clinical rule without overstating the transfer pathway. Mothership routing is the decision point where bridging-vs-direct is a real choice at the bedside (the patient hasn’t received IVT yet); the paper shows the benefit is concentrated there. For transfer patients, IVT is already a fait accompli by the time the mothership team gets the call, so the choice is moot. Mandating bridging selectively at the mothership node honors both the positive finding and the negative subgroup interaction, and it doesn’t require waiting for an RCT to act.

Overvalue Warning

Where this analysis is most likely to overstate the result, and biases to watch for:

Single-test significance inflation. The ordinal mRS shift is the only P<0.05 result in the paper. With four dichotomized mRS cutoffs, mortality, recanalization (3 tiers), and sICH, there are ≥9 secondary comparisons. The chance that one lands below alpha by chance in a 539-patient registry is non-trivial.Era effect. The MR CLEAN Registry enrolled from 2014–2018. Thrombectomy workflows, door-to-puncture times, aspiration-first adoption (which rose from 13% in 2014 to 36–37% by 2017–2018), and contrast agents have all evolved. The IVT+EVT benefit observed here may be partially explained by less efficient EVT in the bridging-eligible patients, not by IVT itself.Selection-bias asymmetry. “We must consider that patients who responded to IVT and did not require EVT were excluded from the analysis.” This bias runs in the opposite direction from what the conclusion implies — it should make the IVT+EVT arm look worse, not better, because fast IVT responders never reach the EVT cohort. That the arm still looks better is either a very strong signal or a residual confounding artifact. The paper doesn’t quantify how many IVT responders were excluded.Sample size in subgroups. The transfer subgroup analysis has substantially less power than the mothership subgroup. The 95% CI on the transfer aOR (0.60–2.34) is consistent with both substantial benefit and substantial harm from bridging. Treating P interaction = 0.02 as evidence of differential benefit requires acknowledging that the interaction is being driven by an imprecise transfer estimate.Author conflicts. Multiple senior authors hold industry relationships with Stryker, Medtronic, Penumbra-affiliated companies, and tenecteplase-distributing entities. While the funding source (Toegepast Wetenschappelijk Instituut voor Neuromodulatie) is non-conflicted, the disclosures signal that equipoise on this question is not uniform across the investigator group.Novelty/recency bias in citation framing. The Discussion leans on BRIDGE-TNK’s overall positive result (52.9% vs 44.1%, P=0.04) while noting BRIDGE-TNK had too few M2 patients (n=38) to be conclusive. A reader should not infer that BRIDGE-TNK supports this MR CLEAN M2 finding when BRIDGE-TNK explicitly does not have the M2 power to do so.Sample size overall is 539. Mechanism plausibility (microvascular reperfusion from IVT in M2 territory) is plausible but not directly measured. The paper is from a 2025 issue of Stroke and aligns with a broader narrative that mothership routing + bridging is the optimal M2 strategy — a narrative that IRIS and the BRIDGE-TNK interpretation were already eroding.