MRI DWI-FLAIR mismatch as a "tissue clock" — deciding thrombolysis eligibility when the true stroke onset time is unknown
Roughly one in four ischemic strokes is discovered when a patient wakes up with a new neurological deficit that was not present at bedtime. Because standard IV thrombolysis eligibility has historically depended on a clearly documented time-of-onset (≤4.5 hours), these patients were routinely excluded from treatment — not because their infarct was necessarily old, but because nobody could prove it was young. MRI-based tissue clocks were developed specifically to solve this documentation gap.
IV alteplase (recombinant tissue plasminogen activator, tPA) dissolves the occluding clot and can restore blood flow to still-salvageable brain tissue — the "ischemic penumbra." Its benefit is powerfully time-dependent:
• 0–3 hours: strongest benefit, lowest bleeding risk relative to benefit • 3–4.5 hours: extended window (ECASS III, 2008), narrower benefit margin, stricter exclusion criteria • >4.5 hours (clock-based): historically considered too risky — irreversibly infarcted tissue treated with a thrombolytic bleeds
The entire eligibility framework was built around clock time because, for decades, it was the only proxy available for "how much of the penumbra is still alive." The problem: clock time is a proxy, not the thing that actually matters. What actually matters is the biological age of the infarct.
Wake-up stroke is not a distinct stroke subtype — it is a subset of otherwise ordinary ischemic strokes distinguished only by the timing of discovery. Several factors make sleep a vulnerable window:
• Nocturnal blood pressure dipping and relative hypotension reduce collateral flow in patients with pre-existing stenosis • Increased vagal tone and reduced cardiac output overnight • Undiagnosed atrial fibrillation with early-morning embolic events • Circadian peak in platelet aggregability and blood viscosity in the early morning hours
Because the patient is asleep, the deficit is discovered anywhere from minutes to many hours after it actually began. The last-known-well time (typically "when they went to bed") is often 6–8 hours before discovery, but the true onset could have occurred at any point in that window — including just minutes before waking.
A stroke discovered at 7:00 AM in a patient who went to bed at 11:00 PM could be anywhere from a few minutes old to eight hours old. Clock time alone cannot distinguish a highly treatable hyperacute infarct from an already-completed one — only the tissue itself can.
Diffusion-weighted imaging (DWI) is the single most sensitive sequence for detecting acute ischemia. It exploits a simple biophysical fact: when neurons lose ATP-dependent ion pump function, water shifts from the extracellular to the intracellular compartment (cytotoxic edema), where its Brownian motion is more restricted. This restricted diffusion produces unmistakable bright signal on DWI — often within 10–30 minutes of vessel occlusion, long before any structural change is visible on conventional imaging.
DWI measures the random thermal (Brownian) motion of water molecules using paired diffusion-sensitizing gradients. In healthy tissue, water diffuses relatively freely through both intra- and extracellular space, producing signal loss on the diffusion-weighted image. In acutely ischemic tissue:
• Na+/K+-ATPase pump failure (energy depletion) allows Na+ and water to flood into cells • Cells swell (cytotoxic edema); the extracellular space shrinks • Water trapped inside cells and tortuous shrunken extracellular channels diffuses more slowly • Restricted diffusion → less signal loss from the diffusion gradient → the voxel appears BRIGHT on DWI
The apparent diffusion coefficient (ADC) map, calculated from multiple b-values, provides the quantitative counterpart: ADC drops by 30–50% in the ischemic core, confirmed as a dark region on the ADC map corresponding to the bright DWI lesion. This DWI-bright / ADC-dark signature is the classic signature of acute cytotoxic edema and is considered a highly reliable marker of the ischemic core.
DWI is exquisitely sensitive but poorly specific for lesion age. Once restricted diffusion appears, it typically persists for days: DWI lesions remain hyperintense from roughly 30 minutes post-occlusion out to 1–2 weeks (pseudonormalization occurs later, beyond the acute treatment window). This means:
• A DWI-positive lesion could be 20 minutes old or 20 hours old — DWI positivity by itself is essentially universal in acute stroke, regardless of exact onset time • DWI establishes THAT an infarct exists and roughly WHERE, but not WHEN it started • A second, independently time-sensitive sequence is required to estimate lesion age
This is precisely the gap that FLAIR imaging — acquired in the same MRI session — is used to fill.
FLAIR (Fluid-Attenuated Inversion Recovery) nulls the signal from free water (CSF) while remaining sensitive to tissue water content. Unlike DWI, FLAIR does not detect the earliest cytotoxic changes of ischemia — it responds to vasogenic edema and blood-brain-barrier breakdown, processes that accumulate gradually over hours. This built-in lag is what allows FLAIR, paired with DWI, to function as an imaging "stopwatch" for infarct age.
The FLAIR sequence combines T2-weighting with an inversion recovery pulse tuned to null the signal from free fluid (CSF). Visible FLAIR hyperintensity in ischemic tissue requires a later, slower pathological cascade than the one DWI detects:
• Cytotoxic edema (DWI-positive, minutes): intracellular water shift with little net increase in total tissue water — barely changes T2/FLAIR signal • Blood-brain-barrier breakdown (hours): tight junction failure allows protein-rich plasma to leak into the extracellular space • Vasogenic edema accumulation (hours): net increase in total tissue water content, now including the extracellular compartment • T2-prolongation effect: the accumulated water increases T2 relaxation time enough to become visible as hyperintensity even after CSF nulling
This pathophysiological delay is remarkably consistent across patients: pooled analysis of the WAKE-UP trial cohort and predecessor studies (Aoki et al., Petkova et al.) found the median time to visible FLAIR change is approximately 5–6 hours post-onset, with a sensitivity/specificity trade-off around the 4.5-hour treatment threshold.
In the pivotal validation cohorts, roughly 85–90% of infarcts imaged within 4.5 hours of a witnessed onset were FLAIR-negative — meaning FLAIR-negativity is a reasonably strong (though imperfect) proxy for "this infarct is young enough to treat."
FLAIR interpretation is not purely binary and requires attention to subtle findings:
• Subtle/parenchymal FLAIR hyperintensity: faint, ill-defined signal change corresponding to the DWI lesion counts as "FLAIR positive" even if less obvious than DWI • Sulcal FLAIR hyperintensity: hyperintense CSF in sulci overlying the ischemic territory (from slow flow / early leptomeningeal collateral congestion) is a distinct, sometimes earlier sign and is scored separately from parenchymal change • Old white matter disease: chronic small-vessel ischemic changes can obscure or mimic acute FLAIR signal — comparison with the DWI lesion location and shape is essential • Inter-rater reliability: visual FLAIR scoring has good but imperfect agreement (kappa ~0.6–0.8); quantitative signal-intensity ratio methods (relative FLAIR signal intensity, rFLAIR) are being developed to reduce subjectivity
The DWI-FLAIR mismatch concept is elegantly simple: overlay the DWI lesion and the FLAIR image of the same slice. If the DWI-positive region shows no corresponding FLAIR signal change, the biological lag between cytotoxic and vasogenic edema implies the infarct is very likely still within the treatable window — regardless of what the clock says. If FLAIR change has caught up to DWI ("mismatch negative" / "match"), the infarct has probably matured beyond 4.5 hours.
The WAKE-UP trial (and its predecessor observational studies) operationalized DWI-FLAIR mismatch with explicit imaging criteria, read centrally by trained neuroradiologists blinded to clinical data:
1. A clear, unambiguous acute ischemic lesion visible on DWI, corresponding to the clinical deficit 2. No, or only very subtle/ill-defined, parenchymal hyperintensity in the corresponding region on FLAIR 3. Reader confidence: visual assessment classified as clearly negative, subtle-but-positive, or clearly positive; only "no or subtle" qualifies as mismatch-positive for trial eligibility 4. Lesion must not be so large or so posterior-fossa-located that risk outweighs benefit under separate exclusion criteria
This is fundamentally a pattern-recognition task performed by comparing signal intensity, sharpness of margins, and homogeneity between the two co-registered sequences — increasingly assisted by quantitative software (e.g., automated FLAIR-DWI mismatch algorithms used in some stroke networks) to reduce inter-reader variability.
It is important to be precise about what DWI-FLAIR mismatch actually estimates:
• It is a probabilistic marker, not a precise timestamp — mismatch-positive infarcts are more likely, not certain, to be <4.5h old • Sensitivity (~60%) means some genuinely early infarcts will already show subtle FLAIR change and be misclassified as "too old" — these patients may be denied treatment unnecessarily • Specificity (~78%) means most mismatch-positive infarcts really are early, but a meaningful minority are false positives — genuinely older infarcts that happen to still look FLAIR-negative • Individual variability in edema kinetics (collateral status, glucose, temperature, infarct size) shifts the DWI-to-FLAIR transition time earlier or later for any given patient
Because of this imperfect but favorable trade-off, mismatch functions as a risk-stratification tool for a treatment decision under uncertainty — not a replacement for clinical judgment, and not a guarantee of safety or benefit for any individual patient.
Mismatch should be read together with lesion size, NIHSS severity, and vascular imaging (CTA/MRA) for large-vessel occlusion — a patient can be simultaneously "mismatch positive" for thrombolysis AND a thrombectomy candidate if a proximal occlusion is present.
The WAKE-UP trial (Thomalla et al., New England Journal of Medicine, 2018) provided the first randomized, placebo-controlled evidence that MRI-selected wake-up and unknown-onset stroke patients benefit from IV alteplase. It transformed DWI-FLAIR mismatch from a research concept into a guideline-endorsed treatment pathway, extending a proven therapy to a population previously excluded by default.
WAKE-UP randomized 503 patients with unknown-onset stroke (including classic wake-up strokes) who had a DWI lesion but no marked FLAIR change, to IV alteplase or placebo, in addition to standard care (including endovascular thrombectomy when indicated).
Key result: patients treated with alteplase had significantly better functional outcome at 90 days (modified Rankin Scale 0–1, i.e. no or minimal disability) — 53.3% versus 41.8% with placebo. This benefit is comparable in magnitude to that seen in classic time-window trials (e.g. NINDS, ECASS III), supporting the core hypothesis that tissue state — not clock time — determines who benefits from thrombolysis.
Safety: symptomatic intracranial hemorrhage occurred in 2.0% of the alteplase group versus 0.4% of placebo — a real but modest and expected increase consistent with thrombolysis risk in general, and it did not translate into a significant difference in mortality.
WAKE-UP directly changed international stroke guidelines: patients with unknown onset (including wake-up stroke) and a favorable DWI-FLAIR mismatch are now an accepted indication for IV thrombolysis in AHA/ASA and ESO guidance, using imaging rather than the clock as the entry criterion.
MISMATCH PRESENT (DWI+ / FLAIR− or only subtle change): • Proceed toward IV alteplase per standard contraindication screening (blood pressure, coagulation status, lesion size/extent, prior hemorrhage, anticoagulant use) • Simultaneously evaluate CTA/MRA for large-vessel occlusion — thrombectomy candidacy is assessed independently and can be pursued alongside or instead of alteplase • Time from imaging to needle should still be minimized ("imaging-to-needle" time) — mismatch removes the clock-time exclusion, it does not remove urgency
MISMATCH ABSENT (FLAIR positive / signal change matches DWI): • Infarct is probabilistically more likely mature (>4.5h) — IV alteplase carries an unfavorable risk-benefit balance and is generally not given on this basis alone • Manage with standard secondary-prevention pathway: antiplatelet therapy, risk-factor control, further workup for stroke etiology • Large-vessel occlusion should still be evaluated for mechanical thrombectomy — extended-window thrombectomy trials (DAWN, DEFUSE-3) use perfusion-core mismatch (not DWI-FLAIR mismatch) and can extend eligibility to 16–24 hours in carefully selected patients, independent of the FLAIR result
Indeterminate cases (subtle/equivocal FLAIR change) require multidisciplinary discussion, weighing NIHSS severity, lesion volume, and patient/family goals of care.