The first CT in intracerebral hemorrhage is a snapshot of a process that may still be unfolding. A patient can arrive with a seemingly manageable hematoma and deteriorate while we arrange an ICU bed, clarify the admitting service, or wait for a transfer decision. My approach is to treat those first hours as a shared emergency: establish the diagnosis, control the conditions that favor continued bleeding, reverse clinically relevant anticoagulation, identify the underlying cause, and bring the medical and surgical teams into the same conversation early. The practical challenge is making those actions happen together.

Time matters because hematoma expansion is concentrated early. The lecture that prompted this discussion uses a rapidly declining bleeding curve: some hemorrhages achieve hemostasis within minutes, others continue growing through the first hour, and progressively fewer remain actively expanding over the next several hours. Its estimates—roughly 40% stopping within 15–20 minutes, expansion being essentially complete within an hour in about half, and only 10–15% still expanding around three hours—are teaching approximations, not reliable probabilities for an individual patient. The useful principle is that the opportunity to prevent additional injury shrinks quickly. Anticoagulant exposure, uncertain onset, and imaging evidence of active bleeding make assumptions about completed hemostasis especially dangerous.

FAST-MAG illustrates how much can change before definitive hospital care. In its exploratory analysis, a decline of at least two Glasgow Coma Scale points between the initial prehospital assessment and early postarrival assessment occurred in 30.8% of patients with intracranial hemorrhage, compared with 6.1% with cerebral ischemia. That observation supports urgency, but declining consciousness does not prove that the hematoma expanded. Hydrocephalus, seizures, aspiration, and other evolving complications can also change the examination. Repeated assessment must accompany rapid treatment.

The background disease also matters. ICH has historically had fewer proven treatment options than ischemic stroke, but therapeutic pessimism is increasingly difficult to justify. Hypertension, cerebral small vessel disease, cerebral amyloid angiopathy, anticoagulation, and structural vascular lesions create different patterns of bleeding and different risks. U.S. population data show higher ICH hospitalization incidence in Black and Hispanic populations than in White populations; those findings should not be recast as a single worldwide racial comparison or an intrinsic biological explanation. Prevention, blood pressure burden, and access to care belong in that discussion.

Lobar and deep hemorrhages should therefore be distinguished from the beginning. Deep hemorrhage commonly raises concern for arteriolosclerotic small vessel disease, whereas lobar hemorrhage broadens consideration of amyloid angiopathy and structural causes. Location is an important clue rather than a diagnosis. It also affects surgical access, the consequences of tissue injury along the operative corridor, and how closely a patient resembles the populations enrolled in recent evacuation trials.

The first treatment decision may occur before CT is available. A mobile stroke unit can establish hemorrhage in the field and allow treatment to follow the actual diagnosis, but these units remain resource intensive. Blood biomarkers could make prehospital differentiation more scalable. GFAP is promising because astroglial injury can produce an early signal in ICH, and the prospective DETECT study evaluated measurement on a point-of-care device. However, diagnostic accuracy varies with patient characteristics and assay thresholds. A negative result cannot safely establish ischemic stroke, and biomarker performance does not establish that a biomarker-directed treatment pathway improves outcomes. GFAP remains an emerging diagnostic strategy whose clinical implementation requires further validation.

INTERACT4 explains why this distinction has clinical consequences. Conducted in China, it initiated ambulance-based treatment with urapidil within two hours of suspected stroke onset, with a median onset-to-randomization time of 61 minutes. Mean systolic pressure on hospital arrival was 159 mmHg with intervention and 170 mmHg with usual care. The overall trial was neutral, but effects differed by the subsequently established stroke type: functional outcomes favored treatment in hemorrhagic stroke and favored usual care in cerebral ischemia. The finding supports rapid diagnosis and early treatment of confirmed hemorrhage; it does not support intensive pressure reduction in every patient with an undifferentiated stroke syndrome.

The physiologic explanation is plausible. Lower arterial pressure may reduce continued bleeding from an injured vessel, while ischemic tissue with impaired autoregulation may depend on systemic pressure to sustain collateral flow. Nevertheless, plausibility alone cannot define the treatment target. The ESO blood pressure guideline update published in 2026 advises against routine prehospital blood pressure lowering in suspected stroke. Stroke subtype, reperfusion status, baseline pressure, and competing indications must remain part of the decision.

The hospital response should preserve the same urgency. Two hours from onset is a treatment opportunity, not an acceptable door-to-CT interval. Imaging should occur immediately when feasible, with medication history, blood sampling, monitoring, and preparation for treatment proceeding in parallel. The ATACH-2 analysis cited in the lecture examined patients receiving intravenous nicardipine within two hours, rather than simply patients randomized within that interval. Intensive treatment in that subgroup was associated with less hematoma growth and better functional outcomes, but this was an exploratory post hoc analysis of a trial whose primary result was neutral. It strengthens a timing hypothesis without converting it into definitive proof.

A complementary body of evidence comes from INTERACT3, a cluster-randomized care-bundle trial. It combined early blood pressure control with management of glucose, temperature, and warfarin-associated anticoagulation. Functional outcomes improved with the bundle. Its design does not isolate the contribution of blood pressure lowering, and achieving the target within an hour of treatment initiation is different from treating within an hour of symptom onset. The broader lesson is that organized, prompt delivery of several treatments can matter more than selecting one intervention in isolation.

Moderation cannot be defined by a universal 30–45 mmHg systolic reduction. A starting pressure of 170 and a starting pressure above 220 require different consideration. Moving directly from above 220 to 120 mmHg warrants particular concern about overshoot. Chronic hypertension may shift the autoregulatory operating range toward higher pressures, while elevated intracranial pressure reduces the margin for maintaining cerebral perfusion: CPP is approximately MAP minus ICP. A large, abrupt fall can therefore be hazardous. However, there is no validated rule that every reduction of 60 mmHg or more causes watershed infarction, and a right-shifted curve should not be presumed identical in every hypertensive patient.

The evidence requires a more precise distinction between avoiding overshoot and withholding appropriate treatment. ATACH-2 did not show improved death or disability with its intensive 110–139 mmHg target and reported more renal adverse events. Conversely, ICHADAPT-2, published in 2025, did not demonstrate an increase in diffusion-restricted lesions with a target below 140 compared with below 180 mmHg. Only 79 of 162 randomized patients underwent the primary MRI assessment, limiting precision and generalizability, particularly for severe hemorrhages. Internal watershed infarcts, sometimes appearing in a string-of-pearls pattern, can occur during hypoperfusion, but their appearance alone does not establish an iatrogenic cause. Bilateral watershed injury can produce the man-in-the-barrel phenotype of disproportionate arm weakness. The bedside response is to assess pressure trajectory, examination, renal function, and perfusion context together.

Vascular imaging adds two different kinds of information: why the patient bled and whether further bleeding is likely. Broad use of acute CTA can be a useful institutional strategy, although a mandate for every ICH is stronger than the guideline recommendation. AHA/ASA recommendations specifically prioritize etiologic angiography in defined age, location, and hypertension-history groups, while early CTA can also help assess expansion risk. Macrovascular diagnostic yield varies substantially with the population; a 10–15% figure should not be applied to every patient. CTA can identify an AVM or suggest a dural fistula, but a negative examination does not exclude every structural lesion.

This information is particularly important before planned evacuation. Removing a hematoma associated with an unrecognized vascular lesion creates different operative risks from evacuating a primary small vessel hemorrhage. Suspicion may justify venous imaging, MRI, or catheter angiography, depending on the pattern and clinical circumstances. At the same time, additional imaging must be balanced against an immediate need for lifesaving decompression. An efficient protocol obtains useful vascular information early without allowing the diagnostic workup to become a new source of delay.

The CTA spot sign is evidence of contrast within the hematoma consistent with active extravasation. PREDICT prospectively established its association with subsequent expansion and outcome. It is among the most useful expansion markers, but its sensitivity is incomplete and its predictive value depends on timing and acquisition technique. A several-fold association with expansion should not be converted into one universally applicable risk multiplier. Expansion definitions also differ: commonly used thresholds include at least 6 mL or 33% growth, with approximately 12.5 mL or two-thirds growth used in some analyses of severe expansion. A negative spot sign does not establish that bleeding has stopped.

Spot positivity also informs the surgical conversation. Observational data associate it with active intraoperative bleeding and postoperative rebleeding. A small spot-positive hemorrhage may justify closer monitoring, earlier repeat imaging, and earlier discussion with neurosurgery than its initial volume alone would suggest. It does not independently establish an indication for evacuation. The distinction is between recognizing a potentially unstable lesion and assuming that a particular treatment has been proven beneficial for that lesion.

Multiphase CTA makes the temporal component visible. Additional acquisitions after the first pass show whether a spot appears early, persists, enlarges, or disperses. Three phases over roughly 45 seconds illustrate this approach, although actual intervals depend on the protocol. The AJNR study by Horn and colleagues linked timing, spot volume, and leakage characteristics with hematoma growth. This is a promising way to move beyond a binary positive-or-negative label. It remains observational risk stratification, however, and does not yet define a universally validated protocol for choosing reversal agents, hemostatic drugs, or surgery.

The black-and-white sign combines the CTA spot sign with a co-localized hypodense region on noncontrast CT. The proposed explanation is ongoing leakage into a region containing relatively fresh or less coagulated blood; that remains a mechanistic interpretation rather than direct proof of clot composition. In the original 200-patient cohort, the sign appeared in 14 patients, or 7%, and was strongly associated with expansion. Subsequent multicenter PREDICT validation found it in 9.5%, with substantial expansion risk but lower predictive performance and only moderate agreement between readers. The marker is useful for identifying a high-risk phenotype; it is neither invariably present nor a guarantee of catastrophic growth.

Transport decisions should respond to this evolving physiology. In RACECAT’s secondary analysis of 302 patients ultimately diagnosed with ICH, bypassing the closest stroke center for an endovascular-capable center was associated with worse functional outcomes. Vomiting during transport occurred in 12.8% versus 1.9%, and pneumonia through day five in 35.8% versus 17.6%. Neurological worsening during transport was also more frequent. Imaging occurred later in the bypass group, but the difference in substantial hematoma enlargement was not statistically significant. It would therefore overstate the results to say that RACECAT proved longer transport causes greater expansion.

The study supports concern about delaying diagnosis and stabilization, while the proposed contribution of vehicle motion remains a hypothesis. Its nonurban transport system and ambulances unable to provide intravenous treatment also limit direct application to every modern transfer pathway. The nearest capable hospital may offer faster airway management, blood pressure control, and reversal, but urgent neurosurgical needs and local capability can favor direct transport to a specialist center. The relevant comparison is time to effective care. Neither universal bypass nor mandatory local admission follows from this trial.

Anticoagulant-associated ICH makes coordinated care particularly urgent. The medication, last dose, kidney function, and likelihood of clinically relevant anticoagulant activity need rapid clarification. Reversal should not be postponed while services negotiate responsibility. The same information must accompany the patient during transfer, including the agent given, dose, bolus time, infusion completion, and subsequent examination and imaging. A transfer note that merely says anticoagulation was reversed can conceal clinically important uncertainty.

Andexanet illustrates why a better hemostatic result does not automatically mean a better overall treatment. In ANNEXA-I, hemostatic efficacy occurred in 67.0% with andexanet and 53.1% with usual care; most usual-care patients received PCC. Thrombotic events, particularly ischemic stroke, were more frequent with andexanet, and the trial did not demonstrate better 30-day functional outcome or survival. Describing it simply as superior to PCC loses both the endpoint distinction and the nature of the comparator. It improved control of bleeding under the trial’s conditions, with a clinically important competing risk.

A selective-use argument focuses on patients presenting very early with sizable hemorrhages, spot positivity, or a high estimated pre-scan growth rate. Dividing hematoma volume by elapsed time can identify a rapidly accumulated burden, but it is an average proxy rather than a measurement of the current bleeding rate. A 20–30 mL hematoma within the first couple of hours is an example of potential risk enrichment; validated selection criteria proving net clinical benefit from andexanet remain lacking. Likewise, selective use may reduce expenditure without establishing cost-effectiveness. Functional outcomes, thrombotic complications, drug cost, and health-system context all belong in that analysis.

Duration of reversal is another important qualification. Andexanet acts as a factor Xa inhibitor-binding decoy; it does not eliminate the anticoagulant. Its prescribing information describes a marked reduction in anti-Xa activity during the bolus and two-hour infusion, followed by re-elevation, with activity returning toward placebo levels approximately two hours after the infusion ends. Reversal can therefore wane during a long transfer, but five hours is not a universally valid cutoff for complete disappearance of all effect. Laboratory anticoagulant activity, clinical hemostasis, and operative safety are related but different measurements.

The broader hemostatic story has also moved forward. FASTEST, led by Joseph Broderick, ultimately included 626 participants treated within two hours, and its results were published in 2026. Recombinant activated factor VII reduced hematoma growth but did not improve the primary 180-day functional outcome, and life-threatening thromboembolic complications were more frequent. Stopping additional bleeding is attractive, yet a smaller hematoma on follow-up imaging does not guarantee meaningful recovery. Treatment may arrive after irreversible injury, the absolute volume saved may be insufficient, or benefit may be offset by complications.

Further testing in patients with the greatest likelihood of continued bleeding remains reasonable. Biomarker enrichment may help identify a subgroup with more opportunity for benefit, but this remains a research hypothesis. Preoperative factor VIIa is another proposed research direction. A trial would need to measure operative bleeding, recurrent hemorrhage, thromboembolism, and functional recovery, while accounting for the quality of surgical hemostasis. The existing results support continued investigation rather than routine perioperative administration.

Early surgical involvement should be integrated with medical treatment. Joint assessment by stroke neurology, neurocritical care, and neurosurgery can establish a shared plan at admission. One model described in the referenced discussion uses early helicopter activation and coordinated evaluation on arrival in Cleveland. Those logistics are local practice, but the underlying principle is broadly useful: agree early on stabilization, repeat imaging, transfer urgency, rescue criteria, and the opportunity for planned evacuation. Protocols should make responsibility explicit so that multidisciplinary involvement accelerates decisions instead of creating another handoff.

A volume-only consultation rule cannot capture that complexity. Familiar 10-, 20-, and 30-mL thresholds and much larger 100-, 150-, and 200-mL hemorrhages do not yield a universally appropriate consultation ceiling. Abrupt deterioration as a hematoma grows from 30–40 to 150 mL creates a different decision from presentation with an already massive hemorrhage, but neither pattern determines futility by itself. The reported exceptional recovery after a 200-mL hemorrhage is an anecdote, not an outcome estimate. Examination, trajectory, location, ventricular obstruction, premorbid function, and the patient’s values must remain in the decision. Predefined rescue criteria are particularly useful overnight. Consultation, rescue surgery, and trial-supported elective evacuation are separate questions.

Historical reluctance to evacuate supratentorial ICH partly reflects trials such as STICH II, which did not establish a significant overall functional benefit from early open surgery in its selected superficial lobar population. Those results should neither be ignored nor assumed to settle the question for every newer surgical technique. The 2025 ESO/EANS guideline incorporates subsequent evidence; describing all contemporary guidance as merely outdated STICH-era volume criteria is inaccurate.

ENRICH provides the clearest reason to reconsider selected lobar hemorrhages. It studied early trans-sulcal parafascicular evacuation in patients with 30–80 mL lobar or anterior basal ganglia hemorrhages, with intervention within 24 hours. The overall functional benefit was attributable to the lobar group; enrollment of the anterior basal ganglia group stopped for futility. These findings support building an experienced program around patients resembling the trial population, with age, consciousness, premorbid function, anatomy, and timing considered together. They do not establish that every lobar hematoma, surgical device, or evacuation method benefits. A positive randomized trial should not be equated with a universal class I, level A recommendation for lobar surgery.

Deep hemorrhage remains a different problem. The operative corridor and vulnerability of eloquent pathways can change the balance between decompression and procedure-related injury. Innovation deserves continued study, with experienced surgeons, standardized techniques, and meaningful outcomes. The later MIND trial reinforces the need for caution: despite substantial clot reduction, minimally invasive evacuation within 72 hours did not significantly improve its primary 180-day disability outcome. MIND differed from ENRICH in technique, timing, and case mix and stopped early, so it does not negate ENRICH’s result. It does show why minimally invasive evacuation cannot be treated as a single uniformly effective intervention.

Residual volume is relevant but must be interpreted correctly. MISTIE III was neutral for its primary functional outcome; exploratory analyses linked more successful clot removal, including achieving a residual of 15 mL or less, with better outcomes. Such associations do not prove that an assigned residual target causes benefit. More ambitious 2–8 mL goals remain hypotheses and technical aspirations rather than established standards across techniques. Pursuing smaller residuals must also account for injury to surrounding tissue. A residual below 8 mL does not independently eliminate the need for ICP monitoring or ventricular drainage; hydrocephalus, edema, consciousness, and the broader pressure physiology still matter.

Amyloid angiopathy adds uncertainty rather than an automatic surgical prohibition. Strictly lobar microbleeds and cortical superficial siderosis on susceptibility-sensitive MRI can inform the likely mechanism and future bleeding risk. A high long-term recurrence risk does not demonstrate that operating on the current hemorrhage is futile or that immediate recurrence is inevitable. An analysis of ENRICH imaging for outcomes in patients with probable amyloid angiopathy would be useful, but its primary report does not resolve that subgroup question. Until applicable data are available, operative decisions should account for the present lesion and the patient’s overall situation rather than treating the suspected mechanism as sufficient by itself.

Postoperative blood pressure management returns us to the same physiologic tradeoff. Hemostasis, residual clot, edema, ICP, chronic pressure exposure, kidney function, and examination should inform an explicit target agreed upon by the surgical and critical care teams. Removing clot can relieve mass effect without removing every threat to perfusion. The literature supporting medical targets does not establish one universal postoperative target, a fixed 30–45 mmHg reduction, or a safe residual-volume cutoff. Excessive reduction and prolonged overshoot deserve particular attention, especially when the examination worsens during titration.

The practical sequence is therefore tightly connected: diagnose immediately, treat promptly after confirmation, reverse relevant anticoagulation, obtain appropriate vascular information, repeat the examination and imaging when the clinical course requires it, and resolve transfer and surgical questions together. Cheap prehospital detection, richer imaging markers, and more precise operative selection may improve this pathway, but each must demonstrate that its added information changes outcomes. The first two hours deserve concentrated effort because preventable injury can accumulate quickly. The patient still needs careful reassessment after that window, when bleeding, hydrocephalus, edema, perfusion, and the consequences of our own treatment continue to evolve.

ICH_First_Two_Hours_Post.md