Cerebral Blood Flow, Oxygen Delivery, and Metabolism
The brain receives oxygen in flowing blood, extracts part of that oxygen, and consumes it through metabolism. These are linked processes with different measurements. A satisfactory arterial oxygen tension does not establish adequate oxygen content; adequate delivery does not establish uniform tissue availability; and a tissue oxygen tension does not directly measure ATP production. Understanding these distinctions makes cerebral monitoring more informative and prevents a favorable number from being mistaken for a complete physiological recovery.
60.1 Cerebral blood flow is a distribution of flows
Cerebral blood flow, or CBF, describes the volume of blood passing through brain tissue per unit time. Regional CBF is commonly expressed as mL of blood per 100 g of tissue per minute. Whole-brain flow is expressed as mL/min. Keeping these denominators visible matters: a small region can have severely impaired perfusion while contributing little to the whole-brain average.
For an illustrative brain mass of 1,400 g and spatially averaged flow of 50 mL/100 g/min, whole-brain flow would be 14 × 50 = 700 mL/min. These are constructed teaching values, not a universal physiological normal or a treatment target. Gray and white matter have different baseline metabolic activity and perfusion, and injured tissue adds further regional variation.
Blood reaches the cerebral circulation through the carotid and vertebrobasilar systems, distributes through large arteries and resistance vessels, exchanges gases in the microcirculation, and drains through cerebral veins and venous sinuses. Collateral connections can redistribute supply when a conduit artery becomes compromised. Their presence does not establish that collateral flow is sufficient under a particular pressure, transit time, or metabolic demand.
The classic nitrous oxide method applied conservation of an inert tracer to estimate human cerebral flow. Its dependence on arterial input, cerebral venous output, and tissue equilibration established a lasting principle: cerebral flow is inferred through a measurement model with anatomical and temporal assumptions. Kety and Schmidt explicitly considered how representative jugular blood was of cerebral venous drainage. 1
Cerebral perfusion pressure contributes to the driving conditions for flow, but pressure alone cannot specify flow through a changing vascular network. Vascular tone, downstream conditions, blood properties, and local obstruction also matter. Chapter 59 develops that pressure–flow relationship; this chapter follows what the flowing blood carries and what the tissue does with it.
60.2 Oxygen tension, saturation, and content answer different questions
Arterial oxygen tension, PaO₂, describes the pressure exerted by dissolved oxygen. Arterial hemoglobin saturation, SaO₂, describes the fraction of available oxygen-binding sites occupied by oxygen. Arterial oxygen content, CaO₂, describes the amount of oxygen in a volume of blood. Most arterial oxygen is normally carried on hemoglobin, so hemoglobin concentration is indispensable to the content calculation.
For a teaching estimate, take hemoglobin in g/dL, SaO₂ as a fraction, and PaO₂ in mmHg. Using an oxygen-binding coefficient of 1.34 mL O₂/g and a dissolved-oxygen coefficient of 0.003 mL O₂/dL/mmHg gives:
With hemoglobin 10 g/dL, SaO₂ 0.98, and PaO₂ 100 mmHg, bound oxygen is 1.34 × 10 × 0.98 = 13.132 mL/dL. Dissolved oxygen adds 0.300 mL/dL, giving approximately 13.43 mL/dL. At hemoglobin 7 g/dL with the same saturation and tension, content is 9.1924 + 0.300 = approximately 9.49 mL/dL. The reassuring saturation has remained unchanged while oxygen per unit blood volume has fallen substantially.
The coefficients are approximations, and dyshemoglobins require appropriate co-oximetry and content interpretation. Nevertheless, the causal lesson is robust: raising PaO₂ chiefly enlarges the dissolved component once hemoglobin is nearly saturated, whereas changing functional hemoglobin changes the much larger bound component. Neither observation by itself determines whether a transfusion or a change in inspired oxygen improves outcome.
Oxygen tension also provides the gradient for diffusion from capillary blood toward tissue. Content describes how much oxygen can be carried; tension helps describe its tendency to move. These complementary roles explain why neither can substitute for the other.
60.3 Delivery, extraction, and consumption obey conservation
Cerebral oxygen delivery is the oxygen carried into tissue per unit time. Cerebral metabolic rate of oxygen consumption, CMRO₂, is the oxygen removed from blood and consumed per unit time. Oxygen extraction fraction, OEF, is the proportion of delivered oxygen extracted. When CBF is expressed per 100 g of brain, delivery and consumption retain that same tissue denominator.
Let arterial and representative jugular venous oxygen contents, CaO₂ and CjvO₂, be expressed in mL O₂/dL. The factor of 100 below converts mL of blood to dL:
At CBF 50 mL/100 g/min and CaO₂ 13.432 mL/dL, delivery is 50 × 13.432/100 = 6.716 mL O₂/100 g/min. If consumption is 3.0 mL O₂/100 g/min, required extraction is 3.0/6.716 = approximately 0.447, or 44.7%. With unchanged flow and consumption but content 9.4924 mL/dL, delivery becomes 4.7462 and required extraction rises to approximately 63.2%.
This illustrates extraction reserve: tissue can sometimes maintain consumption despite declining delivery by removing a larger fraction. Reserve is finite and regionally variable. An extraction fraction below 100% is not automatically safe, and 100% is an algebraic boundary rather than a generally achievable physiological operating point.
The Fick interpretation requires sufficiently steady conditions, contemporaneous measurements, and venous blood representative of the tissue supplied by the measured flow. Oxygen storage changes and mismatched sampling undermine that interpretation. Regional positron emission tomography can estimate flow, extraction, and consumption within a different, explicitly modeled measurement framework. 2
High extraction can indicate threatened delivery relative to demand. Low extraction can reflect abundant supply, depressed demand, shunting of flow, or impaired utilization. The ratio needs its numerator, denominator, and clinical setting.
60.4 Neural activity recruits blood flow through neurovascular coupling
Neuronal signaling consumes energy through synaptic activity, ion transport, and restoration of transmembrane gradients. Activity also initiates signaling among neurons, astrocytes, vascular cells, and surrounding tissue that changes local vascular resistance. This neurovascular coupling normally directs additional blood toward active regions.
The vascular response is not a rigid oxygen replacement meter. Flow can increase more than oxygen consumption during activation, lowering extraction and increasing venous oxygenation despite increased metabolic work. In a small paired human positron-emission study, Fox and Raichle found that focal activation increased regional flow substantially more than oxygen metabolism. Their findings establish the possibility of physiological uncoupling; the reported proportional changes are not universal conversion factors. 3
This distinction helps explain the blood-oxygen-level-dependent magnetic resonance signal. Changes in deoxyhemoglobin depend on the interaction of flow, blood volume, and oxygen use. A stronger signal is not a direct measurement of increased neuronal oxygen consumption. Conversely, an abnormal response can arise from disturbed vascular signaling even when some neural activity remains.
In critical illness, contusion, ischemia, inflammation, endothelial injury, seizures, and altered arterial gases can change the coupling between activity and perfusion. Sedation may reduce activity and demand while also changing blood pressure or vascular tone. Therefore, a change in CBF after an intervention cannot automatically be assigned to altered metabolic demand.
The operating question is whether supply responds appropriately to the demands of a particular region. An apparently normal global flow may be inadequate for a seizing cortex; a lower flow may accompany appropriate demand reduction in another setting. Neither statement permits the inference that all low flow is benign or all high flow protective.
60.5 Carbon dioxide and oxygen change vascular conditions
CO₂ readily influences the brain’s acid–base environment. Increased arterial CO₂ generally favors cerebral vasodilation, whereas a fall favors vasoconstriction. Consequently, changing ventilation can alter cerebral flow even when systemic oxygen saturation remains excellent. Classic human experiments demonstrated major flow responses to altered inspired CO₂ without a corresponding requirement for increased cerebral oxygen consumption. 4
A fall in PaCO₂ can reduce cerebral blood volume and thereby influence intracranial pressure, yet simultaneously reduce oxygen delivery if flow falls. A favorable pressure response therefore does not prove a favorable regional oxygen balance. With rising PaCO₂, the potential flow benefit must likewise be considered alongside blood-volume, intracranial-pressure, systemic, and regional consequences.
There is no fixed percentage change in CBF per mmHg of PaCO₂ that applies across all patients, baseline tensions, time courses, and injuries. Reactivity can differ between regions and evolve over time. Prolonged changes also alter acid–base adaptation, so an acute challenge cannot simply be extrapolated indefinitely.
Arterial hypoxemia can recruit vasodilation, but that compensatory response may be limited by vascular disease, injury, or unfavorable perfusion conditions. Hyperoxia raises dissolved oxygen and capillary oxygen tension while potentially producing vasoconstriction; Kety and Schmidt observed reduced flow during high-oxygen breathing in healthy subjects. 4
Thus, a higher tissue oxygen reading after increasing inspired oxygen can arise from a stronger diffusion gradient without demonstrating increased flow or increased mitochondrial oxygen consumption. The net delivery response depends on both content and actual flow. These mechanisms explain competing effects; they do not identify a universal ventilation or oxygen target.
60.6 Metabolic demand depends on activity, substrates, and temperature
ATP production supports ion gradients, neurotransmission, cellular maintenance, and repair. Oxygen consumption reflects oxidative metabolism, while glucose uptake includes pathways that need not immediately consume oxygen. Glucose delivery, glycolysis, lactate exchange, and mitochondrial oxidation therefore cannot be inferred from one another through a fixed ratio in every condition.
Lactate can be produced while oxygen is available and can also be taken up and oxidized by the brain. In a small human tracer study, van Hall and colleagues demonstrated uptake and oxidation of circulating lactate under the studied resting, infusion, and exercise conditions. This establishes a substrate role, not a recommendation to reproduce those interventions in brain-injured patients. 5
An elevated cerebral extracellular lactate concentration can consequently reflect increased glycolysis, altered delivery or removal, substrate exchange, or impaired oxidative handling. Pairing lactate with pyruvate, glucose, oxygenation, and the physiological trajectory is more informative than labeling all cerebral lactate as anaerobic waste.
Seizures can abruptly increase energy requirements. If the vascular and substrate responses cannot meet that requirement, metabolic stress may accompany electrical activity. In a monitored traumatic-brain-injury cohort, electrographic seizures were associated with increased intracranial pressure and elevated extracellular lactate-to-pyruvate ratios. The observational design does not establish that any particular seizure-prevention strategy improves recovery. 6
Temperature and drugs also alter demand. Cooling may lower metabolic rate, while fever can raise requirements; neither supplies a universal quantitative correction for injured tissue. Sedative effects depend on agent, dose, electrical state, circulation, and temperature. Reducing measured oxygen consumption can represent useful suppression of excessive activity, but it can also accompany loss of viable or functioning tissue. Metabolic suppression and neurological recovery are different endpoints.
60.7 Regional oxygen failure can be hidden within a global average
Between an arterial sample and mitochondrial oxygen use lie regional arteries, capillary flow paths, interstitial diffusion, cell membranes, and cellular metabolic machinery. Total delivery can be adequate while one region receives insufficient effective supply. Conversely, low regional flow can accompany depressed metabolism without the same supply–demand imbalance as active ischemia.
Microvascular obstruction, heterogeneous capillary transit, and altered diffusion distances can impair extraction from otherwise oxygenated blood. Edema may alter the path between flowing erythrocytes and cells, but its effect depends on local structure and perfusion rather than tissue water alone. A high venous oxygen saturation can therefore coexist with a threatened region whose oxygen is not adequately extracted or whose low flow contributes little to the sampled venous mixture.
There is also a distinction between inadequate oxygen availability and impaired use of available oxygen. Mitochondrial dysfunction is biologically plausible after severe injury, but bedside patterns rarely isolate it from substrate limitation, cellular loss, diffusion barriers, or measurement location.
In a combined microdialysis and positron-emission study of patients with traumatic brain injury, Vespa and colleagues found metabolic-crisis patterns that often occurred without PET-defined ischemia. This demonstrates that an elevated lactate-to-pyruvate ratio need not be synonymous with the study’s ischemic PET pattern. It does not prove that all tissue was adequately oxygenated at every location or that mitochondrial failure was the sole mechanism. 7
The consequences are practical. A globally reassuring measurement should be tested against lesion location, vascular territories, local oxygen and metabolic signals, and electrical activity. Discordance can reveal heterogeneity rather than identify a defective monitor. It can also arise from measurements made at different times or tissue sites.
60.8 Flow measurements depend on tracers, transit, and vessel geometry
Different technologies estimate different parts of the cerebral circulation. Oxygen-15 positron emission tomography can combine tracer models to estimate regional flow, oxygen extraction, and CMRO₂. This provides a powerful physiological comparison, but requires specialized acquisition and modeling and is not continuous bedside surveillance. Spatial resolution and changing conditions still constrain interpretation. 2
Arterial spin labeling uses magnetically labeled arterial water as an endogenous tracer. The signal depends on delivery of that labeled water to the sampled tissue. When arterial arrival is delayed, the chosen acquisition timing can underestimate tissue perfusion or leave label in arteries. A clinical study in patients with unilateral arterial stenosis showed how multiple inflow times could help address prolonged transit, while retaining method-dependent artifacts and imperfect agreement with a contrast-based comparator. 8
CT and contrast-enhanced magnetic resonance perfusion also use the passage of an indicator and mathematical models. Arrival delay, arterial input selection, collateral pathways, tissue coverage, and processing influence the maps. A perfusion map is not a direct map of oxygen consumption or of irreversible tissue injury. Its value comes from interpreting the measured pattern with anatomy, timing, and the method’s assumptions.
Transcranial Doppler measures blood velocity in an insonated vessel. Volumetric flow equals mean velocity multiplied by cross-sectional area; velocity alone tracks flow only to the extent that geometry and measurement conditions permit. In a human study using magnetic resonance measurements during CO₂ changes, middle cerebral artery area changed, and velocity underestimated changes in calculated flow. 9
This does not make Doppler uninformative. It makes its signal specific: velocity and waveform changes must be interpreted alongside vessel caliber, insonation, systemic circulation, and the suspected process. Comparing serial measurements requires comparable technique as well as comparable physiology.
60.9 Jugular oxygen and tissue oxygen sample different compartments
Jugular bulb oxygen saturation reflects oxygen remaining in sampled cerebral venous outflow. With appropriate catheter position and representative drainage, it helps assess the aggregate relation between supply and extraction. It is not mixed systemic venous saturation, and the two cannot be substituted in the cerebral Fick equation.
Venous mixing is weighted by blood flow. A well-perfused region can dominate the sample while a severely underperfused region contributes little blood. Drainage asymmetry, catheter location, extracranial contamination, and changing arterial content further affect interpretation. High jugular saturation can reflect high flow relative to demand, reduced metabolism, or impaired extraction; low saturation identifies increased aggregate extraction but does not locate its cause.
Brain tissue oxygen tension, PbtO₂, is a local partial pressure measured near an implanted probe. It reflects the balance among oxygen arriving in local microvessels, diffusion toward the sensor, and tissue consumption. Probe location in relatively preserved tissue, near a lesion, or within severely damaged tissue changes the biological question being sampled.
In a physiological study of traumatic brain injury, Rosenthal and colleagues examined local flow, arterial and jugular measurements, and PbtO₂ during pressure, oxygen, and CO₂ challenges. Their findings emphasized diffusion-related influences on PbtO₂ rather than its equivalence to global oxygen delivery or metabolism. These were contextual physiological relationships, not a universal equation for converting probe pressure into oxygen consumption. 10
A rise in PbtO₂ may result from increased arterial tension, improved local perfusion, reduced demand, or a combination. A persistently low value requires examination of signal validity and anatomy as well as delivery. Neither a single local probe nor one global venous sample can certify oxygen sufficiency throughout the brain.
60.10 Metabolic, electrical, and optical signals add context
Cerebral microdialysis samples extracellular solutes near a catheter over collection intervals. Glucose, lactate, pyruvate, and their trajectories provide local metabolic information, but concentrations depend on production, use, transport, and sampling recovery. The lactate-to-pyruvate ratio is related to cellular redox conditions through lactate dehydrogenase chemistry; an extracellular sampled ratio is not a direct assay of every cell’s cytosolic redox state.
A rising ratio with low pyruvate and poor oxygen availability can support a different mechanistic interpretation from a rising ratio with preserved pyruvate and no corroborating supply deficit. Neither pattern alone proves a single diagnosis. Local concentration measurements also do not provide a whole-brain substrate flux without additional flow and exchange information.
Electroencephalography contributes another domain: electrical activity. Seizures may reveal increased demand that explains worsening metabolic balance. Suppression can accompany sedation, injury, temperature change, or other causes. EEG is not a direct calorimeter, and absence of seizures does not establish adequate oxygen delivery.
Near-infrared cerebral oximetry estimates a regional optical oxygenation signal containing contributions from vascular compartments and overlying tissues. Its arterial–venous weighting, depth sensitivity, and processing depend on the device and conditions. In a healthy-volunteer experiment, scalp ischemia and hypoxia changed readings from three cerebral oximetry systems, demonstrating extracranial influence with different magnitudes between technologies. 11
Interpretation becomes stronger when independent domains tell a coherent story: evolving electrical activity, local substrate stress, falling tissue oxygen, and a compatible perfusion change. Apparent corroboration is weaker when several displayed variables are derived from the same underlying signal. Discordance should prompt examination of location, timing, and measurement assumptions before forcing every observation into one diagnosis.
60.11 Physiological improvement and clinical benefit require separate evidence
A monitor can identify a physiological problem without proving that its associated treatment strategy improves recovery. In the 119-patient BOOST-II phase II trial, management informed by PbtO₂ and intracranial pressure reduced brain hypoxia burden compared with pressure-guided management alone. The trial was not powered to establish neurological efficacy. 12
OXY-TC randomized 318 patients with severe traumatic brain injury; its primary modified intention-to-treat analysis included 291. Adding PbtO₂-guided management did not reduce poor six-month neurological outcome: 51% with intracranial-pressure monitoring alone versus 52% with dual monitoring. Catheter dysfunction and catheter-related intracerebral hematoma were more frequent with dual monitoring. This is a finding about the tested strategy and population, not proof that all monitoring approaches are equivalent. 13
Hemoglobin illustrates the same distinction. Its effect on content is predictable, but transfusion also changes blood properties and exposes patients to other consequences. HEMOTION randomized 742 adults with traumatic brain injury and anemia to strategies using transfusion thresholds of 10 versus 7 g/dL. The primary analysis did not establish a statistically significant reduction in unfavorable six-month neurological outcome with the liberal strategy; the confidence interval also did not establish equivalence. 14
TRAIN randomized 850 patients with traumatic brain injury, subarachnoid hemorrhage, or intracerebral hemorrhage. Its liberal strategy, using a 9 rather than 7 g/dL threshold, reduced unfavorable 180-day neurological outcome in the primary analysis. Population, thresholds, and outcome definitions differed from HEMOTION. 15
SAHARA subsequently reported 742 randomized patients with aneurysmal subarachnoid hemorrhage and anemia. Mandatory transfusion at hemoglobin ≤10 g/dL did not significantly reduce unfavorable twelve-month neurological outcome compared with a strategy allowing optional transfusion at ≤8 g/dL. Its comparator was therefore different from a mandatory 7 g/dL trigger, and its result should retain that distinction. 16
These trial thresholds describe tested interventions, not a single hemoglobin prescription for every neurological illness. Physiology identifies plausible benefits and competing effects; comparative trials estimate the resulting clinical balance in specified patients.
60.12 An integrated case tests the whole oxygen balance
Consider a constructed serial case using the same units throughout. Initially, hemoglobin is 10 g/dL, SaO₂ 0.98, PaO₂ 100 mmHg, and measured average CBF 50 mL/100 g/min. Estimated content is 13.432 mL/dL and delivery 6.716 mL O₂/100 g/min. Assume consumption is independently estimated at 3.0 mL O₂/100 g/min, requiring extraction of approximately 44.7%.
After bleeding or dilution, hemoglobin becomes 7 g/dL while arterial gases remain unchanged. If flow and consumption stayed fixed, delivery would fall to 4.7462 and required extraction would rise to approximately 63.2%. Maintaining saturation has not maintained delivery. If flow instead rose to 70 mL/100 g/min, calculated delivery would be 70 × 9.4924/100 = 6.64468, almost restoring the initial value. This is a conditional calculation, not a prediction that the brain will produce that response or a proposed flow target.
Now suppose actual average flow falls to 40 while a demand assessment suggests that 4.0 mL O₂/100 g/min would be required to sustain the current activity. Delivery is 40 × 9.4924/100 = 3.79696. The required extraction would be 4.0/3.79696 = approximately 1.053, or 105.3%, which is impossible. These assumptions cannot all coexist with sustained consumption at the stated requirement. Supply must increase, actual consumption must fall, or the measurements and demand estimate do not represent the same tissue and time.
A normal global jugular saturation would not resolve the contradiction if threatened tissue contributed little venous outflow. A higher local PbtO₂ after increasing inspired oxygen would show a local tension response, while flow, consumption, and regional vulnerability would still require reassessment. Seizures, temperature, arterial CO₂, systemic circulation, and measurement location supply possible causal explanations rather than automatic interventions.
The useful endpoint of interpretation is a coherent, revisable account of supply, extraction, demand, and tissue location. Cerebral physiology becomes actionable when each observation is assigned the question it can answer.
References
1. Kety SS, Schmidt CF. The nitrous oxide method for the quantitative determination of cerebral blood flow in man: theory, procedure and normal values. J Clin Invest. 1948;27:476–483. Original article.
2. Frackowiak RS, Lenzi GL, Jones T, Heather JD. Quantitative measurement of regional cerebral blood flow and oxygen metabolism in man using ¹⁵O and positron emission tomography: theory, procedure, and normal values. J Comput Assist Tomogr. 1980;4:727–736. Original article.
3. Fox PT, Raichle ME. Focal physiological uncoupling of cerebral blood flow and oxidative metabolism during somatosensory stimulation in human subjects. Proc Natl Acad Sci USA. 1986;83:1140–1144. Original article.
4. Kety SS, Schmidt CF. The effects of altered arterial tensions of carbon dioxide and oxygen on cerebral blood flow and cerebral oxygen consumption of normal young men. J Clin Invest. 1948;27:484–492. Original article.
5. van Hall G, Strømstad M, Rasmussen P, et al. Blood lactate is an important energy source for the human brain. J Cereb Blood Flow Metab. 2009;29:1121–1129. Original article.
6. Vespa PM, Miller C, McArthur D, et al. Nonconvulsive electrographic seizures after traumatic brain injury result in a delayed, prolonged increase in intracranial pressure and metabolic crisis. Crit Care Med. 2007;35:2830–2836. Original indexed report.
7. Vespa P, Bergsneider M, Hattori N, et al. Metabolic crisis without brain ischemia is common after traumatic brain injury: a combined microdialysis and positron emission tomography study. J Cereb Blood Flow Metab. 2005;25:763–774. Original article.
8. Martin SZ, Madai VI, von Samson-Himmelstjerna FC, et al. 3D GRASE pulsed arterial spin labeling at multiple inflow times in patients with long arterial transit times: comparison with dynamic susceptibility-weighted contrast-enhanced MRI at 3 Tesla. J Cereb Blood Flow Metab. 2015;35:392–401. Original article.
9. Coverdale NS, Gati JS, Opalevych O, Perrotta A, Shoemaker JK. Cerebral blood flow velocity underestimates cerebral blood flow during modest hypercapnia and hypocapnia. J Appl Physiol. 2014;117:1090–1096. Original article.
10. Rosenthal G, Hemphill JC III, Sorani M, et al. Brain tissue oxygen tension is more indicative of oxygen diffusion than oxygen delivery and metabolism in patients with traumatic brain injury. Crit Care Med. 2008;36:1917–1924. Original article.
11. Davie SN, Grocott HP. Impact of extracranial contamination on regional cerebral oxygen saturation: a comparison of three cerebral oximetry technologies. Anesthesiology. 2012;116:834–840. Original article.
12. Okonkwo DO, Shutter LA, Moore C, et al. Brain oxygen optimization in severe traumatic brain injury phase-II: a phase II randomized trial. Crit Care Med. 2017;45:1907–1914. Original article.
13. Payen JF, Launey Y, Chabanne R, et al. Intracranial pressure monitoring with and without brain tissue oxygen pressure monitoring for severe traumatic brain injury in France (OXY-TC): an open-label, randomised controlled superiority trial. Lancet Neurol. 2023;22:1005–1014. Original article.
14. Turgeon AF, Fergusson DA, Clayton L, et al. Liberal or restrictive transfusion strategy in patients with traumatic brain injury. N Engl J Med. 2024;391:722–735. Original article.
15. Taccone FS, Rynkowski CB, Møller K, et al. Restrictive vs liberal transfusion strategy in patients with acute brain injury: the TRAIN randomized clinical trial. JAMA. 2024;332:1623–1633. Original article.
16. English SW, Delaney A, Fergusson DA, et al. Liberal or restrictive transfusion strategy in aneurysmal subarachnoid hemorrhage. N Engl J Med. 2025;392:1079–1088. Original article.