Gastrointestinal, Hepatic, and Splanchnic Perfusion

The splanchnic circulation links several organs in parallel and then delivers much of their venous blood to the liver in series. Its function cannot be summarized by cardiac output, one venous saturation, or one liver enzyme. Understanding the network requires separating arterial supply, portal drainage, hepatic oxygen delivery, mucosal function, and the liver’s many metabolic and excretory tasks.

69.1 The splanchnic circulation is a connected network

The gastrointestinal tract, pancreas, spleen, and liver receive blood through distinct arterial territories. The celiac, superior mesenteric, and inferior mesenteric systems supply organs with different functions and metabolic demands. Venous drainage from much of this region enters the portal vein before returning to the systemic circulation through the liver.

The arrangement combines parallel and serial relationships. Different intestinal segments can receive different proportions of mesenteric flow. Their venous outflows then contribute to portal inflow, alongside blood from other abdominal organs. A change in one territory can alter portal volume, oxygen content, nutrient delivery, and signaling without producing an identical change everywhere else.

The liver receives an additional direct arterial inflow. It therefore encounters both systemic arterial blood and portal blood that has already exchanged oxygen and substances with upstream tissues. This dual supply must be represented explicitly when hepatic oxygen delivery or extraction is calculated.

Regional vessels respond to sympathetic activity, circulating mediators, local metabolism, endothelial signals, and transmural pressure. Vasoconstriction during an acute circulatory threat can redistribute blood toward other organs, while local metabolic signals oppose restriction within active tissue. The net response depends on the threat, vessel, and time course rather than a fixed percentage of cardiac output assigned to the gut.

Venous capacity also matters. The splanchnic reservoir can accommodate or mobilize substantial blood through changes in tone and pressure. Redistribution changes useful filling without creating blood volume. Venous congestion, abdominal pressure, and hepatic resistance can simultaneously hinder drainage and increase tissue filtration.

The practical consequence is heterogeneity at several levels: between abdominal organs, between intestinal layers, and within the hepatic microcirculation. Normal global flow cannot establish uniform regional supply. Conversely, a local abnormality should not automatically be generalized to every splanchnic territory. The network must be reconstructed before a pressure, flow estimate, or biochemical result is assigned a mechanism.

69.2 Hepatic oxygen delivery has two inflows

Portal blood normally has a lower oxygen content than systemic arterial blood because upstream organs have extracted oxygen. Its contribution to hepatic oxygen delivery nevertheless depends on both that content and its flow. The hepatic artery carries a different combination: generally higher oxygen content through a separate flow pathway.

Let QHA and QPV be hepatic arterial and portal venous flows in L/min. Let CaO₂ and CpvO₂ be their oxygen contents in mL O₂/dL. Multiplying liters by 10 converts them to deciliters:

For constructed values, choose hepatic arterial flow 0.30 L/min with content 18 mL/dL and portal flow 0.90 L/min with content 12 mL/dL. Arterial delivery is 0.30 × 10 × 18 = 54 mL O₂/min. Portal delivery is 0.90 × 10 × 12 = 108. Total hepatic delivery is therefore 162 mL O₂/min through 1.20 L/min of blood.

If the inflows mix completely without intervening exchange, their combined content is 162 divided by 12 dL/min, or 13.5 mL/dL. This is a flow-weighted content calculation. Averaging arterial and portal oxygen tensions would not conserve oxygen amount, because hemoglobin binding makes tension and content nonlinearly related.

Portal oxygen content is itself a changing variable. Increased upstream extraction can lower it even when portal volume flow is preserved. Anemia, hypoxemia, altered intestinal metabolism, and changes in flow distribution can therefore alter hepatic supply through different pathways.

The teaching values are not normal flow fractions or targets. Actual inflow proportions vary with feeding, circulation, vascular disease, surgery, and liver pathology. Measurements must also distinguish blood flow from plasma flow and whole-blood oxygen content from saturation alone. Dual supply provides two routes of delivery, not a guarantee that either can fully replace the other under stress.

69.3 The hepatic arterial buffer response provides partial compensation

When portal flow decreases, hepatic arterial resistance can fall and arterial flow can increase. This hepatic arterial buffer response, or HABR, limits part of the change in total hepatic flow. It is not an assurance of full compensation, and it does not imply an equivalent reciprocal portal response when arterial flow is lost.

The adenosine-washout hypothesis supplies a mechanism. Adenosine produced near hepatic arterial resistance vessels can be removed into nearby flowing blood. Reduced portal flow decreases that washout, allowing local adenosine to accumulate and dilate the arterial vessels. The signal concerns local production and removal rather than a direct measurement of whole-liver oxygen need.

Lautt and colleagues tested this mechanism in anesthetized cats through portal-flow reduction and pharmacological manipulation of adenosine responses. Their findings supported adenosine-related partial arterial buffering. The measured compensation belonged to that preparation and should not become a fixed percentage assigned to every patient. 1

Ezzat and Lautt subsequently found adenosine-related hepatic arterial pressure–flow autoregulation in cats. This connects arterial autoregulation with local washout behavior while retaining the limits of the experimental vascular and pressure conditions. HABR and arterial autoregulation should not be treated as proof that hepatic flow remains constant throughout shock. 2

The available arterial pressure and remaining vasodilatory reserve constrain the response. Hepatic vascular disease, altered sinusoidal conditions, systemic vasoactive signals, and major anatomical changes can further modify it. Even an observed arterial increase does not establish what happened to regional sinusoidal perfusion or oxygen extraction.

Return to the constructed inflows. If portal flow falls from 0.90 to 0.45 L/min while arterial flow rises from 0.30 to 0.40, total flow becomes 0.85 rather than 1.20 L/min. Arterial recruitment has replaced only part of the lost portal volume. Its oxygen contribution may be proportionately greater because arterial blood carries more oxygen per unit volume, but the actual delivery still requires both flow–content products.

69.4 Mucosal oxygenation depends on microvascular geometry and demand

Intestinal blood flow is distributed among mucosa, submucosa, muscular layers, and other structures. These compartments perform different work. Absorption, epithelial ion transport, secretion, motility, and immune activity do not impose identical oxygen requirements or receive interchangeable vascular supplies.

Within a villus, arterial inflow and venous return lie close enough for oxygen to diffuse between pathways before reaching more distal tissue. Such countercurrent exchange can reduce effective delivery toward the tip. Tissue consumption along the vascular route also lowers oxygen availability, even without substantial direct arteriovenous shunting.

Shepherd and Kiel’s original villus model examined these convective and diffusive interactions. It showed that countercurrent exchange could impair oxygenation under specified conditions, but also that a base-to-tip oxygen gradient was not itself proof of shunting. The model’s geometry and steady-state assumptions do not provide a universal human villus oxygen tension or failure threshold. 3

Villus-tip enterocytes can be particularly vulnerable when supply falls relative to demand. Reduced flow, low arterial content, uneven erythrocyte distribution, and tissue swelling may converge on that region. Increased total mesenteric flow does not establish that these local constraints have been removed.

Human experimental observations also show that injury and repair are dynamic. In patients undergoing pancreatic surgery, Derikx and colleagues studied a short jejunal segment destined for removal. Temporary ischemia produced enterocyte injury, followed during reperfusion by shedding of damaged cells and restoration of epithelial continuity. This selected short-duration surgical model does not predict how rapidly a diffusely shocked, edematous, or persistently ischemic intestine will recover. 4

The clinically useful distinction is between an anatomical vulnerability and a measured injury state. A villus arrangement explains why regional oxygen limitation is possible; it does not diagnose ischemia from anatomy alone. Likewise, a repaired epithelial surface does not independently establish normal absorption, microvascular flow, or long-term barrier function.

69.5 Shock changes regional flow in ways global measurements can miss

Low cardiac output and reduced arterial pressure can restrict splanchnic supply, while sympathetic and vasoactive responses redistribute it. In sepsis, endothelial activation, altered tone, leukocyte interactions, and microvascular obstruction can disturb distribution even when total output is preserved. The two patterns can overlap.

A systemic pressure increase can improve flow where perfusion is pressure limited. The intervention producing that increase can simultaneously change cardiac output and local resistance. It is therefore inaccurate to assume either that every vasoconstrictor necessarily worsens gut supply or that restoring arterial pressure guarantees its recovery.

Intestinal microvascular measurements demonstrate this regional limitation. Edul and colleagues studied postoperative patients with abdominal sepsis and an ostomy, permitting comparison of intestinal and sublingual microcirculation. Their responses after a historical fluid challenge were dissociated: changes in accessible oral microvessels did not reliably describe intestinal changes. 5

The study used hydroxyethylstarch under its historical protocol. That exposure is part of the experiment’s context and is not a current fluid recommendation. The selected postoperative anatomy and small sample also prevent treating the intestine as universally isolated from systemic circulation. The defensible conclusion is that one sampled territory cannot certify another territory’s response.

Occlusive and nonocclusive ischemia require different explanations. Arterial embolism, thrombosis, or venous obstruction can create a focal anatomical problem. Low flow, severe vasoconstriction, and microvascular dysfunction can threaten bowel without an occluded main artery. A normal-looking proximal vessel or an improved cardiac output does not exclude the latter processes.

Duration, collateral pathways, baseline disease, oxygen content, and demand determine the consequences. Increasing extraction may initially preserve metabolism, but regional reserve is finite. Global lactate may be normal early or elevated from other tissues; neither state independently rules intestinal injury in or out. The trajectory must be interpreted with the abdominal and vascular context.

69.6 Venous and tissue pressures constrain abdominal perfusion

Blood must leave an organ for sustained arterial inflow to continue. Increased portal resistance can raise upstream intestinal venous pressure. Increased hepatic venous pressure can elevate sinusoidal pressure and impede liver drainage. Right-heart dysfunction, pericardial constraints, and thoracic pressure transmission can therefore influence the abdominal circulation from downstream.

These effects extend beyond a smaller arterial-to-venous pressure difference. Elevated capillary pressure favors filtration, and bowel-wall or hepatic edema can increase the distance and resistance between vessels and cells. In a constrained environment, rising interstitial pressure can compress small vessels and further restrict flow.

Intra-abdominal pressure changes several conditions simultaneously. It can compress vessels, raise surrounding tissue pressure, alter venous drainage, and transmit pressure toward the chest. Its effects depend on the arterial pressure, abdominal compliance, regional anatomy, and time course. A single subtraction of abdominal pressure from mean arterial pressure (MAP) is consequently only a descriptor, not a complete measurement of local organ perfusion.

Diebel and colleagues directly measured hepatic arterial, portal, and microcirculatory responses during increased abdominal pressure in anesthetized pigs. Hepatic perfusion deteriorated despite maintenance of systemic arterial pressure and cardiac output under the experimental conditions. The study demonstrates a downstream and tissue-pressure mechanism; its pressure steps do not define universal human thresholds or a resuscitation protocol. 6

Central venous pressure (CVP) requires the same care. It is referenced intravascular pressure, with respiratory and hydrostatic influences, rather than a direct measurement of every hepatic or portal pressure. An externally transmitted component still has consequences within connected vessels. It should not be discarded through a fixed positive end-expiratory pressure correction or substituted automatically for the effective downstream pressure of the intestinal microcirculation.

This explains why edema, high venous pressure, and reduced organ function can coexist with an apparently acceptable arterial pressure. Additional fluid may improve useful filling in one setting while aggravating filtration and congestion in another. Responsiveness and tolerance remain different questions.

69.7 Epithelial continuity, absorption, and motility are different functions

The intestinal epithelium separates luminal contents from the internal environment while selectively transporting nutrients, water, and electrolytes. Tight-junction behavior, epithelial renewal, mucus, immune activity, and the microbiota contribute to this interface. Barrier dysfunction can permit abnormal molecular passage, but does not automatically establish clinically important bacteremia or identify its source.

Absorption requires additional processes. Luminal digestion and mixing must make substrates available; transporters move them across enterocytes; cellular energy supports ion gradients and processing; blood or lymph then removes the absorbed material. A continuous surface can therefore coexist with impaired functional uptake.

Deane and colleagues measured glucose absorption and intestinal transit in critically ill patients using an intraduodenal nutrient exposure. Absorption was reduced compared with healthy participants, while transit measurements did not explain the difference. The design separates absorption from simply delivering nutrient beyond the stomach; its tracer and population do not describe every nutrient or every intensive care unit (ICU) state. 7

Related human and mouse work found reduced glucose absorption associated with lower expression of intestinal glucose-transporter and sensing transcripts. This supplies a cellular correlate, but transcript abundance is not a direct measurement of transporter activity, and the findings do not isolate one cause of impaired uptake. 8

Motility is another domain. Gastric emptying can be delayed by illness, drugs, metabolic conditions, and neural disturbances. Distension, vomiting, or a large gastric residual may indicate impaired handling, but none directly measures mucosal perfusion. Diarrhea also has multiple causes, including medication, infection, secretion, and altered absorption.

The reverse inference is equally limited: receiving a prescribed feed volume without obvious symptoms does not prove complete absorption or adequate regional oxygenation. Assessment should preserve the distinction between what entered the tube, what moved through the lumen, what crossed the epithelium, and what the body ultimately used. This functional separation also prevents a barrier-injury marker from becoming a substitute for nutritional or perfusion assessment.

69.8 Feeding increases work, while nutrition trials test complete strategies

Digestion and absorption increase cellular transport, secretion, processing, and metabolic activity. Local vascular responses normally support this work through postprandial hyperemia. Sodium-linked absorption, for example, depends indirectly on adenosine triphosphate (ATP)-consuming ion-gradient maintenance. Nutrient delivery therefore adds demand as well as substrate.

A small physiological study by Sim and colleagues compared ventilated critically ill adults older than 65 years with healthy older participants. Small-intestinal glucose produced a smaller mesenteric flow increment and reduced tracer glucose absorption during critical illness. These associated responses support examining vascular recruitment and uptake together; they do not establish a causal flow threshold or predict every patient’s feeding response. 9

When circulatory reserve is limited, the normal response may be constrained by low flow, abnormal distribution, venous congestion, or cellular dysfunction. This supplies a physiological reason to examine tolerance during acute shock. It does not establish that every use of enteral nutrition during vasopressor treatment causes ischemia, or that a specific infusion rate defines safety.

NUTRIREA-2 randomized 2410 ventilated adults receiving vasopressors for shock to early enteral or parenteral nutrition with similar normocaloric aims. Enteral feeding did not reduce 28-day mortality or ICU-acquired infections and produced more digestive complications, including uncommon but more frequent bowel ischemia. Recruitment stopped after an interim assessment judged further enrollment unlikely to change the result. These findings concern the tested early route/intensity strategies in that population, not every amount of enteral feeding or a permanent advantage of parenteral nutrition. 10

NUTRIREA-3 addressed a different question: lower versus standard combined calorie and protein targets during the first ICU week in ventilated adults with shock. It randomized 3044 patients, with 3036 remaining after consent withdrawals. Lower targets did not significantly reduce 90-day mortality, but shortened time to readiness for ICU discharge and reduced several gastrointestinal complications. 11

Calories and protein changed together, so the trial cannot isolate the effect of either nutrient. Readiness for discharge is also different from actual discharge, and digestive events were secondary outcomes rather than direct measurements of mucosal oxygen balance. Neither trial proves a particular perfusion mechanism for every complication.

Acute-phase observations do not define requirements throughout recovery. The resulting lesson is to retain route, intensity, timing, population, and outcomes when discussing feeding physiology. A demand–supply rationale and a clinical feeding strategy need complementary evidence rather than being treated as the same claim.

69.9 Hepatic extraction and clearance depend on more than flow

The liver removes some circulating substances through cellular uptake, metabolism, biliary excretion, or combinations of these processes. The fraction removed during one passage is the extraction ratio. Clearance expresses removal as an equivalent volume of blood cleared per unit time; it is not the physical volume of blood diverted out of the circulation.

For a simplified steady-state model, let Cin and Cout be representative inlet and outlet whole-blood concentrations in mg/L, Q hepatic blood flow in L/min, E the dimensionless extraction ratio, and CL clearance in L/min. With matching compartments and no unaccounted production or storage:

Choose constructed values Cin 10 mg/L, Cout 6 mg/L, and Q 1.20 L/min. Extraction is 0.40, clearance 0.48 L/min, and removal 4.8 mg/min. For dual inflow, Cin must appropriately represent both incoming streams; an arterial sample alone can be insufficient when portal concentration differs.

Flow and intrinsic cellular capacity interact. A substance efficiently removed from arriving blood can become strongly delivery limited. When uptake or metabolism is relatively slow, changing cellular capacity or unbound availability can have a larger influence than a comparable flow change. Binding, transport, saturation, shunting, and internal mixing modify these relationships.

Pang and Rowland’s original theoretical comparison showed why well-stirred and parallel-tube models require explicit assumptions about mixing, binding, enzyme activity, and diffusion. Their agreement in some circumstances does not make their internal concentrations or predictions universally interchangeable. 12

Experimental perfused-liver work by Shand and colleagues also demonstrated that route of entry changes the relation between flow, intrinsic removal, and measured systemic exposure. This model does not supply a bedside dosing rule for an unstable patient. 13

Consequently, a reduced disappearance rate of an administered marker can reflect several mechanisms. It cannot independently distinguish reduced flow from altered extraction, transport, or functional tissue mass. Plasma measurements must additionally be matched to plasma flow or converted appropriately before being combined with blood-flow estimates.

69.10 Injury, synthesis, and cholestasis provide different liver signals

Aminotransferases primarily signal cellular injury and enzyme release, not a direct rate of hepatic perfusion or a complete measure of function. Their trajectory reflects injury, release, distribution, and removal. A declining concentration can accompany recovery but does not alone prove restored synthetic or metabolic capacity.

The liver’s organization helps explain regional vulnerability. Blood moves from portal regions through sinusoids toward central venous outflow, with oxygen consumption and zonated metabolic work along the pathway. Pericentral tissue can become vulnerable to combinations of impaired delivery, hypoxemia, congestion, and altered use. This gradient is not a uniform map of every injured liver.

Henrion and colleagues described differing circulatory and respiratory contexts in a clinical series of hypoxic hepatitis. The observations reinforce that a recorded episode of profound systemic hypotension is not required in every case. The clinical patterns do not independently diagnose mitochondrial dysfunction or quantify regional oxygen use in an individual patient. 14

Synthetic signals ask another question. Coagulation-factor production can deteriorate, but prolonged clotting tests also reflect anticoagulants, vitamin K status, consumption, and assay conditions. Albumin is influenced by inflammation, distribution, loss, and turnover as well as synthesis. Neither is a specific perfusion assay, and different proteins respond over different timescales.

Cholestasis concerns bile formation and transport. Inflammation can alter uptake and canalicular handling without requiring the same degree of hepatocyte necrosis. Trauner and colleagues demonstrated endotoxin-related downregulation of a hepatic bile-acid transporter in rats, supporting a regulated transport mechanism. It does not establish that every bilirubin increase in critical illness has that cause. 15

Bilirubin additionally depends on production, uptake, conjugation, and excretion. Hemolysis, obstruction, drugs, congestion, and pre-existing disease can modify the pattern. Thus, a delayed bilirubin rise after improved circulation need not contradict earlier hemodynamic recovery, but it still requires an explanation. Injury markers, synthetic behavior, and cholestatic measurements must remain distinct observations before being integrated.

69.11 Bedside evidence samples different parts of the system

Clinical examination, abdominal imaging, systemic hemodynamics, vascular Doppler, and biochemical trends provide complementary information. Each samples a different scale. Vessel patency does not establish normal mucosal microflow, while a local mucosal observation does not quantify total portal delivery or hepatic extraction.

Doppler velocity requires vessel geometry to become a volumetric flow estimate. Respiratory motion, sampling alignment, pulsatility, shunts, and changing diameter affect interpretation. Hepatic and portal venous waveforms can reveal transmitted cardiac or congestive influences without directly measuring total body volume or proving fluid intolerance in every organ.

Lactate requires particular care. The liver contributes substantially to its metabolism, while production occurs in many tissues and the kidneys also participate in disposal. A falling systemic concentration represents net production, distribution, and removal rather than an isolated hepatic clearance measurement. Splanchnic organs can also change between net uptake and release under different metabolic conditions.

In a secondary cohort analysis, Sterling and colleagues associated liver dysfunction with altered lactate normalization during early sepsis resuscitation. This supports context-sensitive interpretation, but the study’s serial percentage changes were not direct measurements of hepatic clearance in L/min. It does not establish a different universal lactate target or justify assigning every persistent elevation to liver dysfunction. 16

Markers of enterocyte injury likewise describe cellular release rather than directly measuring all bowel perfusion or absorption. Timing, anatomical extent, distribution, and removal affect the signal. An accessible biomarker cannot independently identify bowel viability throughout the abdomen.

The most informative assessment therefore asks whether observations made at comparable times support a coherent mechanism. New congestion, falling output, rising abdominal pressure, a focal vascular lesion, and an absorption problem imply different explanations even when several accompany the same lactate or bilirubin trend. Discordance should prompt reconsideration of tissue location, timing, and measurement assumptions rather than forcing every result into one perfusion label.

69.12 Integrate inflow, drainage, demand, and organ function

Consider the constructed liver receiving portal flow 0.90 L/min at oxygen content 12 mL/dL and arterial flow 0.30 L/min at 18 mL/dL. Initial delivery is 162 mL O₂/min. Now portal flow falls to 0.45 while arterial flow rises to 0.40. Holding both contents fixed for the example, portal delivery becomes 0.45 × 10 × 12 = 54 and arterial delivery 0.40 × 10 × 18 = 72. Total delivery is 126 mL O₂/min.

Arterial buffering has occurred, yet delivery has fallen by 36 mL O₂/min, or approximately 22.2% of the initial value. Total flow has decreased from 1.20 to 0.85 L/min. The combined inlet content actually rises to 126/8.5 = approximately 14.82 mL/dL because the arterial proportion is larger. A higher mixed inflow content has therefore accompanied lower total oxygen delivery.

These assumed observations do not identify why portal flow fell. Reduced upstream arterial supply, altered intestinal resistance, venous or abdominal-pressure effects, and changing systemic output are possibilities. The actual portal oxygen content could also change, making the fixed-content assumption inappropriate in a real patient.

Suppose systemic arterial pressure subsequently improves while bowel edema and bilirubin rise. Improved pressure may have helped a pressure-limited region, while congestion, permeability, transport dysfunction, or evolving injury persists elsewhere. A lower lactate would add a net metabolic observation without certifying restored intestinal perfusion or hepatic extraction.

Feeding tolerance would provide another domain: gastric handling, absorption, substrate use, and demand. Neither prescribed intake nor the absence of vomiting would resolve the vascular questions. Conversely, the nutrition trials do not permit a precise prediction from these constructed flows.

The coherent interpretation follows the network rather than one number. Arterial inflow, portal content and drainage, hepatic arterial reserve, local microvascular conditions, metabolic demand, and the particular liver or intestinal function being measured must be reconciled over time. This approach explains competing responses without inventing a universal perfusion target or feeding protocol.

References

1. Lautt WW, Legare DJ, d’Almeida MS. Adenosine as putative regulator of hepatic arterial flow (the buffer response). Am J Physiol. 1985;248:H331–H338. Original experiment.

2. Ezzat WR, Lautt WW. Hepatic arterial pressure-flow autoregulation is adenosine mediated. Am J Physiol. 1987;252:H836–H845. Original experiment.

3. Shepherd AP, Kiel JW. A model of countercurrent shunting of oxygen in the intestinal villus. Am J Physiol. 1992;262:H1136–H1142. Original model.

4. Derikx JPM, Matthijsen RA, de Bruïne AP, et al. Rapid reversal of human intestinal ischemia-reperfusion induced damage by shedding of injured enterocytes and reepithelialisation. PLoS One. 2008;3:e3428. Original study.

5. Edul VSK, Ince C, Navarro N, et al. Dissociation between sublingual and gut microcirculation in the response to a fluid challenge in postoperative patients with abdominal sepsis. Ann Intensive Care. 2014;4:39. Original study.

6. Diebel LN, Wilson RF, Dulchavsky SA, Saxe J. Effect of increased intra-abdominal pressure on hepatic arterial, portal venous, and hepatic microcirculatory blood flow. J Trauma. 1992;33:279–282; discussion 282–283. Original experiment.

7. Deane AM, Summers MJ, Zaknic AV, et al. Glucose absorption and small intestinal transit in critical illness. Crit Care Med. 2011;39:1282–1288. Original study.

8. Deane AM, Rayner CK, Keeshan A, et al. The effects of critical illness on intestinal glucose sensing, transporters, and absorption. Crit Care Med. 2014;42:57–65. Original study.

9. Sim JA, Horowitz M, Summers MJ, et al. Mesenteric blood flow, glucose absorption and blood pressure responses to small intestinal glucose in critically ill patients older than 65 years. Intensive Care Med. 2013;39:258–266. Original study.

10. Reignier J, Boisramé-Helms J, Brisard L, et al. Enteral versus parenteral early nutrition in ventilated adults with shock: a randomised, controlled, multicentre, open-label, parallel-group study (NUTRIREA-2). Lancet. 2018;391:133–143. Original trial.

11. Reignier J, Plantefeve G, Mira JP, et al. Low versus standard calorie and protein feeding in ventilated adults with shock: a randomised, controlled, multicentre, open-label, parallel-group trial (NUTRIREA-3). Lancet Respir Med. 2023;11:602–612. Original trial.

12. Pang KS, Rowland M. Hepatic clearance of drugs. I. Theoretical considerations of a “well-stirred” model and a “parallel tube” model. Influence of hepatic blood flow, plasma and blood cell binding, and the hepatocellular enzymatic activity on hepatic drug clearance. J Pharmacokinet Biopharm. 1977;5:625–653. Original model analysis.

13. Shand DG, Kornhauser DM, Wilkinson GR. Effects of route of administration and blood flow on hepatic drug elimination. J Pharmacol Exp Ther. 1975;195:424–432. Original indexed experiment.

14. Henrion J, Schapira M, Luwaert R, Colin L, Delannoy A, Heller FR. Hypoxic hepatitis: clinical and hemodynamic study in 142 consecutive cases. Medicine (Baltimore). 2003;82:392–406. Original series.

15. Trauner M, Arrese M, Lee H, Boyer JL, Karpen SJ. Endotoxin downregulates rat hepatic ntcp gene expression via decreased activity of critical transcription factors. J Clin Invest. 1998;101:2092–2100. Original experiment.

16. Sterling SA, Puskarich MA, Jones AE. The effect of liver disease on lactate normalization in severe sepsis and septic shock: a cohort study. Clin Exp Emerg Med. 2015;2:197–202. Original cohort analysis.