The effect of plasma versus crystalloids on intravascular volume, tissue perfusion and tissue oxygenation: a randomized controlled trial in a human endotoxemia model

Abstract

Background

Fluid resuscitation is a cornerstone of shock treatment to ensure sufficient tissue perfusion. However, large volumes of crystalloids are associated with adverse outcomes and plasma has been suggested as an alternative fluid. This trial investigated the effect of plasma transfusion on circulation, tissue perfusion, and mitochondrial oxygen tension (mitoPO2) compared with balanced crystalloids in a human endotoxemia model.

Methods

Twelve healthy volunteers were subjected to 2 ng/kg lipopolysaccharide (LPS) infusion. Six participants received 10 ml/kg of plasma. In the control group, 10 ml/kg of balanced crystalloids was infused. Non-invasive cardiac index (CI), markers of the microcirculation and mitoPO2 were measured at prespecified time points. Blood volume was estimated by hemoglobin levels at each timepoint. Differences in groups over time were tested using linear mixed modeling. Wilcoxon rank-sum test was used to test the statistical difference between groups at each timepoint.

Results

LPS infusion resulted in a systemic inflammatory reaction with an increase in temperature and leukocytes. Heart rate and CI increased, while systemic vascular resistance (SVRI) decreased. Capillary refill time, mitoPO2 and perfused sublingual vessels all deteriorated, indicative of hyperdynamic vasodilation with redistribution of blood volume. Plasma increased intravascular volume by 4 ml/kg one hour after infusion, whereas intravascular volume remained at baseline with crystalloids (p < 0.01). Subsequently, CI increased more in the plasma group compared to the crystalloid group (p < 0.05). Tissue perfusion was improved in the plasma group, with shorter capillary refill time (p < 0.05) and a higher proportion of perfused vessels compared to the crystalloid group (p < 0.05). After fluid infusion, mitoPO2 increased less in the plasma group compared to the crystalloid group (p < 0.05).

Conclusions

In healthy endotoxemic volunteers without shock, plasma infusion augmented intravascular blood volume and cardiac output, resulting in improved microcirculatory tissue perfusion compared to crystalloids. However, plasma did not restore the reduction in skin mitoPO2 as much as crystalloids.

Trial registration

EudraCT ID: EUCTR2020-005045-17-NL. Date of registration: 14-12-2021

Supplementary Information

The online version contains supplementary material available at 10.1186/s13054-026-05885-2.

Introduction

Fluid resuscitation is essential in the treatment of shock and sepsis to restore tissue perfusion [1]. Current guidelines recommend intravenous balanced crystalloids [2]. However, crystalloids only partly remain intravascular [3], increasing the risk of tissue edema. Furthermore, crystalloids are associated with glycocalyx degradation, capillary leakage and edema, which have been hypothesized to impair cellular oxygen delivery and lead to worse patient outcomes [4–6].

Plasma has been suggested as an alternative resuscitation fluid as it more effectively increases intravascular volume, decreases vascular leakage by maintaining endothelial barrier integrity and restores microvascular flow better than other resuscitation fluids in experimental shock models [7–10]. However, it is unknown what the effects of plasma are on intravascular blood volume and perfusion in humans. Particularly in sepsis, an increase in endothelial permeability alters fluid and protein distribution across compartments [5]. The sublingual microcirculation is disturbed in sepsis and is associated with worse outcomes [11, 12].Taken together, the effects of plasma on global hemodynamic parameters in inflammatory conditions remain unknown. In addition, little is known of the effect of plasma on microvascular perfusion and subsequent tissue oxygenation, which ultimately is the goal of fluid resuscitation. The protoporphyrin IX-triplet state lifetime (PpIX-TSLT) technique non-invasively measures the mitochondrial oxygen tension (mitoPO2) within the skin which corresponds to mitoPO2 in various organs [13, 14]. In healthy volunteers and critically ill patients, a low mitoPO2 and has been shown to correspond to end-organ tissue hypoxia [15, 16].

In a randomized, controlled, non-blinded human endotoxemia model characterized by a hyperdynamic systemic inflammation, we investigated the effect of plasma transfusion on hemodynamics, tissue perfusion and mitoPO2 compared to infusion with crystalloids. We hypothesized that plasma transfusion would restore the microcirculation more effectively and improve tissue perfusion and oxygenation compared to a similar volume infusion with balanced crystalloids.

Methods

Study design

This study was approved by the Medical Ethical Committee Amsterdam University Medical Center (NL74983.018.21) and is in accordance with the Declaration of Helsinki. The ‘Resuscitation of Endothelial Permeability in Endotoxemia’ study was a healthy volunteer, open-label, randomized study conducted at Amsterdam University Medical Center. The findings on host response of this study are previously published [17]. The participants in this and the aforementioned paper are the same subjects. All enrolled participants provided written informed consent before enrollment. The study was conducted and reported with the CONSORT guidelines [18].

Inclusion of participants

Between April 2023 and October 2023, healthy male nonsmoking participants not taking any medication, aged 18 to 35 years old, were included. Prior to enrollment, participants were screened on medical history, physical examination, blood work, urine tests and electrocardiogram (ECG). participants were excluded in case of any abnormality during screening. Participants who dropped out of the study between the screening and subsequent visit were replaced.

Blinding, randomization and plasma products

Participants were allocated to one of two groups: either receiving 10 mL/kg infusion of Plasma-lyte (148 Baxter, Deerfield, IL) or 10 mL/kg Omniplasma (Octapharma GmbH, Lachen, Switserland). Due to evaluating safety of a novel batch of LPS, the initial two participants received Plasma-lyte. Due to evaluating safety of transfusion of plasma, the subsequent two received Omniplasma. The remaining participants were randomly distributed between the groups. Researchers were not blinded, as this was not feasible. The plasma product was Omniplasma, which is a pooled (600–1200 donors), solvent detergent treated and prion-reduced plasma product used as the standard plasma product nationwide. Plasma products were freshly thawed and sourced from the local blood bank on the day of the experiments. Fluids were infused at 10 ml/kg/hour during 1 h. This was chosen based on the surviving sepsis campaign guidelines while minimizing the risk of fluid overload [2].

Human endotoxemia model

The night before the experiment participants self-administered an ALAcare plaster (Photonamics GmbH, Pinneberg, Germany) on the sternum to enable mitoPO2 measurement. The morning of the experiment, a radial arterial catheter was inserted for continuous measurement of mean arterial blood pressure and arterial blood gas and blood sampling. Participants were admitted to the research unit of the intensive care department and were continuously monitored during the entirety of the experiment. Subjects were kept fasted throughout the experiment to prevent adverse effects and confounding of volume effects. A peripheral intravenous catheter was inserted for the administration of study products. Vital parameters were continuously monitored using five-lead ECG and pulse oximetry (Philips, Eindhoven, the Netherlands). Temperature was measured hourly using a tympanic thermometer (Genius II, Medtronic, Minneapolis, MN). After measuring all baseline values, 2 ng/kg Lipopolysaccharide (LPS) from Escherichia coli O113 (National Institute of Health Clinical Center, Bethesda, MD) was injected as bolus and flushed with normal saline through the peripheral IV catheter (T = 0). The 2ng/kg LPS dose was based on the minimum dose to elicit a sepsis-like response [19].

Thirty minutes following LPS injection (T = 0.5), fluid intervention products were infused over one hour. The experiment concluded 8.5 h after LPS infusion. Participants were discharged home after resolvement of all symptoms and provided with dinner if they wished.

Sample collection

Arterial blood was collected at baseline prior to LPS administration (T = 0), 30 min after infusion of LPS but before transfusion (T = 0.5), and T = 2.5, T = 4.5 and T = 6.5 h after infusion of LPS. Complete blood counts were determined at the same timepoints by the clinical laboratory.

Non-invasive cardiac output

CI and systemic vascular resistance index (SVRI) were measured continuously throughout the experiment using the volume clamp method in the Nexfin device (BMEYE, Amsterdam, The Netherlands) as arterial wave-form estimation of cardiac output was not available. The change of direction between two measurements of Nexfin-CI has been shown to acceptable agreement with invasive cardiac output measurements and can therefore reliably indicate trends of changes in CI [20, 21]. An unused finger cuff was placed on the index or middle finger according to the manufacturer’s instructions. The mean CI and SVRI of 10 min was automatically recorded for each time point. Capillary refill time was measured by applying pressure on the sternum for 5 s followed by release. Refill was measured using a digital stopwatch.

Sublingual microcirculation

The sublingual microcirculation was recorded using a handheld video microscope with sidestream darkfield imaging (SDF) using the MicroScan (MicroVision Medical, The Netherlands, Amsterdam). Sublingual measurements were done by a researcher with experience in microcirculatory image recording at baseline prior to LPS administration (T = 0) and T = 2.5, T = 4.5 and T = 6.5 h after infusion of LPS. Measurements at T = 0.5 have been excluded as they were not completed before start of fluid infusion. Three anatomical sites were recorded: the medial and both lateral parts of the sublingual area. Video captures were at least 5 s long as intermittent flow was not part of the assessment. Sublingual videos were analyzed using validated offline software-aided analysis with AVA 3.2 software (Microvision Medical, Amsterdam, The Netherlands) [22]. The videos were analyzed by a trained investigator with experience in SDF image recording, after being trained by experienced instructors. The investigator was blinded to the allocation group. Good quality captures required an absence of pressure artifacts, defined by the presence of continuous flow in large vessels (> 20 μm). For sufficiently focused images, red blood cells should have been clearly seen in the capillaries. Images complied with the second consensus requirements of sublingual microcirculatory image analysis [23].

Images were only evaluated in case the quality was sufficient. The proportion of (small) perfused vessels (PPV), total vessel density, and the perfused DeBacker density were calculated by the software after manual identification and selection of vessels. These variables are required to assess the effect of fluid administration [23]. The perfused DeBacker Density are all vessels with visible microcirculatory flow. The Percentage of Perfused Vessels (PPV) is calculated as the percentage of perfused vessels in relation to the total number of all vessels and for small vessels in particular (PPV small). Small vessels (capillaries) have a diameter less than 20 μm. The mean values of three recordings of each parameter were used for final analysis.

MitoPO2

MitoPO2 was measured with the COMET device (Photonics Healthcare B.V.) and computed from the mean of at least 5 mitoPO2 measurements during 5 min at prespecified time points.

Derivation of intravascular volume expansion

We used the change in Hb to derive the degree of intravascular volume expansion, described comprehensively in validation studies [24]. The following formula was used:

Where 50 ml/kg was assumed as the baseline intravascular blood volume for our participants.

Sample size calculation

This study answers a secondary objective of an investigation that was powered to detect a significant difference in syndecan-1 levels. Also, no previous study has examined the effect of crystalloid versus plasma resuscitation on tissue perfusion and mitoPO2 in human endotoxemia. As such, a formal sample size calculation was not possible.

Statistical analysis

Normally distributed data are expressed as mean ± standard deviation and non-normally distributed data as median with interquartile ranges. Normality of data was assessed visually and using the Shapiro-Wilk test. Baseline differences between groups were calculated using either the unpaired Student’s t-test or the Mann-Whitney U test as appropriate. To assess within-group changes over time, a paired Student’s t-test or the Wilcoxon ranked sums test was used to compare values at T = 0 with the maximum or minimum values observed during the study period. To analyze differences in continuous variables between groups over time, linear mixed models were performed, incorporating time point and group as fixed effects and subject ID as a random effect. A p-value of less than 0.05 was considered statistically significant. Missing data was assumed to be missing at random and handled using mixed modelling for repeated measures.

Results

Of 15 volunteers undergoing medical screening, one was ineligible due to bigeminal ventricular premature complexes on ECG and two eligible subjects withdrew from the study after screening before the scheduled appointment day. All subjects that received LPS were included in the final analysis. Baseline characteristics are shown in supplementary Table 1.

Inflammatory parameters

All participants had a maximum change in vital parameters 3.5 h after LPS infusion. Median body temperature increased from 37.0 (36.6–37.4) to 38.4 (35.9–39.0) °C and heart rate increased from 67 (49–85) to 96 (72–110) beats per minute (Figs. 1 and 2). In addition, leukocyte counts and CRP increased after 4 h after LPS administration [17]. All participants met the systemic inflammatory response syndrome (SIRS) criteria.

Fig. 1.

Line plots showing temperature over time. Median and IQR are plotted. The yellow data points denote the plasma group. Blue data points denote the crystalloid group. Vertical black line at t = 0.5 denotes start of volume infusion

Fig. 2.

Evolution of hemodynamic variables over time. Median and IQR are plotted. (A) systolic blood pressure, (B) diastolic blood pressure, (C) heart rate, (D) cardiac index in percentage change from baseline, (E) systemic vascular resistance index in percentage change from baseline, (F) hemoglobin. The yellow line denotes the plasma group. Blue line denotes the balanced crystalloid group. Vertical black line at t = 0.5 denotes start of volume infusion

Global hemodynamics

Directly after LPS infusion, heart rate and CI increased in both groups while SVRI decreased, indicative of a hyperdynamic vasodilatory condition (Fig. 2). The CI (p < 0.05) and heart rate (p < 0.01) increased more in the plasma group compared to the crystalloid group. Plasma resulted in a Hb decrease of 1 g/dl whereas Hb remained constant with crystalloids (p < 0.01). The decrease in SVRI was more pronounced in the plasma group compared to the crystalloid group (p < 0.05). Post-hoc pairwise testing revealed no specific time points where CI and SVRI differed between groups.

Mitochondrial oxygen tension

In both groups, mitoPO2 decreased after LPS, despite fluid infusion, to a nadir of 14 mmHg [7, 18] mmHg with plasma and 16 mmHg [10, 22] with crystalloid expansion. At 1 h after completion of fluid therapy, at 2.5 h, mitoPO2 increased more with crystalloids than with plasma (Fig. 3A, p < 0.05). Overall, mitoPO2 was higher in the crystalloid group compared to the plasma group (p < 0.05).

Fig. 3.

Line plots showing the evolution of variables of tissue perfusion over time. Median and IQR are plotted. (A) mitochondrial oxygen tension, (B) Capillary refill time, (C) Lactate, D) De Backer score, E) Proportion of perfused vessels (PPV), F) Total vascular density. The yellow line denotes the plasma group. Blue line denotes the balanced crystalloid group. Vertical black lines denote start of fluid infusion

Markers of tissue perfusion

Capillary refill time was 1.0 [0.8, 1.2] seconds at baseline and increased to 2.7 [2.3, 3.0] seconds in the balanced crystalloid group and to 2.1 [1.9, 2.3] seconds in the plasma group (Fig. 3). Over the complete time period, CRT was faster in the plasma group than the crystalloid group (p < 0.01). Specifically, 1 h after completion of fluid infusion, capillary refill time was lower in the plasma group compared to the crystalloid group. Similarly, as a parameter of convective blood flow, the percentage of perfused vessels (PPV) remained higher in the plasma group compared to the crystalloid group after LPS and volume expansion (p = 0.04; Fig. 3). There was large baseline variation in total vessel density and the DeBacker score between groups, but when the change from baseline was considered, total vascular density decreased significantly in the crystalloid group compared to the plasma group (p = 0.04). Lactate remained unchanged during the study period (Fig. 3).

Markers of fluid overload

Atrial natriuretic peptide (ANP) increased after plasma infusion but not with balanced crystalloid group (Fig. 4, p < 0.01). The derived intravascular volume, as calculated by the drop in Hb, increased in parallel.

Fig. 4.

Line plots showing markers of fluid overload and endothelial integrity over time. Median and IQR are plotted. (A) Atrial natriuretic peptide, (B) derived intravascular volume expansion as percentage, The yellow line denotes the plasma group. Blue line denotes the balanced crystalloid group. Vertical black lines denote start of fluid infusion

Discussion

This model of human endotoxemia induced a vasodilatory hyperdynamic circulation with mild disturbances in skin perfusion. This study shows that in a hyper inflammatory condition, plasma leads to more effective volume expansion with an increase in cardiac index with subsequent microcirculatory perfusion. Crystalloids neither expanded volume effectively nor did they improve the microcirculation. Nonetheless, plasma was less effective in restoring skin mitoPO2 than crystalloids.

Our model is characterized by a vasodilatory hyperdynamic circulation, with increased CI, vasodilation and increased heart rate. Capillary refill time increased but remained within normal clinical range, reflecting redistribution of blood volume with normal arterial lactate. We observed the same magnitude of decrease in PPV and TVD as seen in endotoxemia and mildly septic patients [25–29]. The observed clinical picture satisfies the SIRS criteria and induced physiological changes common to acute infectious and inflammatory syndromes [30, 31]. Symptoms experienced by participants were mild with our dose of LPS and consisted mostly of headache and photophobia [17].

In this inflammatory state, plasma is an effective lasting volume expander as seen by the drop in Hb. In contrast, only 5% of crystalloids remained in the intravascular space after 2 h of infusion, which is lower than in literature [32, 33]. Possibly, the rapid rate of infusion increased the proportion of fluid leakage and decreased the reuptake of crystalloids from the interstitial space, making the intravascular expansion effect of crystalloids more transient [33, 34]. Plasma infusion lastingly increased stressed blood volume and cardiac preload seen in increased ANP levels, suggesting that endotoxemia and SIRS did not affect the rate of extravasation. This corresponds to previous results showing a half life of 3 h of plasma infusion in human volunteers with the same volume as in our study [35].

Volume expansion after plasma infusion translated to an augmented CI. In contrast, crystalloids showed an increase in cardiac output which was not accompanied by an increase in intravascular volume. The infused volume of 10 ml/kg/hour of crystalloids is considered sufficient to increase stroke volume in preload dependent subjects but we did not observe increase in intravascular volume, perhaps not being detected due to the long interval of Hb measurements [36]. Alternatively, the increase in CI after crystalloids might have due to the endotoxemia mediated tachycardia as part of the SIRS reaction [37]. Interestingly, a greater decrease in SVRI was observed in the plasma group. This might be explained by a decrease in viscosity by hemodilution, reducing systemic vascular resistance.

Importantly, plasma but not crystalloids also resulted in an improvement in microcirculatory perfusion. Previous studies in patients with sepsis have shown variable effects in microcirculatory parameters after crystalloid fluid resuscitation, in part depending on the phase of sepsis [11, 38–41]. The volume expansion by plasma and subsequent increase in stroke volume may have been the major prerequisite for microcirculatory improvements found previously in septic patients [34–36, 39]. Interestingly, the effect of plasma on PPV seem to occur earlier than improvement in cardiac output, possibly suggesting an additional protective mechanism of plasma, aside from improvements in systemic hemodynamics. A concern with colloid and plasma infusion may be that increased leakage of protein into the interstitial space in sepsis contributes to tissue edema [42]. However, there were no clinical or microcirculatory signs, of tissue edema in these endotoxemic participants [43]. In summary, in human endotoxemia, volume expansion by plasma is more effective in improving both the macro- and microcirculation.

The induced systemic inflammatory response resulted in a sharp decrease in mitoPO2 values concurrently with a prolongation in CRT as found before [44]. This suggests that cutaneous mitoPO2 decreases during redistribution of blood volume during SIRS. Also, the changes in mitoPO2 precede those of PPV, likely because blood flow is diverted away from the skin before the sublingual circulation [29, 45]. Plasma did not improve tissue oxygenation as measured with mitoPO2. MitoPO2 is primarily determined by blood flow, oxygen saturation, diffusion and oxygen consumption [46]. As blood flow improved with plasma and SaO2 remained unchanged, the lower mitoPO2 compared to crystalloids would be explained by limited diffusion or oxygen consumption. Alternatively, plasma or hemodilution might have exerted differential effects on the sublingual versus cutaneous microvasculature [47]. The capillary hemodilution by plasma may have worsened the diffusion distance for oxygen to the surrounding parenchymal tissue, which could have been further exacerbated by increased venous pressure. Unfortunately, the capillary hematocrit could not be measured with our SDF imaging software. Alternatively, cellular oxygen consumption may have increased by the administration of protein and albumin. In vivo animal models, plasma infusion in the liver increases tissue oxygen consumption by increasing glycolysis [48, 49]. MitoPO2 reflects the balance between oxygen supply and consumption in mitochondria. As consumption increases, mitoPO2 decreases.

In our model with mild tissue hypoperfusion and circulatory redistribution, improved tissue perfusion did not translate to an increased skin mitoPO2, possibly due to hypermetabolic or hemodilutery effects of plasma in this mild inflammatory model without overt shock.

Limitations

Although the cardiovascular, hematological and metabolic derangements in endotoxemia mimic those in SIRS and resembling some of the early alterations in sepsis, it remains a model of systemic inflammation without infection [37, 50]. Therefore, the model may have been too mild for the potential protective effects of plasma to become apparent on tissue oxygenation. This limits clinical applicability of the results in later stages of sepsis, as the hypotension and organ dysfunction induced by prolonged inflammation in sepsis and septic shock is not replicated in this model [19, 50]. Second, the accuracy of Nexfin compared to invasive cardiac output measurements is limited, especially for estimating absolute values of CI which is why only direction of change in CI could be reliably reported [20, 21]. In addition, the Nexfin device no longer is technically supported by the manufacturer. Third, because we chose not to withhold fluid therapy in the control group of participants, the model cannot distinguish between the short-lived effect of LPS versus the specific effect of fluid resuscitation. Lastly, due to the explorative nature of the study, a sample size calculation was not possible, requiring validation of the findings in a larger cohort. The open-label design might also have introduced bias into the subjective capillary refill time measurements. However, the changes in capillary refill time are corroborated by PPV and the De Backer score, which were evaluated blinded to the allocation group.

Conclusions

In this model of human endotoxemia characterized by a hyperdynamic circulation with mild disturbances in skin perfusion, plasma is a more effective method of volume expansion than crystalloids for improving hemodynamics and tissue perfusion. Despite this, skin mitoPO2 increased less in the plasma group compared to the crystalloid group.

Abbreviations

ANP

Atrial natriuretic peptide

5-ALA

Aminolevulinc acid

CO

Cardiac output

CRT

Capillary refill time

LPS

Lipopolysaccharide

MitoPO2

Mitochondrial oxygen tension

PpIX-TSLT

Protoporphyrin-IX triplet singlestate lifetime technique

SDF

Sidestream darkfield imaging

SIRS

Systemic inflammatory response syndrome

SVRI

Systemic vascular resistance index

Author contributions

DvdB, DK and NJ designed the study. DvdB and BH included and screened patients, conducted the study, performed clinical and laboratory measurements and analyzed the data. RK and NJ supervised the medical care of participants. DvdB, BH and NJ drafted the first version of the manuscript. All authors read and approved the final version of the manuscript.

Funding

Not applicable.

Data availability

All data analysed during this study are included in this published article and its supplementary information.

Declarations

Ethics approval and consent to participate

The study protocol was approved by the medical ethical review board of the Amsterdam UMC under registration number NL74983.018.21. We obtained informed consent from all volunteers who participated in the study.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.