Effects of Fluid Therapy on Mesenteric Microcirculation Using New Probe-Based Confocal Laser Endomicroscopy (Cellvizio®) in a Porcine Model of Endotoxic Shock

Microcirculation represents the anatomical structure where oxygen and nutrients exchange and the largest surface organ in the human body. During sepsis, pathophysiological alterations in microcirculation occur [1], inducing tissue hypoperfusion and cellular hypoxia [2, 3], leading to multiorgan failure.

Endothelial dysfunction is both a marker and the cause of organ failure. Johannson et al. [4] introduced the concept of “shock-induced endotheliopathy” which gathers endothelial alterations found in different states of shock, strongly associated with poor outcome [5]. Microcirculatory alterations have been observed at the early phase of sepsis, even though macrocirculation seems preserved, i.e., loss of hemodynamic coherence [6, 7]. Usually, these alterations are described as a decrease in vascular density (“capillary rarefaction”) and a decrease in capillary perfusion all together called functional capillary density (FCD), the main determinant of tissue oxygenation. During sepsis, a wide heterogeneity in microcirculation alterations has been described, as all vascular territories may not be affected in the same way. This heterogeneity has been associated with an altered and low capillary blood flow [8].

Most of the previous studies focusing on microcirculation analysis in septic shock evaluated sublingual microcirculation, using several evaluation tools such as the sidestream dark field, incident dark field, orthogonal polarization spectral, or near-infrared spectroscopy [9, 10]. Although these studies have resulted in important observations on the prognostic implications of microcirculatory alterations in critically ill patients, it remains questioned whether sublingual microcirculation alterations may adequately reflect microcirculatory intestinal alterations, at least at the early phase of sepsis [11].

During sepsis, it is known that the gut can become a pro-inflammatory organ that promotes systemic inflammatory response syndrome and leads to multiple organ failure in a vicious circle [12]. Besides, a real-time in vivo analysis of the microcirculation with these tools seems more challenging and time consuming as it requires a significant reprocessing of the images.

The probe-based confocal laser endomicroscope (Cellvizio®) firstly developed to guide abdominal and pulmonary biopsies has the advantage to allow a real-time in vivo visualization of the capillaries while fluorescent dyes such as FITC dextran are concomitantly used [13]. Laemmel et al. [14] showed that fiber confocal fluorescence microscopy (FCFM) allows to study the microcirculation in vivo, with a very high correlation (R² = 0.98) with intravital fluorescence microscopy. FCFM brings advantages such as confocality (avoiding the superposition of vessels), portability due to the small size of the probe, and good spatial resolution. FCFM allows to study the organ in its physiological environment [14]. To date, only one study has reported Cellvizio® data in a porcine model of septic shock, but its use was mainly endoluminal and only focused on histological analysis of the intestinal mucosa [15].

Another way to monitor microcirculation is through endothelial cell-related biomarkers such as soluble vascular endothelial growth factor receptor (sVEGFR), syndecan, soluble thrombomodulin, and angiopoietins 1 and 2, highly correlated with glycocalyx damages [16]. We chose an endotoxic shock model induced by an LPS infusion, to get as close as possible to the capillary hyperpermeability linked to sepsis and the hyperdynamic shock. Andersson et al. [17] showed that LPS perfusion leads to decreased ileal microcirculatory perfusion despite a superior mesenteric artery flow higher after LPS infusion than before.

There is increasing evidence that normalization of the microcirculation and maintenance of end-organ perfusion may be of great value in sepsis management [18]. Whether fluid administration improves microcirculation in the acute phase of septic shock remains controversial [19]. We thus sought to perform an experimental study to assess the effects of crystalloid fluid therapy on mesenteric microcirculation, using a new device, the confocal laser endomicroscope Cellvizio®, in a porcine model of endotoxic shock.

This study was conducted as a prospective experimental trial in a piglet model. The Animal Care and Use Committee Languedoc-Roussillon (CEEA-LR-12013) approved the protocol, and all experiments were performed in an authorized animal research laboratory. All facilities and transport comply with current legal requirements.

Twenty-one female piglets weighing 32 ± 3 kg were fasted overnight prior to the experiment, with free access to water. The animals were intravenously anesthetized with sufentanil (0.3 μg/kg) and propofol (a bolus of 3 mg/kg followed by 1 mg/kg/h) after intramuscular premedication with ketamine (15 mg/kg) and midazolam (1 mg/kg) and mechanically ventilated after surgical tracheotomy with an inspired fraction of oxygen of 0.21, a tidal volume of 8 mL/kg, and a positive end-expiratory pressure of 5 cm H₂O delivered by using a Drager^TM ventilator. An internal jugular central venous line and a PiCCO^TM femoral arterial catheter were inserted under ultrasound guidance. A urinary catheter was surgically placed in the bladder by a midline minilaparotomy.

After surgical preparation, animals were stabilized for 30 min. Baseline measurements were performed, and piglets were then assigned to one of the 3 study groups: 6 piglets in the sham group (no LPS injection, no fluid), 9 in the control group (LPS infusion, no fluid), and 6 in the experimental crystalloid group (LPS infusion and fluid resuscitation with crystalloids). Basal hydration with polyionic glucose solution at 100 mL/h was infused during the whole experiment.

At T0, endotoxic shock was induced by infusing an Escherichia coli O111:B4 lipopolysaccharide (4%) at a dose of 150 μg/kg over 30 min [20]. At T120, 0.9% saline fluid resuscitation was started as a bolus of 20 mL/kg over 30 min followed by a 10 mL/kg/h infusion rate for 3.5 h in the crystalloids group.

A protocolized hemodynamic resuscitation was performed by trained intensivists in all study groups: a vasopressor support with norepinephrine and dobutamine was administered when necessary, to maintain mean arterial pressure (MAP) above 65 mm Hg, according to current recommendations on sepsis management [21]. Euthanasia was performed at the end of the experiment under general anesthesia with an overdose of thiopental (60–80 mg/kg) (Fig. 1).

From the transpulmonary thermodilution PiCCO^TM analysis, the following parameters were recorded at T0, T60, T180, and T360: cardiac output, global end diastolic volume, extravascular lung water, pulmonary vascular permeability index, indexed stroke volume, and systemic vascular resistances. From the pulse contour PiCCO^TM analysis, stroke volume variations, pulse pressure variations, and dP/dt max (an index of cardiac contractility) were collected. Additional hemodynamic data were recorded at T0, T60, T180, and T360: MAP, heart rate, central venous pressure, and urine output.

We used a probe-based confocal laser endomicroscope (Cellvizio®; Mauna Kea technology, Paris, France) with the concomitant administration of 1 mL of an intravascular dye, FITC dextran (70 kDa), to assess the mesenteric microcirculation. The procedure was standardized: the mesenteric network was visualized by affixing the laser probe in gentle contact with the mesentery without damaging the vessels, at the ileal segment level accessed from the minilaparotomy performed for the urinary catheter. The mesentery was maintained hydrated with warmed saline throughout microcirculation analysis. Cellvizio® was used at T0, T180, and T360. Images were recorded in vivo in a real-time manner by using the CellViewer software as digital sequences of at least 60-s duration (12 frames per second) [22] and analyzed thereafter using the CellViewer® software to collect FCD, mean vessel diameter, and total vessel length, according to the cardiovascular and dynamics section of the ESICM latest recommendations [23]. The vessel detection algorithm detects the medial axis and the border of tubular structures represented by FITC-induced contrast in a specified region of interest. The length, area, and diameter of the capillaries were calculated as a mean of these tubular structures. For quantification of FCD, the quotient of the capillary area divided by the total area of a given region of interest was calculated. Blood flow within the capillaries was also visually scored according to the POEM score as previously described [24].

In brief, the POEM scoring system is a 5-point ordinal scale that integrates assessments of both flow and heterogeneity. It enables assessing real-time microcirculatory parameters determined on 4 video clips from the same piglet at the same time point according to the POEM methodology, i.e., from 1 (critically impaired) to 5 (normal flow without heterogeneity) (online suppl. material 1; see www.karger.com/doi/10.1159/000519693 for all online suppl. material). In order to evaluate POEM score agreement and reliability, all videos were analyzed twice by the same observer blinded to study group allocation and results of first assessment. A second observer independently analyzed the videos and performed POEM scoring.

At T0 and T360, sVEGFR levels were measured in duplicate using a commercial enzyme-linked immunosorbent assay kit (Quantikine; R&D Systems, Minneapolis, MN, USA) according to the manufacturer’ s instructions.

Statistical analysis was performed using R software (version 3.3.2) [25]. Data are presented as medians and interquartile range (25th–75th percentiles) for continuous variables and absolute values and percentages for categorical variables. Groups were compared using the Mann-Whitney rank-sum test. Because of multiple testing, to ensure a global type I error of 0.05 (5%), for each test, the level of significance was set at 0.016 (1.6%) thanks to the Bonferroni’s correction [26]. To assess interrater and intrarater agreement for the POEM score, we computed the proportion of overall and specific agreement [27]. We checked the reliability of this scoring thanks to the weighted Cohen’s kappa [28, 29]. We also calculated 95% confidence intervals with a nonparametric bootstrap approach consisting in resampling 10,000 times with no replacement and using the accelerated bias-corrected percentile method [30].

At baseline, hemodynamic, biological, and microcirculatory parameters were similar between the 3 study groups (Table 1). No death occurred within the 6 h of the experiment.

MAP was maintained above 65 mm Hg during the 6 h of experiment for 90.5% of the piglets (19/21). At T360, CO was significantly lower in the control group compared to the sham group (p = 0.005) (Table 2). There was no significant difference in CO at T360 between the sham and crystalloids groups (p = 0.132). Urine output was significantly lower in both control and crystalloids groups compared to the sham group (p = 0.009 and p = 0.004, respectively).

We obtained analyzable images using the Cellvizio® endomicroscope in 90.5% of the cases (Fig. 2). We did not find any significant difference in FCD and mean vessel diameter between the 3 study groups at any time, whereas total vessel length was not evaluable due to respiratory variations (Table 3). At T360, the POEM score was significantly lower in the control and crystalloids groups (p = 0.005 and p = 0.011, respectively), whereas it remained unchanged in the sham group. The proportion of interrater agreement was 45.6% (95% CI: [33.3–57.9]). The weighted kappa coefficient was 0.68 (95% CI: [0.50–0.81]).

The proportion of intrarater agreement was 57.9% (95% CI: [45.6–70.2]). The weighted kappa coefficient was 0.84 (95% CI: [0.71–0.92]).

Biological data are shown in Table 4. At T360, bicarbonates were lower in the crystalloids group compared to the sham group (p = 0.008). At T360, sVEGFR1 levels were significantly higher in both control and crystalloids groups compared to the sham group (p = 0.001 and p = 0.004, respectively). There was no significant difference between control and crystalloids group (p = 0.8662) (Fig. 3).

To our knowledge, this is the first study reporting the effects of crystalloids infusion on mesenteric microcirculation assessed by probe-based confocal laser endomicroscopy (Cellvizio®) in early experimental sepsis. Our results show no significant improvement of microcirculatory alterations after crystalloids resuscitation despite improvement in macrocirculation. Microcirculation assessment tools should be considered to guide hemodynamic resuscitation in sepsis. We report the feasibility of using probe-based confocal laser endomicroscopy Cellvizio® in the experimental setting to better understand microcirculatory alterations.

Sepsis treatment guidelines [21] mainly focus on the normalization of macrocirculatory alterations, with less attention to restore microcirculatory dysfunction that has been shown to worsen outcomes [31]. Increased microcirculatory blood flow during sepsis resuscitation has been correlated with reduced organ failure at 24 h even while macrocirculation remains unchanged, supporting the hypothesis that therapeutic interventions should specifically target microcirculation goals to potentially improve organ failure in sepsis [32]. In septic patients, fluid resuscitation is a vital component of patient management. However, in our study, crystalloids fluid resuscitation did not improve microcirculatory dysfunction. These findings are in accordance with previous experimental studies focusing on the effects of fluid resuscitation on microvascular dysfunction in septic shock. Lopez et al. [33] previously reported persistent sublingual microvascular dysfunction despite early hemodynamic resuscitation using crystalloids infusion and vasopressors in an endotoxic shock. Similarly, Ergin et al. [34] demonstrated that fluid administration partially resolves LPS-induced shock whatever the type of fluid considered (0.9% NaCl and hydroxyethyl starch-ringer acetate solution), whereas Damiani et al. [35] showed that only 20% albumin restored microcirculatory hemodynamics. On the opposite, clinical studies found an improvement in microcirculatory parameters such as microvascular flow index after fluid therapy among patients with an initial significant alteration of microvascular flow index <2.6 [36] especially at the early phase of sepsis [37]. However, none of these studies have evaluated the impact of fluid resuscitation on the intestinal microcirculation. Experimental studies focusing on microcirculatory alterations in sepsis reported a disturbed regulation of regional tissue perfusion leading to mesenteric microcirculatory dysfunction [38]. Experimental sepsis is associated with ubiquitous changes in microcirculation, expressed as heterogenous microcirculatory flow, decreased FCD, and mismatched local oxygen supply and demand [39]. Studies in septic rats suggest these alterations may occur early, before the occurrence of macrocirculatory dysfunction [6, 40].

Several tools have been proposed to assess microcirculatory alterations during experimental sepsis. Handheld microscope devices using high-resolution image sensors and incident dark field imaging allowing automatic analysis of images have been widely studied [41]. Only one study reported microcirculatory assessment using the probe-based confocal laser endomicroscopy Cellvizio® in a porcine model of septic shock [15]. In this experimental study, the authors observed a decrease in FCD during sepsis that was reversible with fluid administration. These results differ from what we observed mainly because microcirculation was assessed at the mucosal perfusion level, while in the present study, we focused on studying mesenteric circulation. Furthermore, Schmidt et al. [15] used colloids resuscitation that has been shown to have greater effects on microcirculation compared to saline [42].

Interestingly, our results confirm the feasibility of using the Cellvizio® endomicroscope in the experimental setting allowing the in vivo and real-time visualization of pathological microcirculatory flows using a relatively simple score that would be easier to implement at bedside. It has been previously reported that a visual assessment by a trained operator is a reliable surrogate of a calculated value [43]. In the present study, interrater agreement was good especially for the extreme values of the POEM score (1 or 5) enabling to reproducibly categorize microcirculation in “altered” or “normal” microcirculation.

In addition to goal-directed resuscitation procedures aimed at optimizing systemic hemodynamic variables, a microcirculatory-guided fluid resuscitation procedure would be needed to ensure the ultimate purpose of fluid resuscitation, which is to maximize oxygen transport to the tissue cells by optimizing convective and diffusive transport of oxygen. Since hypovolemia induces convection limitation (low convective flow) and fluid overload diffusion limitation (large diffusion distance), defining the optimal fluid volume should be the first concern of the intensivist [44].

However, in the present study, the microcirculation remains profoundly altered in the groups of septic animals, whether they receive crystalloids or not. There was no relationship between changes in microvascular perfusion and initial arterial pressure or cardiac index or changes during fluid administration. These data confirm the loss of hemodynamic coherence observed between macro- and microcirculation in septic shock. Fluid therapy may not always be adequate to improve microcirculation dysfunction suggesting that other therapeutic interventions should be considered.

Current macrocirculatory goals do not seem sufficient for effective resuscitation, aimed at reducing organ failures. Previous studies suggest that improvement of microcirculation at the early phase of resuscitation is associated with a decrease in organ failure at 24 h [32]. These notions have brought out the concept of “microcirculation-guided fluid therapy” [44]. However, further studies are needed to confirm whether microcirculation-guided fluid therapy can improve organ dysfunction. Other treatments such as different vasopressors have been tried in order to restore normal microcirculation, without success [45]. Development of assessment tools easily implemented at bedside is another prerequisite to microcirculation-guided therapy in sepsis.

Unfortunately, lactate levels were not available during the experiment to correlate with microcirculatory alterations. Nevertheless, signs of organ hypoperfusion were recorded, and VEGFR1 serum levels were higher in both control and fluids groups, compared to the sham group. Yang et al. [46] showed that plasma sVEGFR1 levels had good correlation with organ hypoperfusion such as renal dysfunction, metabolic acidosis, and hematologic dysfunction.

We only studied crystalloids resuscitation as these fluids are recommended as the first-line treatment in international guidelines (SSC). However, it would be interesting to compare different type of fluids (crystalloids, colloids, and albumin, for example) and their effect on the POEM score, especially in an experimental setting [47‒49].

In our study, we used an endotoxic shock model, inducing early macro- and microcirculatory alterations and severe and lethal shock. These results cannot be extrapolated to all types of shock. We were not able to quantify the capillary leak.

Recommended microcirculatory parameters did not reflect significant changes in microcirculation in the present experimental study. However, the visual estimation of microcirculatory blood flow applying the POEM score is highly demonstrative of microcirculatory impairment as you may see in the video clips attached. The Cellvizio technology provides a good visualization of the microcirculation; however, microcirculatory evaluations such as FCD and capillary leak quantification could not be reliable due to difficulties in stabilizing images.

There was no significant improvement of microcirculatory alterations after crystalloids resuscitation despite improvement in macrocirculation hemodynamics, assessed by probe-based confocal laser endomicroscopy (Cellvizio®) in early experimental sepsis on a piglet model. Targeting macrocirculation and fluid therapy goals may not be sufficient to prevent and treat organ failures in sepsis. Microcirculation-guided fluid therapy should be considered for sepsis resuscitation.

We thank the Nîmes University Hospital and the Nimes University of Medicine for their support.

The Animal Care and Use Committee Languedoc-Roussillon (CEEA-LR-12013) approved the protocol, and all experiments were performed in an authorized animal research laboratory. All facilities and transport comply with current legal requirements.

The authors declare that they have no competing interests.

This study was supported by academic funding from the Nimes University Hospital.

C.D. analyzed and interpreted the data and wrote the protocol, C.R. also analyzed the videos and was a major contributor in writing the manuscript, B.L. performed statistical analysis, and G.L. and M.P. have carried out all the experiments.

The datasets used and analyzed during the current study are available from the corresponding author on reasonable request.