Macromolecular Capillary Leakage Is Involved in the Onset of Anaphylactic Hypotension

• Release of mediators such as histamine and nitric oxide may contribute to vasodilation and subsequent hypotension during anaphylactic shock

• Mediator release during anaphylaxis also incites increase in microvascular permeability

• Real-time, in situ  capillary permeability measured with confocal fluorescence microscopy by means of contrast-enhanced imaging was increased by 25-fold within minutes of the onset of anaphylactic shock

• Increase in microvascular permeability was correlated with a drop in arterial blood pressure, suggesting that these two phenomena may be mechanistically linked during early anaphylaxis

ANAPHYLACTIC shock has a distributive profile in which vascular dysfunction results in inadequate regional oxygen delivery.1Clinical guidelines for anaphylaxis recommend epinephrine as first-line therapy to counteract the vasoplegic component because of its beneficial effects mediated by α1- and β1-adrenergic receptors activation.2–4Nevertheless, the role of the hypovolemic component secondary to the inflammation-induced microcirculatory changes with subsequent intravascular fluid transfer into the interstitial space remains debated. However, it may participate in the inaugural anaphylactic arterial hypotension. The investigation of the microcirculatory function during anaphylaxis is therefore essential as its role is central for adequate tissue oxygenation and consequently organ function.5 

The in situ , real-time changes in capillary permeability can be evaluated by fibered confocal fluorescence microscopy (FCFM), by means of microscopic dynamic contrast-enhanced imaging technique. FCFM has allowed to assess changes in microvascular permeability in a limited number of experimental studies such as histamine-induced macromolecular leakage,6sepsis,7–8normal mesentery, and xenografted tumors.9A few experimental works have demonstrated that allergic activation of mast cells resulted in increased vascular permeability as well as leukocyte adhesion and platelet aggregation in the microcirculation, thereby amplifying the inflammatory response.10Microcirculatory changes during ongoing anaphylaxis have not been assessed using FCFM. Nevertheless, the volume loss was quantified in a few anaphylactic patients.11–12In these patients, changes in vascular permeability might permit a 50% transfer of the intravascular fluid into the interstitial space, within 15 min following anaphylactic shock onset.12 

Based on these studies, we hypothesized that the microcirculation could be relevant diagnostic and therapeutic targets for treatment of anaphylaxis. Accordingly, important questions need to be addressed. Does capillary leakage really occur in the course of anaphylactic shock? And, if so, in which time frame does it occur? Is this capillary leakage quantifiable? How is the kinetic profile of this capillary leakage? In order to answer these questions, we used a dynamic contrast-enhanced acquisition using FCFM in an anesthetized ovalbumin-induced Brown Norway rat model of anaphylactic shock to assess the microcirculatory changes during ongoing anaphylaxis.

We used 10-week-old male Brown Norway rats (Janvier, Le Genest-St-Isle, France). The animals were managed in accordance with the American Physiologic Society Institutional Guidelines and position of the American Heart Association on Research Animal Use. Animal care and use was performed by qualified individuals and supervised by a veterinarian, and all facilities and transportation complied with current legal requirements. The study was approved by the Animal Ethics Advisory Committee of the University Paris Descartes. Rats were sensitized by grade VI chicken egg albumin (Sigma-Aldrich, Saint-Quentin Fallavier, France) at day 0 (D⁰), D⁴and D¹⁴, as previously described.1,13–14 

The Cell-viZio device (Mauna Kea Technologies, Paris, France) is a fibered FCFM allowing in vivo , real-time fluorescence microscopy. Tissues are illuminated by a laser source at a defined wavelength. Fluorescent molecules, either spontaneously present or injected as a contrast agent, absorb this light and reemit a new fluorescent light at a longer wavelength. A 1.5-mm-diameter flexible probe consisting of a bundle of 30,000 optical fibers is placed in contact with the organ of interest, with a spatial resolution of 1.5 µm²in-plane, and a section thickness of approximately 15 µm. The laser unit provides an excitation wavelength of 488 nm and emission band wavelengths between 500 and 650 nm. The fluorescence signal is collected by the same fibers as those used for illumination, and video data are recorded at a temporal resolution of 12 frames/s. In the present study, we used fluorescein isothiocyanate-dextran (FITC-dextran), which is a macromolecular fluorescent contrast agent with a molecular weight of 70 kDa (Sigma-Aldrich). It has respective excitation and emission wavelengths of 488 nm and 520 nm, compatible with our imaging system.

The surgical procedure was performed under general anesthesia on day 21 (D²¹) using 60 mg/kg intraperitoneal sodium pentobarbital (Sigma-Aldrich) and maintained with intravenous additional doses (2 mg/kg) when required. Rectal temperature was maintained at 38 ± 0.5°C by intermittent warming with a heating blanket (Harvard Apparatus, Les Ulis, France). A fluid-filled polyethylene catheter (internal diameter: 0.58 mm, external diameter: 0.96 mm) (Harvard Apparatus) was inserted in the right common carotid artery for mean arterial blood pressure (MAP) and heart rate monitoring. Another fluid-filled catheter was inserted in the left external jugular vein for administration of drugs and fluid maintenance (10 ml . kg^–1. h^–1) with 0.9% sodium chloride injection USP (Baxter, Maurepas, France). Tracheotomy was performed and the lungs were mechanically ventilated with 100% O²as recommended by the different guidelines on perioperative anaphylaxis.2–4We used a Harvard Rodent respirator model 683 (Harvard Apparatus) (rate: 60 cycles/min, tidal volume: 1 ml). MAP and heart rate were recorded using a strain gauge catheter transduction coupled to a computer via  Emka hardware and IOX Software (EMKA Technologies, Paris, France) for continuous digital recording. Hemodynamic values were allowed to stabilize for 30 min (stabilization period). The mesentery was surgically exposed through a mid-abdominal incision for in vivo  visualization. All exposed tissue was covered with warmed and saline-soaked gauze to maintain tissue temperature and avoid tissue dehydration. The mesentery was placed on a homemade platform to minimize respiratory motion. The optical probe of the fluorescence microscope was placed in contact of the mesentery and stabilized, to ensure a steady image during and after FITC-dextran intravenous injection (150 mg/kg diluted in 200 µl 0.9% saline, administered over 30 s) allowing in vivo  visualization of the mesenteric microvascular network.9After a period of 5 min allowing for stabilization of the contrast agent concentration, a 5-s movie was recorded to obtain baseline signal intensity (SI) in capillaries and interstitium. The laser source was turned off between each movie to minimize tissue heating and contrast agent bleaching (i.e. , irreversible fluorescence loss when the contrast agent is submitted to high energy and prolonged illumination). Pre-bleaching was not performed because it would result in an important loss of signal. There would then not be enough fluorescence signal left in the tissues for imaging.

Animals were randomly allocated into two groups: control or anaphylaxis (n = 6 in each group), with 0.9% saline (500 µl) or ovalbumin (1 mg diluted in 500 µl saline) respectively, injected intravenously over 1 min. Time 0 (T0) corresponds to the beginning of saline or ovalbumin injection (shock onset). Preanaphylactic shock or control values of MAP and heart rate were recorded just before (T0–5 min/baseline) and during ovalbumin or saline injection (T0). Hemodynamic variables were then continuously measured until T0 + 30 min (fig. 1).

In both groups, baseline images of the mesenteric microcirculation were recorded at T0–5 min. Starting at T0, 5-s movies were then recorded every 30 s for 30 min (T0 + 30 min). Following the end of experiments, rats were killed by exsanguination. Visual examination of the movies was first performed, to evaluate the absence or presence of contrast agent leakage into the interstitium as well as its spatial distribution (homogeneous or heterogeneous). Raw amplitude movies were transferred to a workstation for processing using a Matlab-based (Matlab version 7.4 Mathworks, Natick, MA) custom-made software (PhysioD3D),15allowing measurements of fluorescent SI over time reflecting the quantity of contrast agent in the tissue. A region of interest, including both vessels and interstitium, was drawn on the whole image (90,195–92,070 pixels) and automatically propagated to all the images in the relevant movies. The region of interest propagation yielded a SI curve over time reflecting changes of contrast agent quantity in tissues during the experiment. SI was averaged over the first five images of each movie to improve the signal-to-noise ratio and was divided by the SI in the image at T0–5 (SI⁰), yielding normalized SI (SI/SI⁰) curves over time for each animal. The shape of the SI curves was analyzed for SI increase in total tissue reflecting capillary leakage from the vessels into the interstitium. Two parameters were calculated for quantification of capillary leakage: the delay between T0 and the beginning of capillary leakage, also called minimum time-to-leak (min); and the magnitude of contrast agent leakage was assessed as the ratio of perivascular intensity (Ip) to intravascular intensity (Ii) and expressed as follows: index leakage (%) = Σ[(Ip1/Ii1) + (Ip2/Ii2) +….. + (Ipn/Iin)]× 100/n; where n is the number of measured portions in each rat.16Thus, the SI within three different capillaries and contiguous interstitial areas was averaged at each of the following time-points: T0–5 min, T0, and every minute until T0 + 10 min.

Data are expressed as mean ± SD or mean ± SD (95% CI). This work is a pilot study designed to make a proof of concept that capillary permeability was increased in anaphylaxis, therefore no a priori  sample size calculation was performed. Intra- and between-groups comparisons of hemodynamic variables and index leakage were performed using a linear mixed model test in which the factors were group, time, and group x time (R, version 2.14.1, Vienna, Austria), taking into account the repeated measures for the factor time. Each time-point was compared with a baseline value (intragroup) and the synchronous time-point of the other group (between-groups). A Tukey post hoc  test was used to account for multiple comparisons. Individual values of index leakage were plotted against individual values of MAP for each time-point (T0–5 min, T0, and every minute until T0 + 10 min) and were correlated using a nonparametric Spearman rank test (Matlab version 7.4 Mathworks). All P  values were two-tailed and P < 0.05 was considered significant.

Twelve ovalbumin- or saline-treated Brown Norway rats (weight on the day of, but before shock induction: 302 ± 6 g) were studied (6 in each group). There were no missing data for the reported variables.

The MAP values did not significantly differ between the two groups during the preshock (T0–5 min/baseline) and T0 time-points (fig. 2). MAP values remained unchanged in the control group throughout the entire study (P = 0.22). In anaphylactic rats, MAP values showed a continuous decrease over time when compared with baseline, starting at T0 + 1 min (P = 0.0001). At T0 + 5 min and T0 + 30 min, mean MAP values showed a decrease of 60% and 70% compared with baseline (P = 0.0001), significantly lower when compared with control rats (P = 0.0001). The between-group difference (P = 0.0001) appeared at T0 + 1 min and persisted thereafter.

Heart rate remained unchanged in both groups from T0–5 min through T0 + 5 min. Heart rate values remained unchanged in the control group throughout the entire study. In anaphylactic rats, a significant (P = 0.04) decrease was observed by T0 + 10 min compared with baseline heart rate, and this decrease persisted for the remainder of the in vivo  recording (data not shown).

In the control group, FITC-dextran remained in the capillaries during the entire study period (fig. 3Aand 3B). In anaphylactic rats, a progressive fluorescence increase in the interstitium was observed over time (figs. 3Cand 3D). In both groups, SI curve showed an early decrease probably because of initial bleaching of the contrast agent. In controls, no subsequent SI increase was observed over time. In anaphylactic rats, the initial SI curve decrease was followed by a constant and progressive global increase reflecting the increase in fluorescence in the interstitium (fig. 4). This increase of SI in the interstitium did not allow differentiation of interstitium from capillaries after 10 min. Finally, in four anaphylactic rats, a heterogeneous distribution of the contrast agent extravasation was observed, with interstitial areas of high fluorescence enhancement and others where no leak of contrast agent into the interstitium could be seen.

In the control rats, mean index leakage values did not significantly differ from baseline at each time-point (P = 0.45). In anaphylactic rats, mean index leakage showed a significant difference from baseline starting at T0 + 3 min (P = 0.0001), with a mean minimum time-to-leak of 5 ± 3 min (95% CI: 4–12 min) after shock onset. Mean index leakage calculated at each time-point differed between control and anaphylactic rats starting at T0 + 2 min (respectively, 20 ± 11 vs.  170 ± 127%; P = 0.0001) and progressively increasing (fig. 5) up to 25-fold during the time of observation. A good correlation was found between individual values of index leakage and arterial blood pressure in both groups (r = –0.75, P = 0.0001) (fig. 6).

The main findings of this study were: interstitial capillary leakage occurred a few minutes after anaphylactic shock onset, and this macromolecular capillary leakage showed up to 25-fold increase when compared with control conditions. Interestingly, this leakage correlated to the drop in arterial blood pressure during the early stage of anaphylaxis. These results have both mechanistic and therapeutic implications.

For this study, we selected FITC-dextran 70 KDa, as its molecular weight is close to that of serum albumin, therefore mimicking physiologic macromolecules. Like albumin, FITC-dextran remains in the vascular compartment under normal conditions.9In cases of increased permeability, the macromolecule leaks and accumulates in the interstitium.6Real-time changes in capillary permeability underlying anaphylactic shock were visualized by means of FCFM that provides in vivo  microvascular observations to evaluate the magnitude of macromolecular leakage as an indicator of endothelial alterations.

Macromolecular capillary leakage started within the first 3 min after anaphylactic shock onset. It was progressive, after-shock onset with an up to 25-fold increase during the first 10 min of observation, and was followed by a further global capillary leakage in the mesenteric compartment during the remaining study period. A possible explanation of our findings could have been that excitation of FITC-dextran per se  influences capillary permeability.17However, under control conditions, macromolecular leakage did not develop over time, suggesting that the responses were specific to anaphylaxis and not to the illumination itself. Interestingly, the minimum time-to-leak and the onset of cardiovascular disturbances occurred within the same time frame, i.e. , 1 or 2 min, after shock onset, suggesting a relationship between the two phenomena. Accordingly, the magnitude of the macromolecular capillary leakage was highly correlated to the decrease in arterial blood pressure. It was consistently low in controls, whereas in anaphylactic rats, it progressively increased with the arterial blood pressure drop, starting 3 min after shock onset and persisting until 10 min after. Past the first 10 min of observation, massive extravasation did not allow further analysis of SI. These findings are consistent with previous clinical reports where the component resulting from plasma losses, 10 and 15 min after anaphylactic shock onset, has been estimated to be respectively as high as 35% and 73% of the circulating blood volume.11–12Others have reported the occurrence of bradycardia, also called “paradoxical bradycardia,” despite severe hypotension in as many as 10% of patients experiencing perioperative anaphylaxis.3This “paradoxical bradycardia” allows the ventricles to fill before they start contracting again despite a massive hypovolemia.18–19Consequently, the occurrence of bradycardia during the early course of anaphylaxis might reflect the ongoing marked macromolecular capillary leakage with subsequent hypovolemia. We therefore suggest that the inflammation-induced microcirculatory changes with subsequent intravascular fluid transfer into the interstitial space might be involved in the onset of anaphylactic hypotension. However, our study setup does not allow us to prove this.

Finally, in most (66%) anaphylactic rats, the progressive enhancement in the interstitium had a heterogeneous spatial distribution, with zones seeming to leak more intensively than others. This may reflect differential changes in capillary permeability according to microvessels similar to spatial heterogeneity of capillary perfusion and decrease in functional capillary density as reported during sepsis.20 

Anaphylaxis is an acute inflammatory IgE-mediated response to a foreign antigen because of massive release of inflammatory preformed mediators from sensitized mast cells and basophils such as histamine and de novo  synthesized mediators.19One of the major objectives of this study was to more closely examine the contribution of the microvascular alterations elicited by anaphylactic shock. Previous studies showed transient increase in venular permeability in the rat mesenteric microcirculation, beginning 1 or 2 min after histamine suffusion, with peaks between 5 and 15 min after, and return to control levels at 20–30 min.21These changes were attributable to venular junctional gaps via  activation of endothelial H¹-receptors.22But other mechanisms may also be involved.22Histamine also leads to increased nitric oxide production from endothelial cells,23whereas endothelial nitric oxide synthase increases microvascular permeability to macromolecules in response to inflammatory agents.24Thus, endothelial nitric oxide synthase has been suggested to be the primary vasodilatator during anaphylactic shock,25suggesting that histamine might induce an increase in microvascular permeability to macromolecules via  nitric oxide26at an early stage of anaphylaxis. Further experimental studies are required to validate these hypotheses.

There are some limitations to our study. First, the experimental model and the study design were chosen to reproduce the clinical conditions experienced by patients under anesthesia when anaphylactic shock occurs. We therefore deliberately decided to perform our study in rats experiencing true in vivo  anaphylaxis and not in a model that attempts to mimic acute inflammation-like conditions using histamine as previously shown.23However, extrapolation of our results to humans should be cautious. Second, our study allows a proof of concept but does not elucidate the mechanisms of capillary leakage occurring during anaphylaxis. Third, optical imaging signal intensity is the result of a combination of complex phenomena including reflection, absorption, and fluorescence of photons. Despite turning off the laser between acquisitions, an important SI decrease was observed in the first minute because of photobleaching of the contrast agent. This signal loss was so intense that it precluded detection of any other optical phenomenon, and it is therefore possible that capillary leakage of macromolecules occurred before the minimum time-to-leak we observed but was undetectable. The main limit of the complex relationship between SI and concentration of contrast agent is that it does not allow absolute quantification of volume or rate of macromolecular leakage. In the present study, macromolecular leakage was qualitatively analyzed by SI curves over time and semiquantitatively using the index leakage. Finally, it may be noted that the index leakage is not null in baseline conditions (approximately 20% before saline or albumin injection), because of the presence of spontaneous signal in the interstitium despite the absence of contrast agent.27 

The present study suggests therapeutic implications. Because capillary leakage occurs immediately after anaphylactic shock onset and correlates to the drop in arterial blood pressure, it may contribute to the decreased preload and the onset of inaugural hypotension. The exact requirement for fluid therapy remains unknown during anaphylaxis and has not been investigated. Although the use of crystalloids followed by colloids is usually recommended by current clinical guidelines,2–4some authors estimate that there is no evidence that one is better than the other.3 

In conclusion, we provide information on microcirculatory changes during ongoing anaphylaxis and demonstrate that in this experimental model, increased capillary leakage occurs early during anaphylactic shock. In addition to the vasoplegic component of anaphylaxis, the inflammation-induced microcirculatory changes with subsequent intravascular fluid transfer into the interstitial space (hypovolemic component) might be involved in the onset of anaphylactic hypotension. Further work is needed to study the mechanisms of capillary leakage during anaphylaxis.