Regional Microvascular Function and Vascular Reactivity in Patients with Different Degrees of Multiple Organ Dysfunction Syndrome

Abstract

The pathophysiology of multiple organ dysfunction syndrome (MODS) is believed to be related to that of microcirculatory dysfunction. We hypothesized that the severity of MODS is determined by measuring regional variables of microvascular function and vascular reactivity in critically ill patients. Therefore, we compared (a) reactive hyperemia response in the forearm using transcutaneous Po₂/Pco₂ electrodes and laser Doppler velocimetry, (b) microvascular permeability assessed by strain-gauge plethysmography in legs, and (c) variables derived from gastric tonometry in hemodynamically stable patients with moderate (n = 15) and severe (n = 15) MODS. There were no differences in systemic oxygen delivery, consumption, and oxygen extraction ratio between the groups. Mortality was 20% in patients with moderate MODS and 60% in patients with severe MODS (P = 0.025). Patients with a high MODS score had significantly larger arterial lactate concentrations (3.81 ± 2.7 mmol/L) than patients with moderate MODS (1.66 ± 0.82 mmol/L; P = 0.006). No significant differences in gastric pHi, gastric regional-to-arterial Pco₂ difference, capillary filtration coefficient, isovolumetric venous pressure, and skin reactive hyperemia response were observed between patients with moderate and severe MODS. Once MODS is established, regional variables of microvascular function and vascular reactivity measured in this study do not reflect severity of organ dysfunction.

In Brief

IMPLICATIONS: This study shows that regional variables of microvascular function and vascular reactivity assessed by different measurement techniques do not reflect the severity of multiple organ dysfunction syndrome in hemodynamically stable, resuscitated, surgical intensive care patients with different mortality rates.

Microcirculatory dysfunction is believed to be a significant contributor to the pathophysiology and development of multiple organ dysfunction syndrome (MODS). In particular, patients with sepsis and septic shock often present with increased levels of global systemic oxygen delivery. At the same time, increased blood lactate concentrations suggest impairment of tissue oxygen delivery at the microvascular level (^1–4).

Several studies of microcirculatory function in critically ill patients have demonstrated significantly reduced regional microvascular functional capacity. Diminished reactive hyperemia (RH) response in the skin has been demonstrated in patients with cardiogenic or septic shock, as well as in surgical intensive care patients (^5–7). In one study, maximal vasodilator response to acetylcholine and sodium nitroprusside in the skin of patients suffering from septic shock was severely depressed in comparison with nonseptic patients and healthy controls (⁸). In patients with septic shock, severe alterations in microvascular blood flow within the tongue have been observed using orthogonal polarization spectral imaging and have been related to subsequent outcome (^9,10).

Interestingly, most previous studies on variables of microcirculatory function focused on the very acute stage of the disease, or were performed on critically ill patients who were still under continuous resuscitation. However, after resuscitation and initiation of appropriate surgical or conservative therapy, some patients enter a phase of stability, with varying degrees of MODS but stabilized hemodynamics and gas exchange.

Because the severity of MODS is directly linked to increasing mortality, one might also hypothesize that the degree of regional circulatory derangement is closely associated with the severity of MODS. Therefore, this prospective, clinical study compares variables of regional microcirculatory function and vascular reactivity in patients suffering from either moderate or severe MODS, thus representing two patient populations with a significant difference in intensive care unit (ICU) mortality. We measured vascular reactivity as RH response in the forearm using transcutaneous Po₂/Pco₂ electrodes and laser Doppler velocimetry, microvascular permeability of the limb using venous congestion plethysmography (VCP), and variables derived from gastric tonometry.

Methods

The study protocol was approved by the ethical committee of the Leopold-Franzens-University of Innsbruck. Written informed consent was obtained, if possible, from all subjects or otherwise from the closest family members. Thirty resuscitated patients with stable hemodynamics suffering from MODS and admitted to a 12-bed surgical ICU were prospectively investigated (^1,11). Before entering the study, all patients were fluid-resuscitated until stroke volume could no longer be increased; pulmonary capillary wedge pressure was then used for the guidance of additional fluid resuscitation. When mean arterial blood pressure remained <70 mm Hg, a norepinephrine infusion was started. When stroke volume index remained <25 mL · min⁻¹ · m² or cardiac index was <2.0 L · min⁻¹· m², a milrinone infusion was started. All patients were mechanically ventilated, analog-sedated, and invasively monitored, including a pulmonary artery catheter at the time of investigation. Patients suffering from moderate MODS (Group I) were investigated after being in the ICU for at least 16 h, whereas patients suffering from severe MODS (Group II) had been in the ICU for at least 24 h before study inclusion. No study patient received hydrocortisone infusion or activated protein C.

Patients were only included in the study protocol if they had no history of peripheral arterial vascular occlusive disease, deep venous thrombosis, diabetes mellitus, traumatic injury to the lower extremities or spinal cord, or previous polyneuropathy. In addition, patients with hypoxia (Pao₂ ≤ 60 mm Hg) or acidosis (pH ≤ 7.35) were excluded. Patients were divided according to severity of MODS into 2 groups: Group I (n = 15; moderate MODS) included patients with a MODS score ≤8 and an expected mean mortality rate of 3.8%. Group II (n = 15; severe MODS) included patients presenting with a MODS score >8 and an expected mean mortality rate of 58% (Fig. 1). Mortality data are collected from data on ICU mortality of 2783 patients admitted to our ICU during the last 4 yr.

Figure 1.: Mortality and multiple organ dysfunction syndrome (MODS) score. Figure 1 presents data on intensive care unit (ICU) mortality of 2783 patients admitted to the Division of General and Surgical Intensive Care Medicine between 1999 and 2002. Patients with a MODS score ≤8 had a highly significant mortality rate when compared to patients with a MODS score >8 (3.81% [94 of 2469] versus 61.78% [194 of 314]; P < 0.005).

The following data were collected from all patients: age, sex, preexisting comorbidity, admission diagnosis, the ASA Classification Score, the Simplified Acute Physiology Score calculated from worst physiologic values within the first 24 h after ICU admission, and the presence or absence of systemic inflammatory response syndrome (SIRS) or sepsis (^12,13).

Hemodynamic variables included heart rate, mean arterial blood pressure, central venous pressure, cardiac index, stroke volume index, mean pulmonary artery pressure, and pulmonary capillary wedge pressure. Arterial and mixed venous acid base status, blood oxygen tension, and arterial lactate concentrations were also determined (Rapidlab 860; Chiron Diagnostics, Medfield, MA). Systemic oxygen delivery index, systemic oxygen consumption index, extraction ratio, and systemic vascular resistance index were calculated according to standard formulas. Laboratory examinations were determined once a day to evaluate hepatic, renal, and hematologic organ function.

RH response after arterial occlusion was measured in the patient's forearm using a transcutaneous Po₂/Pco₂ (Ptco₂/Ptcco₂) combi-electrode (TCM3, Radiometer, Copenhagen, Denmark) heated to 37°C and by laser Doppler velocimetry (Periflux 4001, Perimed, Järfälla, Sweden).

Each measurement started with the calibration of the Ptco₂/Ptcco₂ electrode according to the manufacturer's recommendation. The electrode was then placed on the volar aspect of the forearm using a self-adhesive ring. A solution of 1,2-propanediol in distilled water was used to obtain better contact with the surface. RH can be best recorded at 37°C, at which the coefficient of variation has been reported to be 14% and the correlation coefficient to be 0.90 between duplicate measurements (¹⁴).

Skin microvascular blood flow was assessed by laser-Doppler velocimetry. Laser-Doppler measurements are based on the principle that light, scattered by moving red blood cells, experiences a frequency shift proportional to the velocity of the red blood cells. The Periflux 4001 uses laser light with a wavelength of 770–790 nm. A fiberoptic guidewire (PF407; Perimed) conducts laser light to the tissue and carries back-scattered light to a photodetector. This was placed near the Ptco₂/Ptcco₂ electrode and attached using a self-adhesive ring, as described before. Skin microvascular blood flow was recorded in relative perfusion units (PU). The probe was calibrated against a white surface (PU = 0) and a standard latex solution (PU = 250 ± 5). Forearm ischemia was produced with a sphygmomanometer cuff wrapped around the arm over the brachial artery and inflated to 300 mm Hg for 5 min.

After a resting period of 30 min, preocclusive baseline Ptco₂/Ptcco₂ (B-Ptco₂/B-Ptcco₂) and baseline PU (B-PU) were recorded for 3 min. During reperfusion, postischemic peak Ptco₂/Ptcco₂ (PIP-Ptco ₂/PIP-Ptcco₂), and postischemic peak PU (PIP-PU) were measured. To determine the magnitude of RH, the differences between PIP-Ptco₂ and B-Ptco₂ and PIP-PU and B-PU were calculated. In addition, the elimination rate of carbon dioxide from the skin (ERCO₂) was determined over a period of 3 min according to the formula ERCO₂ = (PIP-Ptcco₂– posthyperemic-Ptcco₂ at 3 min)/(PIP-Ptco₂− B-Ptcco₂) (¹⁵).

For VCP measurements, patients were positioned horizontally and supine. Fluid filtration capacity (K_f) and isovolumetric venous pressure (Pv_i), i.e., venous pressure where capillary filtration starts to increase above normal, were assessed using an electromechanical strain gauge sensor with automated calibration (Filtrass 2001, Domed Medizintechnik GmbH, Munich, Germany). For VCP measurements, the gauge was placed around the limb at a site of measured circumference and automatically stretched to predetermined tension. Changes in the resistance of the gauge result from alterations of limb girth by increasing venous congestion upstream to the strain gauge. Venous congestion was achieved by a tight cuff, coupled to a 0.1 L/s air pump. Cuff pressure was monitored using a pressure transducer. Both VCP and cuff pressure signals were simultaneously monitored. Changes in limb circumference were measured with the strain gauge and continuously recorded by computer. During a continuous series of increasing pressure steps (each pressure step was 10 mm Hg in magnitude, maintained for 270 s; maximum pressure did not exceed diastolic blood pressure), the vascular compliance component and the fluid filtration component of the increase in limb circumference were analyzed. At each pressure step, the cuff pressure in mm Hg and the slope of the slow volume change indicating increased capillary filtration (J_v; mL × 100 mL⁻¹ × min⁻¹) were recorded. The values of J_v, when plotted against corresponding cuff pressures, showed linear dependency. The interception with the x-axis reflects the Pv_i, defined as the vascular pressure where increased capillary filtration starts. The slope of the plotted line corresponds with the microvascular K_f (¹⁶).

Gastric mucosal perfusion was measured using automated recirculating air tonometry combined with capnometry (Tonocap TC-200, Tonometrics, Helsinki, Finland). The Tonocap system analyses the CO₂ content by infrared absorption at preset intervals. All study patients were receiving H₂-blocker treatment for a period of at least 24 h before measurements were taken. In addition, enteral feeding was discontinued 2 h before measurements were taken. Intramucosal pH (pHi) and regional and arterial Pco₂ gap (Pco₂ gap) were calculated from regional measurements and an arterial blood gas analysis.

The measurements were discontinued when stable Ptcco₂, Ptco₂, and PU values were recorded for a period of at least 5 min after reperfusion.

Because the trial was considered to be a pilot study, no sample size estimation was performed. Continuous demographic and clinical data were compared with a paired Student's t-test for Gaussian distribution and with a Wilcoxon's signed rank test for non-Gaussian distribution. A software program was used for data analysis (SYSTAT, Systat Software Inc., Point Richmond, CA). Significance was assumed at P < 0.05. All data are given as mean ± sd, if not indicated otherwise.

Results

Characteristics of patients with moderate (MODS score, 5.3 ± 1.5) and severe (MODS score, 10.5 ± 1.3; P ≤ 0.001) MODS are presented in Table 1. There were no differences in age, sex, and the incidence of SIRS and sepsis/septic shock between groups. Patients with severe MODS demonstrated a significantly higher preoperative ASA score and a three times more frequent ICU mortality (P = 0.025) when compared to patients suffering from moderate MODS. Patients with moderate and severe MODS spent 8 ± 8 days and 10 ± 8 days in the ICU before entering the study, respectively.

Table 1: Characteristics of Patients with Moderate or Severe Multiple Organ Dysfunction Syndrome

Patients with severe MODS demonstrated significantly higher central venous pressures, arterial lactate concentrations, and norepinephrine requirements when compared with patients suffering from moderate MODS (Table 2). There were no differences in mean arterial blood pressure, pulmonary artery pressure, pulmonary capillary wedge pressure, cardiac index, stroke volume index, systemic oxygen transport variables, blood gas variables, or milrinone requirements between groups.

Table 2: Hemodynamics and Systemic Oxygen Transport Variables in Patients with Moderate and Severe Multiple Organ Dysfunction Syndrome at Study Entry

There were no differences between groups in baseline values of Ptco₂ and Ptcco₂, B-PU, PIP-Ptco₂, Ptcco₂, postischemic skin blood flow, the magnitude of the RH response, and the elimination rate of CO₂ within the first 3 min of restoration of arm blood flow (Table 3). There was an insignificant trend towards increased microvascular permeability in patients with severe MODS (P = 0.062). No differences in isovolumetric venous pressure and gastric tonometry-derived variables were detected.

Table 3: Microcirculatory Measurements in Patients with Moderate and Severe Multiple Organ Dysfunction Syndrome

Discussion

We found that regional variables of microvascular function and vascular reactivity measured in this study did not reflect severity of MODS in hemodynamically stable, resuscitated surgical ICU patients.

The extent to which microcirculatory dysfunction affects the development of MODS in critically ill patients is still a matter of intensive discussion. Abnormalities of the microcirculation have been repeatedly demonstrated in various animal models of endotoxemia and bacteremia, whereas other investigations have failed to prove a clear link between microcirculatory failure and tissue oxygen depletion leading to cellular energy failure until the very late stages of septic shock (^2,5,17,18). Some experimental evidence suggests that MODS might, at least in part, be the consequence of enhanced cellular apoptosis or defective mitochondrial respiration (^19,20). Unfortunately, the relative contribution of different pathophysiologic events to the development and progression of MODS remains unknown.

In this study, regional variables of microvascular function and vascular reactivity were assessed using different technologies in stable, resuscitated patients suffering from moderate or severe MODS. At the time of investigation, patients with moderate MODS had a mean of 2 to 3 failing organs, representing a mortality rate of 20%, whereas patients with severe MODS demonstrated a mean of 5 to 6 failing organs, with a subsequent mortality rate of 60%, in our institution.

Studies using different technologies have shown microcirculatory abnormalities in the skin, skeletal muscle, gastrointestinal mucosa, and the tongue of patients suffering from different shock states (^{5,6,9,10,21,22}). Other studies have reported oxygen supply dependency of oxygen consumption for whole body and regional circulation in some critically ill patients, indicating an alteration in microcirculatory function in septic patients (⁹).

Interestingly, we found no differences in vascular reactivity, assessed as the degree of RH response to a defined period of forearm ischemia in the skin and no significant differences in tonometrically derived variables, in particular the Pco₂ gap, representing the overall balance of oxygen supply to demand within the gastric mucosa, between patient groups (^23,24). In addition, there was no difference in the isovolumetric venous pressure between groups, demonstrating that enhanced capillary filtration seems to be more than identical venous filling pressures. However, patients with severe MODS showed a trend towards increased capillary permeability (Table 3).

RH is a measure of regional vascular reactivity in response to tissue hypoxia involving capillaries, arterioles, and small arteries. The increase in regional blood flow after vascular occlusion is directly related to the severity and duration of ischemia (²⁵). Several clinical studies have reported significantly diminished RH response in patients with severe sepsis, septic shock, and cardiogenic shock (^5–7). There is a significant correlation between the magnitude of acetylcholine-induced skin vasodilatation and peak RH response in healthy volunteers (²⁶). Therefore, RH may be limited in patients with SIRS or sepsis by an impaired endothelium-dependent vasodilatation of the regional microvasculature in response to an ischemic stimulus. Decreased endothelium-dependent relaxation has been demonstrated after exposure of vascular endothelium to cytokines in vitro and in patients with systemic cardiovascular diseases (^6,27,28).

Gastric tonometry as a monitoring method to assess adequacy of gastrointestinal oxygen supply has received special attention in recent years. Several authors have reported that tonometrically derived variables, and, in particular the Pco₂ gap, may present an early indicator of mucosal ischemia in critically ill patients (²⁹). In addition, an increased Pco₂ gap has been associated with an unfavorable outcome (²⁴). In contrast, no significant differences in Pco₂ gap were observed between groups in the present study. Considering a normal range for the Pco₂ gap of between 7 and 12 mm Hg, only 3 of 15 patients suffering from severe MODS demonstrated increased values of up to 17 mm Hg. These results are in line with other studies demonstrating no differences in Pco₂ gap or gastric intramucosal pH between survivors and nonsurvivors of critical illness (³⁰).

Increased capillary permeability resulting in tissue edema is a key event observed in certain critically ill patients suffering from different states of shock (³¹). Causes of edema formation include protein leakage, separation of tight junctions between endothelial cells, loss of negative surface charges on the plasma membrane of capillary endothelial cells, and direct endothelial cell injury and disruption (^3,32,33). Development of tissue edema is thought to aggravate microcirculatory failure by increasing diffusion distances and through mechanical compression of capillaries and venules. Increased intestinal permeability has been demonstrated in patients subsequently developing MODS (³⁴). Using VCP, microvascular permeability has been shown to be significantly higher in patients presenting with cardiovascular instability because of infection when compared with patients who were hemodynamically unstable because of other pathophysiologic conditions (³⁵). The present study shows a trend towards higher capillary permeability in patients suffering from severe MODS. However, venous pressure at the point where capillary filtration starts to increase was identical in both groups and was much higher than cardiac filling pressures present during the observation period.

At the time of investigation, there were no differences in systemic oxygen transport variables between groups. However, norepinephrine requirements were significantly larger in patients suffering from severe MODS. In addition, the serum lactate concentration was twice as large, without changes in arterial pH, in patients with severe MODS.

Our study differs markedly from previous studies on microcirculatory function in critically ill patients. We selected patients who were already resuscitated, with stable hemodynamics and gas exchange but with varying degrees of MODS under constant therapy. In addition, we measured several variables of microvascular function and vascular reactivity, assessing regional perfusion, functional vasodilatory capacity, and microvascular endothelial function at the same time. However, our finding of nearly identical regional microvascular function and vascular reactivity between patients with varying severity of MODS does not eliminate the possibility that severe microcirculatory dysfunction is involved in the pathogenesis of MODS.

One general drawback of studies investigating microvascular phenomena in diseased animals and humans is the problem of heterogeneity of regional blood flow and metabolic changes, not only when comparing different organs but also within one particular organ system (³⁶). There may be significant differences in microvascular function and vascular reactivity when comparing tissues or organ systems. Therefore, measurements in the skin, limbs, and stomach may not be representative of other organs. Furthermore, the methods applied in this study do not allow any direct observation of the microvascular bed or permit direct measurements of metabolic changes associated with microcirculatory dysfunction.

The results of our study contradict those of a recent investigation using an orthogonal polarization spectral imaging device, which demonstrated persistent microcirculatory alterations in patients suffering from septic shock and multiple organ failure (¹⁰). Unfortunately, major differences in hemodynamic variables and clinical progression of disease make any comparison of results difficult. In the study of Sakr et al. (¹⁰), most patients (13 of 20) died because of persistent septic shock, and only 7 patients died because of persistent MODS. Presumably, the latter patients more likely resemble patients investigated in our study. In addition, mean systemic blood flow and oxygen delivery were approximately 25% and 35% lower in the study of Sakr et al., which may also have had an impact on final results.

Increased blood lactate concentration is often viewed as evidence of tissue hypoxia, and hence the increased lactate values observed in patients with severe MODS in this study might be interpreted as evidence for tissue hypoxia. However, using the microdialysis technique in patients suffering from septic shock, a recent investigation elegantly demonstrated that increased blood lactate concentrations most likely result from exaggerated aerobic glycolysis through Na⁺-K⁺-adenosinetriphosphatase (ATPase) stimulation, in particular, in skeletal muscle during septic shock (³⁷). Inhibition of Na⁺-K⁺-ATPase by perfusion of microdialysis probes with ouabain, a specific inhibitor of Na⁺-K⁺-ATPase, stopped overproduction of muscle lactate and pyruvate (³⁷).

Other drawbacks of this study may be the small sample size of patients and the lack of measurements in a matched population of healthy volunteers for comparison of variables of microvascular function and vascular reactivity. Because the study was planned as a pilot study, no sample size estimation was performed, and sample size was limited a priori to 15 patients in each group. However, for most variables studied, the results suggest that many more patients would have been required to detect a significant difference. Assuming an α error of 5%, we have calculated that, for example, in the case of capillary filtration coefficient, at least 30 patients in each group would have been required to detect a significant difference with a power more than 80%. Although a comparison of variables with an appropriately matched population of healthy volunteers would have been interesting to demonstrate general differences in regional variables of microvascular function, such a comparison would have merely changed the fact that once MODS is established, regional variables of microvascular function and vascular reactivity do not reflect severity of organ dysfunction.

Despite the above-mentioned limitations, our results challenge the hypothesis that decreased regional tissue oxygen supply caused by alterations in microvascular function and vascular reactivity maintain MODS severity once patients are fully resuscitated and hemodynamically stable. Our study is in line with recent investigations pointing out the importance of mitochondrial dysfunction leading to cytopathic hypoxia or increased apoptosis as significant pathophysiologic events determining severity and duration of the MODS (^38–41). However, our results do not exclude a predominant role of severe microcirculatory derangements in the pathogenesis of MODS, in particular at the beginning of critical illness. In addition, we can also not exclude the possibility that significant microcirculatory derangement may persist in certain patients with established MODS.