Section 1：Splanchnic microcirculatory changes during hemorrhagic shock and
resuscitation in a rat model
1.1 Animal study I: Microcirculatory changes in multiple splanchnic organs during hemorrhagic shock and saline resuscitation.
Macrocirculatory Changes Secondary to Hemorrhagic Shock
The HR did not significantly change in each group. The MAP significantly decreased after hemorrhagic shock (MAP at T1, T2, and T3 was significantly lower than that at T0; p < 0.001), whereas there was no significant MAP change in the S group (Table 1). After fluid resuscitation, the MAP at T1 to T3 was comparable in the S group (Table 1).
Splanchnic Organs and Gracilis Muscle Microcirculatory Blood Flow Intensity Changes Secondary to Hemorrhagic Shock and Fluid Resuscitation
The sequential absolute value changes in microcirculatory blood flow intensity during the measurement period is summarized in Table 2. The sham operation induced no significant changes in microcirculatory blood flow intensity in each investigated organs (Fig 6A and 6D). By comparison, hemorrhagic shock result in a significant decrease in microcirculatory blood flow intensity in the kidney and intestine
(including the mucosa, serosal muscular layer, and Peyer’s patch). Hemorrhagic shock induced heterogeneous sequential percent changes in microcirculatory blood flow
36
intensity among multiple splanchnic organs, with the following reduction differences:
intestine (intestinal mucosa > serosal muscular layer, Peyer’s patch) > liver, kidney, and gracilis muscle (p < 0.05; Figure 7A, 7C, and 7E; Figure 6B and 6E). Compared to the H group, fluid resuscitation only restored the microcirculatory blood flow intensity impairment of the kidney and partially restored the microcirculatory blood flow intensity impairment of intestinal mucosa at T3 (Figure 7E, 6E, and 6F). By comparison, after fluid resuscitation, the intestinal microcirculatory blood flow intensity remained compromised compared to the baseline condition, especially in the mucosa (Table 2; Figure 7E and 6F).
Splanchnic Organs and Gracilis Muscle Tissue Oxygen Saturation Changes Secondary to Hemorrhagic Shock and Fluid Resuscitation
The sequential absolute value changes in tissue oxygen saturation during hemorrhagic shock is summarized in Table 3. There was no significant tissue oxygen saturation change among the organs measured in the S group. Hemorrhagic shock induced significant homogeneous changes of tissue oxygen saturation in each investigated target organ; the reductions in the tissue oxygen saturation of the various splanchnic organs were comparable but significantly greater than those of the gracilis muscle (p
< 0.05; Figure 7B, 7D, and 7F). Compared to the H group, fluid resuscitation did not
37
significantly improve tissue oxygen saturation (Figure 7B, 7D, and 7F).
38
1.2 Animal study II: Effects of different types of fluid resuscitation for hemorrhagic shock on splanchnic organ microcirculation and renal reactive oxygen species formation.
Part I. Changes of microcirculatory blood flow intensity in splanchnic organs Microcirculatory blood flow changes
Changes in the sequential absolute values of the intensity of microcirculatory blood flow during the measurement period are summarized in Table 4. There is no
significant changes in the intensity of microcirculatory blood flow in all investigated organs in the sham group (Table 4; Figs. 8 and 9A–9F). The microcirculatory blood flow intensities in the liver and gracilis muscle did not significantly change after hemorrhagic shock and remained no significant changes after fluid resuscitation (Table 4; Fig. 9A and 9F). Hemorrhagic shock significantly lowered the intensity of microcirculatory blood flow in the kidney and intestine (including the mucosa, serosal muscular layer, and Peyer's patch; Figs. 8B and 9B–9E).
Fluid resuscitation in the NS, HTS, GEL, and HES groups restored intensity of microcirculatory blood flow at T2 (Figs. 8C–8F and 9B). Fluid resuscitation by NS significantly improved microcirculatory blood flow intensities in intestinal mucosal and Peyer's patch compared with that in the control group (Fig. 9C and 9E). However, the microcirculatory blood flow intensity of the intestine (including the mucosa,
39
serosal muscular layer, and Peyer’s patch) in the NS group remained significantly impaired compared with that in the sham group at T2 (Figs. 8C and 9C–9E). By comparison, the intensity of microcirculatory blood flow in the intestine in the HTS (Figs. 8D and 9C– 9E), GEL (Figs. 8E and 9C– 9E), and HES groups (Figs. 8F and 9C– 9E) significantly improved relative to that of the control group and was
comparable with that of the sham group at T2. In addition, Fluid resuscitation in the HTS group, but neither HES nor GEL group, had a significant higher microcirculatory blood flow intensity in the serosal muscular layer than the NS groups at T2 also
reached statistical significance (Fig. 9D).
Macrocirculation and arterial blood gas analysis
Table 5 summarized the macrocirculatory changes following hemorrhagic shock and fluid resuscitation in the part I experiment. There were significant differences in MAP change between the control and other groups (Table 5; p< 0.001). The hemorrhagic shock induced a significant decrease in the MAP from 115 (100–118) mmHg at T0 to 56 (46–67) mmHg at T2 in the control group (p < 0.001). Fluid resuscitation restored the MAP to 80 (76–85) mmHg, 89 (70–102) mmHg, 89 (82–93) mmHg and 85 (72–
99) mmHg at T2 in a comparable extent in the NS, HTS, GEL and HES group
40
respectively (compared to the control group, each p< 005; Table 5). Significant changes in heart rate were not found between each group.
Arterial pH values were significantly decreased after hemorrhagic shock at T1 in each group. The respiratory parameters including the PaO2/FiO2 ratio and PaCO2 were comparable between each group. Hemorrhagic shock induced a comparable severity of anemia in each group that the hemoglobin level and hematocrit decrease from 13.0 (12.5–13.8) g/dL and 39 (38–41) % at T0 to 10.0 (8.6–10.2) g/dL and 30 (26–30) % at T2 in the control group. The absolute hemoglobin and hematocrit values at T2 were higher in the GEL group than those in the other groups, but the changes in values from T0 to T2 were comparable between each group (Table 6). Hemorrhagic shock induced an increase in serum lactate level in each group that the serum lactate level increased from 0.6 (0.5–0.9) mmol/L at T0 to 4.3 (3.0–5.8) mmol/L at T2 in the control group. (Table 6). The serum lactate level at T2 was comparably reduced in all resuscitation groups (Table 6).
Part II. In vivo renal reactive oxygen species formation
There was a significant increase of in vivo renal ROS formation after hemorrhagic shock comparing to the same time point in the sham group [551 (322–955) vs. 155 (155–303) CL counts/10 s in the control and the sham groups respectively; p < 0.05].
41
Fluid resuscitation result in significantly increased in vivo renal ROS formation because of reperfusion syndrome in all resuscitated groups compared with the sham group at T2 [277 (189–480), 1644 (1065–2344), 3918 (2596–4610), and 3119 (1880–
9298) CL counts/10 s, in the NS, HTS, GEL, and HES groups, respectively; p < 0.05 compared with the sham group; Fig. 10]. Reperfusion-induced renal ROS formation was significantly higher in the HTS, GEL, and HES groups than in the control group.
Particularly, the GEL and the HES groups were significantly associated with higher in vivo renal ROS formation compared with that in the HTS and other groups at T2 (Fig.
10).
42
Section 2：Predicting stroke volume and arterial pressure fluid responsiveness in liver cirrhosis patients by using dynamic preload variables
We consecutively enrolled 37 patients; six patients were excluded because of impaired LVEF or arrhythmia. During the investigation period, inotropes or vasopressors were not administered to any patient. Table 7 summarizes patient characteristics. In total, 62 fluid challenges were performed, comprising 23
responders and 39 nonresponders for SV FR and 16 responders and 46 nonresponders for arterial pressure FR (Fig. 11).
Dynamic preload variable changes before and after fluid challenges in responders and nonresponders
Before the fluid challenges, all three dynamic preload variables were significantly higher in the SV responders than in the nonresponders (all P < 0.05; Table 8). After the fluid challenges, a significant decrease was observed in all three dynamic preload variables in the SV responders and nonresponders, except that the SVV did not significantly change in the nonresponders (Table 8). PPV/SVV ratio before the fluid challenges was comparable between the SV responders and nonresponders. No significant change was observed in the PPV/SVV ratio before or after the fluid challenges in the SV responders and nonresponders (Table 8).
Before the fluid challenges, no significant difference was observed in any of the three
43
dynamic preload variables in the arterial pressure responders and nonresponders (Table 9). After the fluid challenges, a significant decrease was observed in all three dynamic preload variables in the arterial pressure responders and nonresponders. The PPV/SVV ratio before the fluid challenges was comparable between the arterial pressure responders and nonresponders. No significant change was observed in the PPV/SVV ratio before or after the fluid challenges in the arterial pressure responders and nonresponders (Table 9).
Prediction of stroke volume and arterial pressure fluid responsiveness by pulse pressure variation, stroke volume variation, plethysmographic variability index, and vascular tone variables in liver cirrhosis patients
The AUROCs (95% confidence intervals) calculated for predicting SV FR in liver cirrhosis patients were 0.794 (0.673–0.886) for PPV, 0.754 (0.628–0.854) for SVV, and 0.800 (0.679–0.891) for PVI (Fig. 12). No significant difference was observed in AUROCs of any dynamic preload variable. The cut-offs for the PPV, SVV, and PVI to predict FR were 10% (sensitivity: 78.3%, specificity: 79.5%), 12% (sensitivity:
69.6%, specificity: 71.8%), and 11% (sensitivity: 95.7%, specificity: 59.0%), respectively. By contrast, the AUROC of CVP did not significantly predict SV FR.
Each vascular tone variable investigated in this study was unable to predict arterial pressure FR in liver cirrhosis patients. A subanalysis showed that these
44
vascular tone variables were unable to predict arterial pressure FR in the SV responders (patients in preload-dependent status) neither.
Correlation among pulse pressure variation, stroke volume variation, and the plethysmographic variability index in liver cirrhosis patients
A significant correlation was observed among the three dynamic preload indices before and after the fluid challenges (Table 10). There were significant correlations between PPV and SVV before and after the fluid challenges (before: r = 0.79, after: r
= 0.90). The PVI and PPV were mildly correlated before the fluid challenges (r = 0.49) and moderately correlated after the fluid challenges (r = 0.60). The PVI and SVV were mildly correlated before the fluid challenges (r = 0.48) and moderately correlated after the fluid challenges (r = 0.62).
Static hemodynamic changes before and after fluid challenges in responders and nonresponders
Compared with the dynamic preload variables, the static hemodynamic changes were less evident. Before the fluid challenges, the heart rate, central venous pressure, SVRI, and perfusion index were comparable between the responders and nonresponders in SV FR (Table 8) and those in arterial pressure FR (Table 9).
45
Compared with the SV responders, the SV nonresponders had a higher baseline SVI (45 ± 10 vs. 38 ± 10 mL m−2, P < 0.01), higher baseline cardiac index (3.80 ± 1.13 vs.
3.12 ± 0.71 L min−1 m−2, P = 0.011), and comparable baseline mean arterial pressure (Table 2). The baseline SVI and cardiac index were comparable between the arterial pressure responders and nonresponders, with the arterial pressure nonresponders showing a higher baseline mean arterial pressure than the responders did (80 ± 18 vs.
66 ± 14 mmHg, P < 0.01; Table 9).
In SV responders, the fluid challenge induced significant increases in the mean arterial pressure (from 72 ± 16 to 77 ± 15 mmHg), SV index (from 38 ± 9 to 47 ± 10 mL m−2), cardiac index (from 3.12 ± 0.71 to 3.76 ± 0.91 L min−1 m−2), and central venous pressure (from 3 ± 3 to 5 ± 3 mmHg) and induced significant decreases in the SVRI (from 1863 ± 678 to 1621 ± 594 mmHg L−1 m−2; all P < 0.05; Table 2). In SV nonresponders, the fluid challenge induced neither significant change in the SVI, cardiac index, mean arterial pressure, nor SVRI. In arterial pressure responders, the fluid challenge induced significant increases in the mean arterial pressure (from 66 ± 14 to 81 ± 16 mmHg), SV index (from 40 ± 9 to 45 ± 10 mL m−2), cardiac index (from 3.31 ± 0.83 to 3.69 ± 0.90 L min−1 m−2), and central venous pressure (from 3 ± 3 to 5 ± 3 mmHg; all P < 0.05; Table 8). The fluid challenge also induced significant increases in SVI and cardiac index the arterial pressure nonresponders, but the mean
46
arterial pressure did not significantly change and that the SVRI significantly decreased from 1805 ± 697 to 1594 ± 696 mmHg L−1 m−2 (P < 0.05; Table 9).
The fluid challenges did not induce significant changes in the heart rate or perfusion index in responders and nonresponders of both the SV (Table 8) and arterial pressure (Table 9).
47
Section 3: Comparison of two stroke volume variation-based goal-directed fluid therapies for supratentorial brain tumor resection
Baseline characteristics
In this study, 90 patients undergoing supratentorial brain tumor resection were initially included, 10 of which were excluded, as shown in Fig. 13. There were total 80 patients ultimately enrolled in this study and completed the study protocol (n = 40 per group; Fig. 13). No enrolled patient was excluded from the analysis. Patients in the two study groups exhibited comparable baseline characteristics (Table 10).
Intraoperative profiles
Table 12 summarizes the intraoperative profiles. The two groups exhibited
comparable surgical profiles, including operation time, blood loss, and transfusion volume. The urine output (2.9 ± 1.8 vs. 2.3 ± 1.6 mL/kg h; p = 0.1315) and the volume of crystalloid infusion (3.8 ± 2.5 vs. 3.1 ± 1.6 mL/kg h; p = 0.1157) were comparable between the two groups. Baseline SVV (14 ± 6 vs. 14 ± 4%; p = 0.8159).
Patients in the low SVV group received a higher volume of colloid infusion than those in the high SVV group (917 ± 385 vs. 591 ± 457 mL; p = 0.0007). Therefore, the low SVV group had a lower average SVV than the high SVV group (9% ± 3% vs. 14% ± 3%; p < 0.0001; Table 12).
The baseline hydration status, including serum osmolality, and BNP levels were
48
comparable between the two groups. However, at the end of the surgery, patients in the high SVV group had a more restrictive hydration status with a significant decrease in BNP between T1 and T2. By comparison, the serum BNP in the low SVV patients did not significantly change between T1 and T2 (Table 12).
The intraoperative hemodynamic profiles, including mean arterial pressure and heart rate, were comparable between the two groups. Furthermore, the number of patients who received intravenous ephedrine bolus was comparable between the two groups (p = 0.1397). No patient received dopamine infusion. The intraoperative average cardiac index of the low SVV group was significantly higher than that of the high SVV group (3.3 ± 0.6 vs. 2.9 ± 0.5 L/min m2; p = 0.0039; Table 12). The serum lactate level significantly increased between T1 and T2 in the high SVV group, but not in the low SVV group (Table 12).
Postoperative outcomes
Table 13 summarizes the postoperative outcomes. No 30-day mortality was observed between the two groups. One patient in the low SVV group and two in the high SVV group expired within 1 year (p = 1.000). Patients in the low SVV group had a
significantly shorter ICU stay than those in the high SVV group (1.5 ± 0.8 vs. 2.6 ± 3.3 days; p = 0.0428); however, the length of hospital stay was comparable between the two groups (11.4 ± 4.7 vs. 12.7 ± 6.3 days; p = 0.2803).
49
Table 13 lists the postoperative neurological and nonneurological outcomes. The number of patients who developed new postoperative neurological events was
significantly higher in the high SVV group than in the low SVV group (4 patients in the low SVV group: 2 with focal seizure, 1 with decreased muscle power and cerebrospinal fluid leak, and 1 with diabetes insipidus; 12 patients in the high SVV group: 3 with focal seizure, 4 with decreased muscle power, 1 with facial palsy, 1 with time disorientation, 1 with diabetes insipidus, and 2 with cerebrospinal fluid leak;
10.0% vs. 30.0%; p = 0.0482). The incidence of nonneurological complications, mainly infections, was comparable between the two groups (2 patients in each group;
p = 1.0000). In the low SVV group, 1 patient developed aspiration pneumonia and 1 patient presented wound infection. In the high SVV group, 1 patient developed bacteremia and 1 patient developed a urinary tract infection.
The number of patients who underwent emergent reoperation or received
prolonged mechanical ventilation was also comparable between the two groups (both p = 1.0000). No significant difference was observed in the number of patients with a Glasgow coma scale score < 15 at both admission and discharge states between the two groups. The low SVV group exhibited a more favorable Barthel’s index score at the discharge state than did the high SVV group (admission state: 93 ± 14 vs. 93 ± 13;
discharge state: 92 ± 13 vs. 83 ± 18; group effect p = 0.074; time effect p = 0.002;
50
group–time interaction p = 0.026; Table 13).
Perioperative changes in serum biomarkers
Figure 3 illustrates the perioperative changes in the serum levels of neuronal biomarkers. A significant difference was observed in the serum NSE level between the two groups (group effect p = 0.016; time effect p = 0.876; group–time interaction p = 0.482; Fig 14A). The post hoc test showed that the NSE levels at T3 (POD1) and T4 (POD2) were significantly higher in the high SVV group (both p < 0.05) than in the low SVV group. In addition, significant perioperative changes were observed in the serum GAFP level in the high SVV group. A significant difference was observed in the serum GAFP level between the two groups (group effect p = 0.030; time effect p = 0.005; group–time interaction p = 0.068; Fig. 14B). The post hoc test showed that the GAFP level at T4 (POD2) was also significantly higher in the high SVV group (p
< 0.05; Fig. 14B) than in the low SVV group. By contrast, no significant difference was observed in the serum S100β levels between the two groups (group effect p = 0.471; time effect p = 0.636; group–time interaction p = 0.929; Fig. 14C). No patient developed postoperative acute kidney injury and no significant changes were
observed in BUN and creatinine levels at T1, T3, and T4 in both groups.