- Email: [email protected]
Comparison of Initial Distribution Volume of Glucose and Intrathoracic Blood Volume During Hemodynamically Unstable States Early After Esophagectomy
Comparison of Initial Distribution Volume of Glucose and Intrathoracic Blood Volume During Hemodynamically Unstable States Early After Esophagectomy* Hironori Ishihara, MD; Hitomi Nakamura, MD; Hirobumi Okawa, MD; Yuichi Yatsu, MD; Toshihito Tsubo, MD; and Kazuyoshi Hirota, MD
Study objective: We have reported that initial distribution volume of glucose (IDVG) measures the central extracellular fluid volume in the presence of fluid gain or loss without apparent modification of glucose metabolism. We hypothesized that IDVG has a close relationship with intrathoracic blood volume (ITBV). We examined whether IDVG can correlate with ITBV during hemodynamically unstable states early after esophagectomy. Design: Prospective clinical study. Setting: General ICU. Patients or participants: Twelve consecutive hypotensive patients who required volume loading during the first 10 postoperative hours after admission to the ICU. Interventions: Indexed ITBV (ITBVI) and cardiac index (CI) were measured by single transpulmonary thermodilution technique using 10 mL of cold saline solution. Indexed IDVG (IDVGI) was then determined by the administration of 5 g of glucose and calculated by applying a one-compartment model. Three sets of measurements were performed: immediately after admission to the ICU, during hypotension, and after subsequent volume loading. Measurements and results: When hypotension developed, stroke volume index (SVI), central venous pressure, and ITBVI were decreased but IDVGI and CI were not. All these variables were increased after volume loading. IDVGI was correlated only slightly with either ITBVI (r2 ⴝ 0.23) or SVI (r2 ⴝ 0.38) but moderately with CI (r2 ⴝ 0.61). Conclusions: Results does not support that IDVGI can be equivalently used as an alternative measure of ITBVI or SVI, but IDVG may be clinically relevant as a measure of the fluid volume affecting CI even during hemodynamically unstable states after esophagectomy. (CHEST 2005; 128:1713–1719) Key words: cardiac output; glucose; hypovolemia; intrathoracic blood volume; measurement techniques; volume loading Abbreviations: AIC ⫽ Akaike information criterion; CI ⫽ cardiac index; CO ⫽ cardiac output; CVP ⫽ central venous pressure; ECF ⫽ extracellular fluid; IDVG ⫽ initial distribution volume of glucose; IDVGI ⫽ indexed initial distribution volume of glucose; ITBV ⫽ intrathoracic blood volume; ITBVI ⫽ indexed intrathoracic blood volume; SVI ⫽ stroke volume index
measurement of intrathoracic blood volA lthough ume (ITBV) has become possible as a more
sensitive measure of cardiac preload than cardiac filling pressures,1,2 its measurement requires a computerized instrument and a specially designed femoral arterial catheter. Consequently, these procedures limit the routine clinical application of this method even in critically ill patients. *From the Department of Anesthesiology, University of Hirosaki School of Medicine, Hirosaki-Shi, Japan. Manuscript received February 13, 2005; revision accepted March 28, 2005. www.chestjournal.org
We have proposed initial distribution volume of glucose (IDVG) as an indicator of dilution volumetry for monitoring the central extracellular fluid (ECF) volume status,3,4 even though glucose distributes not only throughout the ECF compartment but also into the intracellular compartment Reproduction of this article is prohibited without written permission from the American College of Chest Physicians (www.chestjournal. org/misc/reprints.shtml). Corresponding to: Hironori Ishihara, MD, Department of Anesthesiology, University of Hirosaki School of Medicine, 5 Zaifu-Cho, Hirosaki-Shi, 036-8562, Japan; e-mail: [email protected] CHEST / 128 / 3 / SEPTEMBER, 2005
1713
of RBC, brain, and renal medulla.5 IDVG can be approximated simply and rapidly in any ICU using a conventional blood glucose analyzer attached to blood gas analyzer,6 and its repeated measurements can be performed at 30-min intervals.7,8 The central ECF volume consists of the interstitial fluid volume of highly perfused organs such as brain, heart, liver, and kidneys as well as plasma. Its alterations are frequently observed in critically ill patients, associated with various underlying pathologies including hypovolemia or fluid overload, whether or not apparent peripheral signs.9 –11 We have reported that IDVG rather than plasma volume or blood volume is closely correlated with cardiac output (CO) early after esophagectomy independently of an infusion of insulin or plasma glucose levels present before glucose injection.10 Furthermore, we have found that intrathoracic fluid volume correlated with IDVG in patients with fluid accumulation.11 These findings would allow speculation that IDVG has a close correlation with central blood volume or ITBV. Recently Gabbanelli et al12 reported a good relationship between daily IDVG and ITBV in ICU patients. We have also observed a good correlation between IDVG and ITBV in experimental hypovolemia and subsequent volume loading in dogs.13 In these two studies,12,13 however, a considerable time was allowed before each fluid volume measurement after a relatively stable hemodynamic state had been achieved following therapeutic or experimental interventions. Thus, it remains unclear whether IDVG can consistently follow acute hemodynamic changes. According to our experience, hypotension developed in approximately 60% of patients who underwent esophagectomy, requiring subsequent volume loading throughout the first 15-h postoperative period, even though cardiovascular states immediately after surgery are stable and/or postoperative bloody drainage were minimal.14 Although we reported IDVG and CO immediately after admission to the ICU, we did not measure IDVG during hypotension.14 To examine the hypothesis that IDVG has a close correlation with ITBV, and that IDVG can be used as being an alternative clinically relevant measure of fluid volume assessment even during hemodynamically unstable states, these two volumes as well as cardiovascular variables were determined immediately after admission to the ICU, immediately after hypotension developed, and soon after subsequent volume loading in the early postoperative period after esophagectomy. 1714
Materials and Methods The study was approved by our institutional review board, and each patient gave written informed consent before surgery. Although patients with aortic aneurysm were excluded from the study, patients who had a history of diabetes mellitus and/or cardiovascular diseases such as hypertension without apparent ischemic heart disease were included into the study (Table 1). Preoperative echocardiographic examination revealed that left ventricular ejection fraction was ⬎ 60%. Seventeen consecutive surgical patients admitted postoperatively to the general ICU from March 2003 to August 2004 were initially enrolled into the study. Twelve of these patients with hypotension during the first 10 postoperative hours were finally enrolled into the study. Each patient underwent radical surgery for esophageal cancer performed through a right thoracoabdominal approach along with extensive resection of adjacent lymph nodes, subcarinal lymph nodes, and/or cervical lymph nodes, and stayed in the ICU for at least the first 2 postoperative days. Total IV anesthesia with propofol, ketamine, and fentanyl was administered to all patients. Fluid and cardiovascular management decisions during anesthesia, including amounts of crystalloid solution and use of colloidal solutions, blood products, or vasoactive drugs, were made by individual anesthesiologists. Five patients received both packed RBCs and plasma protein fraction in the operating room. Three patients only received plasma protein fraction in addition to crystalloid solution in the operating room. No patients received a continuous infusion of vasoactive drugs during anesthesia. Neither vasoactive drugs nor blood products were administered when patients arrived at the ICU. All patients postoperatively received mechanical ventilatory support of either synchronized intermittent mandatory ventilation or pressure support ventilation with a low positive end-expiratory pressure (⬍ 5 cm H2O) under constant infusions with midazolam and morphine at least until the first postoperative morning. Both 4.3% glucose solution with electrolytes and lactated Ringer solution were infused simultaneously at a constant rate of 1.5 mL/kg/h for the former and 1.0 mL/kg/h for the latter throughout at least 12 h after surgery. No vasoactive drugs were administered throughout the study period. One patient required a continuous infusion of insulin (1 U/h) throughout the study period.
Table 1—Patient Demographics and Fluid Management During Anesthesia and Surgery* Variables
Data
Patients, No. Male/female gender, No. Age, yr Preoperative body weight, kg Preoperative body surface area, m2 Duration of surgery, h Lactated Ringer solution, L Packed RBCs, mL Colloids, mL Urine output, mL Estimated blood loss during surgery, g Blood lactate, mmol/L† Blood temperature, °C†
12 11/1 63 ⫾ 7 (51–73) 51.2 ⫾ 7.1 (41–65) 1.53 ⫾ 0.10 (1.37–1.71) 9.9 ⫾ 1.6 (7.6–12.2) 6.1 ⫾ 1.3 (4.8–9.6) 5 (260–910) 8 (250–1,300) 640 ⫾ 360 (290–1,560) 1,500 ⫾ 870 (580–2,900) 1.7 ⫾ 0.8 (0.8–3.0) 35.5 ⫾ 0.7 (34.5–36.7)
*Data are presented as mean ⫾ SD (range) unless otherwise indicated. †Immediately after admission to the ICU. Clinical Investigations in Critical Care
Clinically Defined Hypotension During the first 10 h postoperatively, volume loading was clinically required when following either condition was met: (1) systolic arterial pressure ⬍ 90 mm Hg lasting ⬎ 5 min without an increase in central venous pressure (CVP), and (2) a reduction of arterial BP ⬎ 40 mm Hg compared with the value immediately after admission to the ICU lasting ⬎ 5 min without an increase in CVP. Measurement of Fluid Volumes Postoperative measurements were made three times: immediately after admission to the ICU on the operative day, during clinically defined hypotension, and 10 min after completion of volume loading with 250 mL of 10% low-molecular-weight dextran over 20 to 30 min. A right subclavian venous catheter had been placed before the operative day. A thermistor-tipped catheter for thermodilution and pulse contour analysis (PV2015L13; Pulsion Medical Systems; Munich, Germany) was inserted into a femoral artery and connected to a monitoring system (PiCCO; Pulsion Medical Systems) in the ICU immediately after surgery, as has been described elsewhere.15 Ten milliliters of cold isotonic saline solution (⬍ 8°C) was injected through the right subclavian venous line to determine ITBV. Measurement was performed by one of the authors (H.I.), triplicated, and averaged. Immediately after measurement of ITBV, CO, and routine cardiovascular variables, 10 mL of 50% glucose solution (5 g) was injected through the same central venous line to calculate IDVG. Blood samples were obtained through a radial artery catheter at the following times: immediately before, and 3, 4, 5, and 7 min after injection. Isotonic saline solution, 2 mL, with a small amount of heparin was used to flush the arterial line immediately following indicator injection, and each 1-mL blood sample was collected into a heparinized syringe. Plasma was separated immediately, and measurements of glucose concentrations were performed within 10 min of sampling. Routine hemodynamic and clinical variables were recorded immediately before each volume measurement. IDVG was calculated from plasma decay curves using a one-compartment model from the increased plasma values from 3 to 7 min after infusion, as described in our previous reports.8 –11,14 Plasma glucose concentrations of all blood samples were measured using amperometry by glucose oxidase immobilized membrane H2O2 electrode (glucose analyzer GA-1150; Arkray Co, Ltd; Kyoto, Japan). Measurements were made in duplicate. Coefficients of variation for repeated measurements were ⱕ 1.0% for plasma glucose (range, 3.0 to 17.0 mmol/L). Akaike information criterion (AIC)16 for IDVG curve was
examined as described previously8 –11,14 to evaluate the exponential term of the pharmacokinetic model. Low AIC values indicate a superiority of the selected model. The mean (⫾ SD) AIC value was ⫺ 24.4 ⫾ 5.8 in this study. As indicated by the AIC values, convergence was assumed for each curve in this study as observed in previous reports.8 –11,14 Statistical Analysis Unless otherwise stated, data are presented as mean (SD) values. IDVG, ITBV, stroke volume, and CO are indexed to body surface area based on the reported preoperative height and body weight. Normally distributed continuous variables were compared by repeated-measures analysis of variance. If significant differences were noted in the analysis of variance test, intergroup comparisons were made using Bonferroni t test. The corresponding nonparametric tests (Student-Newman-Keuls test) were used for data not normally distributed. Regression analysis using either actual or changed values was performed as needed. Actual values were defined as current values at each measurement point. Changed values were defined as current values minus previous values. Thus, the number of data for comparison was 36 for actual values and 24 for changed values.
Results A minimal postoperative bloody drainage on the operative day and a relatively stable cardiovascular state on the first postoperative day without requiring blood transfusion did not support the presence of apparent continued hemorrhage in this study. Hypotension associated with an increase in heart rate occurred at an average of 3.4 ⫾ 2.5 h (range, 1 to 7.6 h) after the first measurement. Six patients whose hypotension occurred during the first 4 h had estimated intraoperative blood loss ⬎ 1,000 g (Fig 1, left), and five of them received both packed RBCs colloidal solutions in the operative room, whereas three patients whose hypotension occurred thereafter had intraoperative blood loss ⬍ 1,000 g, and one of them received colloidal solutions in the operating room. In all patients, peripheral skin was cold on admission to the ICU, and thereafter became warm associated with an apparent increase in blood tem-
Figure 1. Relationship between estimated intraoperative blood loss and time difference between hypotension and the first measurement (left), and between an increase in blood temperature when hypotension developed compared with admission to the ICU and time difference between hypotension and the first measurement (Time) [right]. www.chestjournal.org
CHEST / 128 / 3 / SEPTEMBER, 2005
1715
Figure 2. Changes in CI when hypotension developed compared with the first measurement immediately after admission to the ICU in each individual patient. Time ⫽ time difference between hypotension and the first measurement.
perature particularly in the late study period (Fig 1, right). Hypotension occurred during the first 4 h was associated with a decrease in cardiac index (CI) in seven of eight patients, and hypotension occurred thereafter was associated with an increase in CI in three of four patients as compared with the first measurement (Fig 2). Table 2 shows cardiovascular variables and fluid volumes in three sets of measurement. When hypotension developed, CVP, indexed ITBV (ITBVI), and stroke volume index (SVI) were decreased (p ⬍ 0.01, respectively), but both indexed IDVG (IDVGI) and CI remained statistically unchanged. These five values were increased after volume loading compared with hypotension (p ⬍ 0.001, respectively), even though these values varied obviously among each individual patient as shown in Figure 3 for systolic arterial pressure. Actual IDVGI was correlated only slightly with
actual ITBVI (r2 ⫽ 0.23, n ⫽ 36, p ⫽ 0.003), even though a moderate correlation was observed between these two changed volumes (r2 ⫽ 0.69, n ⫽ 24, p ⬍ 0.0001) [Fig 4]. Actual IDVGI was also correlated only slightly with actual SVI (r2 ⫽ 0.39, n ⫽ 36, p ⬍ 0.0001), even though a moderate linear correlation was observed between these two changed values (r2 ⫽ 0.72, n ⫽ 24, p ⬍ 0.0001) [Fig 5]. Either actual or changed IDVGI was correlated moderately with that of CI (r2 ⫽ 0.61, n ⫽ 36, p ⬍ 0.000, and r2 ⫽ 0.82, n ⫽ 24, p ⬍ 0.0001, respectively). Either actual or changed ITBVI was correlated moderately with that of SVI (r2 ⫽ 0.56, n ⫽ 36, p ⬍ 0.0001, and r2 ⫽ 0.82, n ⫽ 24, p ⬍ 0.0001, respectively). Actual ITBVI was correlated only slightly with actual CI (r2 ⫽ 0.32, n ⫽ 36, p ⬍ 0.001), even though a moderate correlation was observed between these two changed values (r2 ⫽ 0.69, n ⫽ 24, p ⬍ 0.0001). Discussion As judged by clinical conditions including amounts of intraoperative blood loss and peripheral circulatory state when hypotension developed, hypotension during the first 4 h of the study period may be mainly attributable to hypovolemia, even though neither arterial BP nor blood lactate levels immediately after admission to the ICU indicated the presence of apparent oxygen deficit in the peripheral tissues at that time, whereas hypotension thereafter may be mainly attributable to the redistribution of blood from the central to the peripheral tissues as judged by improved peripheral circulation. A simultaneous decrease in CVP, ITBVI, SVI, and CI in this study
Table 2—Cardiovascular Variables and Fluid Volumes* Variables
ICU Admission
Hypotension
Volume Loading
Heart rate, beats/min Systemic arterial pressure, mm Hg Mean arterial pressure, mm Hg CI, L/min/m2 SVI, mL/m2 CVP, mm Hg Hematocrit, % Plasma glucose, mmol/L† IDVGI, L/m2 ITBVI, L/m2
78 ⫾ 17 (50–120) 122 ⫾ 28 (87–174) 90 ⫾ 19 (64–121) 2.6 ⫾ 0.4 (2.0–3.4) 35 ⫾ 9 (17–48) 8 ⫾ 2 (4–12) 30.9 ⫾ 4.6 (24.3–39.1) 8.3 ⫾ 1.4 (5.6–10.1) 3.6 ⫾ 0.4 (3.1–4.2) 0.90 ⫾ 0.1 (0.67–1.12)
92 ⫾ 17 (75–125)‡ 88 ⫾ 9 (76–109)储 63 ⫾ 8 (50–76)储 2.5 ⫾ 0.5 (1.8–3.7) 29 ⫾ 8 (21–44)§ 7 ⫾ 3 (2–12)§ 31.2 ⫾ 5.0 (24.9–41.8) 8.1 ⫾ 1.8 (5.3–11.0) 3.4 ⫾ 0.5 (2.7–4.3) 0.81 ⫾ 0.15 (0.61-1.14)§
85 ⫾ 14 (58–106)¶ 105 ⫾ 16 (82–139)‡# 72 ⫾ 10 (53–91)储 3.5 ⫾ 0.5 (2.9–4.6)储** 42 ⫾ 9 (31–60)§** 9 ⫾ 3 (5–15)§** 25.9 ⫾ 4.3 (20.4–34.9)储** 8.7 ⫾ 1.9 (5.8–11.7) 4.1 ⫾ 0.5 (3.4–4.8)储** 0.95 ⫾ 0.14 (0.75–1.17)**
*Data are presented as mean ⫾ SD (range). †Immediately before glucose injection. ‡p ⬍ 0.05 compared with immediately after ICU admission. §p ⬍ 0.01 compared with immediately after ICU admission. 储p ⬍ 0.001 compared with immediately after ICU admission. ¶p ⬍ 0.05 compared with during hypotension. #p ⬍ 0.01 compared with during hypotension. **p ⬍ 0.001 compared with during hypotension. 1716
Clinical Investigations in Critical Care
Figure 3. Systolic arterial pressure immediately after admission to the ICU, during hypotension, and after fluid volume loading in each individual patient.
would support a decrease in cardiac preload or hypovolemic hypotension, whereas a decrease in CVP, ITBVI, and SVI, but not CI, would support a simultaneous decrease in cardiac preload and afterload. Changes in heart rate and systemic inflammatory response syndrome after esophagectomy17 would also play a role in determining CI at least partly. Presumably, there may be several contributory factors to affect CI in each individual patient of this study. Thus, it would be impossible to draw any definitive conclusion to the different responses of CI during the hypotensive episode in this study. Considering that IDVGI was correlated only slightly with either SVI or ITBVI in this study, IDVGI is not being a relevant surrogate measure of cardiac preload during unstable hemodynamic states. This is contrast to our experimental study13 in which IDVG following hemorrhage and fluid loading had a moderate correlation with ITBV (r2 ⫽ 0.76). Gabbanelli et al12 also demonstrated a moderate
linear correlation between IDVGI and ITBVI in critically ill patients (r2 ⫽ 0.79). In these two studies12,13 both fluid volumes were evaluated during a relatively steady hemodynamic state following experimental or therapeutic interventions. Presumably, the different measurement timing as well as underlying pathophysiology would yield different relationship between the two volumes in this study, even though the relationship between the changed volumes in this study were comparable with these two previous studies. In contrast to ITBVI, both CI and IDVGI were not consistently decreased when hypotension developed. However, the relationship between IDVGI and CI in this study (r2 ⫽ 0.67) was comparable with that of our previous daily based studies after esophagectomy (r2 ⫽ 0.50)10 and early after major burn (r2 ⫽ 0.69).18 We have also observed a linear correlation between IDVG and CO after an infusion of phentolamine that yields fluid redistribution from the central to the peripheral compartment.19 Furthermore, 22 plots of 24 changes in IDVGI and CI moved together to the same direction in this study, as shown in Figure 5. These findings would allow speculation that the IDVGI/CI relationship is consistent even during hypotension. Based on capillary membrane permeability of glucose, the rate at which glucose molecules diffuse through the capillary membrane is approximately 50 times greater than the rate at which plasma itself flows linearly along the capillary,20 suggesting that glucose can rapidly distribute throughout the central ECF compartment even in a low CO state. This is supported by the fact that a relatively large IDVG was observed even in the presence of a small CO in patients with congestive heart failure.4 Accordingly, IDVG may be variable depending on the actual size of the central ECF
Figure 4. Relationship between IDVGI and ITBVI. Shown are actual values (left, A): Y ⫽ 0.14X ⫹ 0.38 (r2 ⫽ 0.23, n ⫽ 36, p ⫽ 0.003); and changed values (right, B): Y ⫽ 0.2X – 0.035 (r2 ⫽ 0.69, n ⫽ 24, p ⬍ 0.0001). www.chestjournal.org
CHEST / 128 / 3 / SEPTEMBER, 2005
1717
Figure 5. Relationship between IDVGI, SVI, and CI. Shown are actual values between IDVGI and SVI (top left, A): Y ⫽ 12X – 8.3 (r2 ⫽ 0.39, n ⫽ 36, p ⬍ 0.0001); changed values between IDVG and SVI (top right, B): Y ⫽ 17X ⫺ 1.3 (r2 ⫽ 0.72, n ⫽ 24, p ⬍ 0.0001); actual values between IDVGI and CI (bottom left, C): Y ⫽ 0.94X – 0.59 (r2 ⫽ 0.61, n ⫽ 36, p ⬍ 0.0001); and changed values between IDVGI and CI (bottom right, D): Y ⫽ X ⫹ 0.14 (r2 ⫽ 0.82, n ⫽ 24, p ⬍ 0.0001).
volume independently of CO states. As IDVG is apparently greater than ITBV, IDVG would reflect not only cardiac preload or ITBV but also rapidly exchanging extravascular fluid volume or the central ECF volume. Nevertheless, a good correlation has been consistently observed between IDVG and CO in critically ill patients without congestive heart failure.4,10,12,14,18 Considering the result of the present study, IDVG measurement cannot be used equally as a measure of cardiac preload. However, it may be of benefit in decision making even immediately after hypotension developed whether volume loading or administration of vasopressors is required for its treatment, since IDVGI has a closer correlation with CI as compared with either ITBVI or SVI in this study. Although we administered fluid volume loading to all patients when hypotension developed in this study, it would have been more appropriate to administer vasoactive drugs such as norepinephrine, instead of fluid volume loading, in some patients associated with an increase in IDVGI or CI. In contrast to ITBV measurements, a considerable 1718
time interval should be allowed between repeated IDVG measurements. We found that the bias of IDVG measurement at a 30-min interval was 0.08 ⫾ 0.32 L in 25 hemodynamically stable patients,8 indicating that IDVG determination can be reliably repeated within a minimum 30 min. Moreover, we reported that IDVG can be approximated simply and rapidly with only two plasma samples: immediately before glucose challenge and 3 min after injection.21 By using the formula we reported,21 IDVG can be approximated even using a conventional glucose analyzer.6 According to our experience of ⬎ 3,700 IDVG determinations in the ICU, it can be performed during the routine fluid management even without taking the greatest care to stabilize plasma glucose levels. We observed a continuous decline in plasma glucose concentration over 60 min after injection, although plasma glucose concentrations at 60 min after injection remained slightly elevated as compared with the preinjection value.8 Hence, IDVG measurement will not induce a continued hyperglycemic state, even in critically ill Clinical Investigations in Critical Care
patients. Diaz-Parejo et al22 suggested that transient hyperglycemia had no adverse effect on outcome in patients with severe traumatic brain lesions and stroke. Therefore, we should be more concerned about normalization in basal plasma glucose concentration than about transient hyperglycemia in these patients. Conclusion Results demonstrate that IDVGI is correlated only slightly with ITBVI or SVI but moderately with CI during hemodynamically unstable states after esophagectomy. Results suggest that IDVGI should not equally be used as an alternative measure of ITBVI but may be useful as a measure of the central fluid volume that may affect CI even during hemodynamically unstable states. ACKNOWLEDGMENT: The authors thank Emeritus Prof. Matsuki A (Hakodate, Japan), Prof. A. H. Giesecke Jr (Dallas, TX), and Dr. P. Hollister (Misawa, Japan) for valuable comments and continued support of the study.
8
9
10
11
12
13
14
References 1 Lichtwarck-Aschoff M, Zeravik J, Pfeiffer UJ. Intrathoracic blood volume accurately reflects circulatory volume status in critically ill patients with mechanical ventilation. Intensive Care Med 1992; 18:142–147 2 Boussat S, Jacques B, Levy B, et al. Intravascular volume monitoring and extravascular lung water in septic patients with pulmonary edema. Intensive Care Med 2002; 28:712– 718 3 Iwakawa T, Ishihara H, Takamura K, et al. Measurements of extracellular fluid volume in highly perfused organs and lung water in hypo- and hypervolaemic dogs. Eur J Anaesthesiol 1998; 15:414 – 421 4 Ishihara H, Takamura K, Koh H, et al. Does the initial distribution volume of glucose reflect the central extracellular fluid volume status in critically ill patients? Infusionsther Transfusionsmed 1996; 23:196 –201 5 Ferrannini E, Smith JD, Cobelli C. Effect of insulin on the distribution and disposition of glucose in man. J Clin Invest 1985; 76:357–364 6 Ishihara H, Nakamura H, Okawa H, et al. Initial distribution volume of glucose can be approximated using a conventional glucose analyzer in the intensive care unit. Crit Care 2005; 9:R144 –R149 7 Mi W, Ishihara H, Sakai T, et al. Possible overestimation of
www.chestjournal.org
15 16 17 18 19 20 21
22
indocyanine green-derived plasma volume early after induction of anesthesia with propofol/fentanyl. Anesth Analg 2003; 97:1421–1427 Rose BO, Ishihara H, Okawa H, et al. Repeatability of measurements of the initial distribution volume of glucose in haemodynamically stable patients. J Clin Pharm Ther 2004; 29:317–323 Ishihara H, Matsuno S, Taguchi S, et al. Evaluation of fluid volume status with a glucose challenge test in a patient with acute adrenal insufficiency. J Anesth 1996; 10:69 –71 Ishihara H, Suzuki A, Okawa H, et al. The initial distribution volume of glucose rather than indocyanine green derived plasma volume is correlated with cardiac output following major surgery. Intensive Care Med 2000; 26:1441–1448 Ishihara H, Suzuki A, Okawa H, et al. Comparison of initial distribution volume of glucose and plasma volume in thoracic fluid-accumulated patients. Crit Care Med 2001; 29:1532– 1538 Gabbanelli V, Pntanetti S, Donati A, et al. Initial distribution volume of glucose as noninvasive indicator of cardiac preload: comparison with intrathoracic blood volume. Intensive Care Med 2004; 30:2067–2073 Nakamura H, Ishihara H, Okawa H, et al. Initial distribution volume of glucose is correlated with intrathoracic blood volume in hypovolemia and following volume loading in dogs. Eur J Anaesth 2005; 22:202–208 Suzuki A, Ishihara H, Okawa H, et al. Can initial distribution volume of glucose predict hypovolemic hypotension after radical surgery for esophageal cancer? Anesth Analg 2001; 92:1146 –1151 Sakka SG, Ru¨hl CC, Pfeiffer UJ, et al. Assessment of cardiac preload and extravascular lung water by single transpulmonary thermodilution. Intensive Care Med 2000; 26:180 –187 Akaike H. A new look at the statistical model identification. IEEE Trans Automatic Control 1974; AC-19:716 –723 Nakanishi K, Takeda S, Terajima K, et al. Myocardial dysfunction associated with proinflammatory cytokines after esophageal resection. Anesth Analg 2000; 91:270 –275 Ishihara H, Otomo N, Suzuki A, et al. Detection of capillary protein leakage by glucose and indocyanine green dilutions during the early post-burn period. Burns 1998; 24:525–531 Matsui A, Ishihara H, Suzuki A, et al. Glucose dilution can detect fluid redistribution following phentolamine infusion in dogs. Intensive Care Med 2000; 26:1131–1138 Guyton AC. The microcirculation and the lymphatic system. In: Guyton AC. Textbook of medical physiology. 10th ed. Philadelphia, PA: WB Saunders, 2000; 162–183 Hirota K, Ishihara H, Tsubo T, et al. Estimation of the initial distribution volume of glucose by an incremental plasma glucose level at 3 min after i.v. glucose in humans. Br J Clin Pharmacol 1999; 47:361–364 Diaz-Parejo P, Stahl N, Xu W, et al. Cerebral energy metabolism during transient hyperglycaemia in patients with severe brain trauma. Intensive Care Med 2003; 29:544 –550
CHEST / 128 / 3 / SEPTEMBER, 2005
1719
Study objective: We have reported that initial distribution volume of glucose (IDVG) measures the central extracellular fluid volume in the presence of fluid gain or loss without apparent modification of glucose metabolism. We hypothesized that IDVG has a close relationship with intrathoracic blood volume (ITBV). We examined whether IDVG can correlate with ITBV during hemodynamically unstable states early after esophagectomy. Design: Prospective clinical study. Setting: General ICU. Patients or participants: Twelve consecutive hypotensive patients who required volume loading during the first 10 postoperative hours after admission to the ICU. Interventions: Indexed ITBV (ITBVI) and cardiac index (CI) were measured by single transpulmonary thermodilution technique using 10 mL of cold saline solution. Indexed IDVG (IDVGI) was then determined by the administration of 5 g of glucose and calculated by applying a one-compartment model. Three sets of measurements were performed: immediately after admission to the ICU, during hypotension, and after subsequent volume loading. Measurements and results: When hypotension developed, stroke volume index (SVI), central venous pressure, and ITBVI were decreased but IDVGI and CI were not. All these variables were increased after volume loading. IDVGI was correlated only slightly with either ITBVI (r2 ⴝ 0.23) or SVI (r2 ⴝ 0.38) but moderately with CI (r2 ⴝ 0.61). Conclusions: Results does not support that IDVGI can be equivalently used as an alternative measure of ITBVI or SVI, but IDVG may be clinically relevant as a measure of the fluid volume affecting CI even during hemodynamically unstable states after esophagectomy. (CHEST 2005; 128:1713–1719) Key words: cardiac output; glucose; hypovolemia; intrathoracic blood volume; measurement techniques; volume loading Abbreviations: AIC ⫽ Akaike information criterion; CI ⫽ cardiac index; CO ⫽ cardiac output; CVP ⫽ central venous pressure; ECF ⫽ extracellular fluid; IDVG ⫽ initial distribution volume of glucose; IDVGI ⫽ indexed initial distribution volume of glucose; ITBV ⫽ intrathoracic blood volume; ITBVI ⫽ indexed intrathoracic blood volume; SVI ⫽ stroke volume index
measurement of intrathoracic blood volA lthough ume (ITBV) has become possible as a more
sensitive measure of cardiac preload than cardiac filling pressures,1,2 its measurement requires a computerized instrument and a specially designed femoral arterial catheter. Consequently, these procedures limit the routine clinical application of this method even in critically ill patients. *From the Department of Anesthesiology, University of Hirosaki School of Medicine, Hirosaki-Shi, Japan. Manuscript received February 13, 2005; revision accepted March 28, 2005. www.chestjournal.org
We have proposed initial distribution volume of glucose (IDVG) as an indicator of dilution volumetry for monitoring the central extracellular fluid (ECF) volume status,3,4 even though glucose distributes not only throughout the ECF compartment but also into the intracellular compartment Reproduction of this article is prohibited without written permission from the American College of Chest Physicians (www.chestjournal. org/misc/reprints.shtml). Corresponding to: Hironori Ishihara, MD, Department of Anesthesiology, University of Hirosaki School of Medicine, 5 Zaifu-Cho, Hirosaki-Shi, 036-8562, Japan; e-mail: [email protected] CHEST / 128 / 3 / SEPTEMBER, 2005
1713
of RBC, brain, and renal medulla.5 IDVG can be approximated simply and rapidly in any ICU using a conventional blood glucose analyzer attached to blood gas analyzer,6 and its repeated measurements can be performed at 30-min intervals.7,8 The central ECF volume consists of the interstitial fluid volume of highly perfused organs such as brain, heart, liver, and kidneys as well as plasma. Its alterations are frequently observed in critically ill patients, associated with various underlying pathologies including hypovolemia or fluid overload, whether or not apparent peripheral signs.9 –11 We have reported that IDVG rather than plasma volume or blood volume is closely correlated with cardiac output (CO) early after esophagectomy independently of an infusion of insulin or plasma glucose levels present before glucose injection.10 Furthermore, we have found that intrathoracic fluid volume correlated with IDVG in patients with fluid accumulation.11 These findings would allow speculation that IDVG has a close correlation with central blood volume or ITBV. Recently Gabbanelli et al12 reported a good relationship between daily IDVG and ITBV in ICU patients. We have also observed a good correlation between IDVG and ITBV in experimental hypovolemia and subsequent volume loading in dogs.13 In these two studies,12,13 however, a considerable time was allowed before each fluid volume measurement after a relatively stable hemodynamic state had been achieved following therapeutic or experimental interventions. Thus, it remains unclear whether IDVG can consistently follow acute hemodynamic changes. According to our experience, hypotension developed in approximately 60% of patients who underwent esophagectomy, requiring subsequent volume loading throughout the first 15-h postoperative period, even though cardiovascular states immediately after surgery are stable and/or postoperative bloody drainage were minimal.14 Although we reported IDVG and CO immediately after admission to the ICU, we did not measure IDVG during hypotension.14 To examine the hypothesis that IDVG has a close correlation with ITBV, and that IDVG can be used as being an alternative clinically relevant measure of fluid volume assessment even during hemodynamically unstable states, these two volumes as well as cardiovascular variables were determined immediately after admission to the ICU, immediately after hypotension developed, and soon after subsequent volume loading in the early postoperative period after esophagectomy. 1714
Materials and Methods The study was approved by our institutional review board, and each patient gave written informed consent before surgery. Although patients with aortic aneurysm were excluded from the study, patients who had a history of diabetes mellitus and/or cardiovascular diseases such as hypertension without apparent ischemic heart disease were included into the study (Table 1). Preoperative echocardiographic examination revealed that left ventricular ejection fraction was ⬎ 60%. Seventeen consecutive surgical patients admitted postoperatively to the general ICU from March 2003 to August 2004 were initially enrolled into the study. Twelve of these patients with hypotension during the first 10 postoperative hours were finally enrolled into the study. Each patient underwent radical surgery for esophageal cancer performed through a right thoracoabdominal approach along with extensive resection of adjacent lymph nodes, subcarinal lymph nodes, and/or cervical lymph nodes, and stayed in the ICU for at least the first 2 postoperative days. Total IV anesthesia with propofol, ketamine, and fentanyl was administered to all patients. Fluid and cardiovascular management decisions during anesthesia, including amounts of crystalloid solution and use of colloidal solutions, blood products, or vasoactive drugs, were made by individual anesthesiologists. Five patients received both packed RBCs and plasma protein fraction in the operating room. Three patients only received plasma protein fraction in addition to crystalloid solution in the operating room. No patients received a continuous infusion of vasoactive drugs during anesthesia. Neither vasoactive drugs nor blood products were administered when patients arrived at the ICU. All patients postoperatively received mechanical ventilatory support of either synchronized intermittent mandatory ventilation or pressure support ventilation with a low positive end-expiratory pressure (⬍ 5 cm H2O) under constant infusions with midazolam and morphine at least until the first postoperative morning. Both 4.3% glucose solution with electrolytes and lactated Ringer solution were infused simultaneously at a constant rate of 1.5 mL/kg/h for the former and 1.0 mL/kg/h for the latter throughout at least 12 h after surgery. No vasoactive drugs were administered throughout the study period. One patient required a continuous infusion of insulin (1 U/h) throughout the study period.
Table 1—Patient Demographics and Fluid Management During Anesthesia and Surgery* Variables
Data
Patients, No. Male/female gender, No. Age, yr Preoperative body weight, kg Preoperative body surface area, m2 Duration of surgery, h Lactated Ringer solution, L Packed RBCs, mL Colloids, mL Urine output, mL Estimated blood loss during surgery, g Blood lactate, mmol/L† Blood temperature, °C†
12 11/1 63 ⫾ 7 (51–73) 51.2 ⫾ 7.1 (41–65) 1.53 ⫾ 0.10 (1.37–1.71) 9.9 ⫾ 1.6 (7.6–12.2) 6.1 ⫾ 1.3 (4.8–9.6) 5 (260–910) 8 (250–1,300) 640 ⫾ 360 (290–1,560) 1,500 ⫾ 870 (580–2,900) 1.7 ⫾ 0.8 (0.8–3.0) 35.5 ⫾ 0.7 (34.5–36.7)
*Data are presented as mean ⫾ SD (range) unless otherwise indicated. †Immediately after admission to the ICU. Clinical Investigations in Critical Care
Clinically Defined Hypotension During the first 10 h postoperatively, volume loading was clinically required when following either condition was met: (1) systolic arterial pressure ⬍ 90 mm Hg lasting ⬎ 5 min without an increase in central venous pressure (CVP), and (2) a reduction of arterial BP ⬎ 40 mm Hg compared with the value immediately after admission to the ICU lasting ⬎ 5 min without an increase in CVP. Measurement of Fluid Volumes Postoperative measurements were made three times: immediately after admission to the ICU on the operative day, during clinically defined hypotension, and 10 min after completion of volume loading with 250 mL of 10% low-molecular-weight dextran over 20 to 30 min. A right subclavian venous catheter had been placed before the operative day. A thermistor-tipped catheter for thermodilution and pulse contour analysis (PV2015L13; Pulsion Medical Systems; Munich, Germany) was inserted into a femoral artery and connected to a monitoring system (PiCCO; Pulsion Medical Systems) in the ICU immediately after surgery, as has been described elsewhere.15 Ten milliliters of cold isotonic saline solution (⬍ 8°C) was injected through the right subclavian venous line to determine ITBV. Measurement was performed by one of the authors (H.I.), triplicated, and averaged. Immediately after measurement of ITBV, CO, and routine cardiovascular variables, 10 mL of 50% glucose solution (5 g) was injected through the same central venous line to calculate IDVG. Blood samples were obtained through a radial artery catheter at the following times: immediately before, and 3, 4, 5, and 7 min after injection. Isotonic saline solution, 2 mL, with a small amount of heparin was used to flush the arterial line immediately following indicator injection, and each 1-mL blood sample was collected into a heparinized syringe. Plasma was separated immediately, and measurements of glucose concentrations were performed within 10 min of sampling. Routine hemodynamic and clinical variables were recorded immediately before each volume measurement. IDVG was calculated from plasma decay curves using a one-compartment model from the increased plasma values from 3 to 7 min after infusion, as described in our previous reports.8 –11,14 Plasma glucose concentrations of all blood samples were measured using amperometry by glucose oxidase immobilized membrane H2O2 electrode (glucose analyzer GA-1150; Arkray Co, Ltd; Kyoto, Japan). Measurements were made in duplicate. Coefficients of variation for repeated measurements were ⱕ 1.0% for plasma glucose (range, 3.0 to 17.0 mmol/L). Akaike information criterion (AIC)16 for IDVG curve was
examined as described previously8 –11,14 to evaluate the exponential term of the pharmacokinetic model. Low AIC values indicate a superiority of the selected model. The mean (⫾ SD) AIC value was ⫺ 24.4 ⫾ 5.8 in this study. As indicated by the AIC values, convergence was assumed for each curve in this study as observed in previous reports.8 –11,14 Statistical Analysis Unless otherwise stated, data are presented as mean (SD) values. IDVG, ITBV, stroke volume, and CO are indexed to body surface area based on the reported preoperative height and body weight. Normally distributed continuous variables were compared by repeated-measures analysis of variance. If significant differences were noted in the analysis of variance test, intergroup comparisons were made using Bonferroni t test. The corresponding nonparametric tests (Student-Newman-Keuls test) were used for data not normally distributed. Regression analysis using either actual or changed values was performed as needed. Actual values were defined as current values at each measurement point. Changed values were defined as current values minus previous values. Thus, the number of data for comparison was 36 for actual values and 24 for changed values.
Results A minimal postoperative bloody drainage on the operative day and a relatively stable cardiovascular state on the first postoperative day without requiring blood transfusion did not support the presence of apparent continued hemorrhage in this study. Hypotension associated with an increase in heart rate occurred at an average of 3.4 ⫾ 2.5 h (range, 1 to 7.6 h) after the first measurement. Six patients whose hypotension occurred during the first 4 h had estimated intraoperative blood loss ⬎ 1,000 g (Fig 1, left), and five of them received both packed RBCs colloidal solutions in the operative room, whereas three patients whose hypotension occurred thereafter had intraoperative blood loss ⬍ 1,000 g, and one of them received colloidal solutions in the operating room. In all patients, peripheral skin was cold on admission to the ICU, and thereafter became warm associated with an apparent increase in blood tem-
Figure 1. Relationship between estimated intraoperative blood loss and time difference between hypotension and the first measurement (left), and between an increase in blood temperature when hypotension developed compared with admission to the ICU and time difference between hypotension and the first measurement (Time) [right]. www.chestjournal.org
CHEST / 128 / 3 / SEPTEMBER, 2005
1715
Figure 2. Changes in CI when hypotension developed compared with the first measurement immediately after admission to the ICU in each individual patient. Time ⫽ time difference between hypotension and the first measurement.
perature particularly in the late study period (Fig 1, right). Hypotension occurred during the first 4 h was associated with a decrease in cardiac index (CI) in seven of eight patients, and hypotension occurred thereafter was associated with an increase in CI in three of four patients as compared with the first measurement (Fig 2). Table 2 shows cardiovascular variables and fluid volumes in three sets of measurement. When hypotension developed, CVP, indexed ITBV (ITBVI), and stroke volume index (SVI) were decreased (p ⬍ 0.01, respectively), but both indexed IDVG (IDVGI) and CI remained statistically unchanged. These five values were increased after volume loading compared with hypotension (p ⬍ 0.001, respectively), even though these values varied obviously among each individual patient as shown in Figure 3 for systolic arterial pressure. Actual IDVGI was correlated only slightly with
actual ITBVI (r2 ⫽ 0.23, n ⫽ 36, p ⫽ 0.003), even though a moderate correlation was observed between these two changed volumes (r2 ⫽ 0.69, n ⫽ 24, p ⬍ 0.0001) [Fig 4]. Actual IDVGI was also correlated only slightly with actual SVI (r2 ⫽ 0.39, n ⫽ 36, p ⬍ 0.0001), even though a moderate linear correlation was observed between these two changed values (r2 ⫽ 0.72, n ⫽ 24, p ⬍ 0.0001) [Fig 5]. Either actual or changed IDVGI was correlated moderately with that of CI (r2 ⫽ 0.61, n ⫽ 36, p ⬍ 0.000, and r2 ⫽ 0.82, n ⫽ 24, p ⬍ 0.0001, respectively). Either actual or changed ITBVI was correlated moderately with that of SVI (r2 ⫽ 0.56, n ⫽ 36, p ⬍ 0.0001, and r2 ⫽ 0.82, n ⫽ 24, p ⬍ 0.0001, respectively). Actual ITBVI was correlated only slightly with actual CI (r2 ⫽ 0.32, n ⫽ 36, p ⬍ 0.001), even though a moderate correlation was observed between these two changed values (r2 ⫽ 0.69, n ⫽ 24, p ⬍ 0.0001). Discussion As judged by clinical conditions including amounts of intraoperative blood loss and peripheral circulatory state when hypotension developed, hypotension during the first 4 h of the study period may be mainly attributable to hypovolemia, even though neither arterial BP nor blood lactate levels immediately after admission to the ICU indicated the presence of apparent oxygen deficit in the peripheral tissues at that time, whereas hypotension thereafter may be mainly attributable to the redistribution of blood from the central to the peripheral tissues as judged by improved peripheral circulation. A simultaneous decrease in CVP, ITBVI, SVI, and CI in this study
Table 2—Cardiovascular Variables and Fluid Volumes* Variables
ICU Admission
Hypotension
Volume Loading
Heart rate, beats/min Systemic arterial pressure, mm Hg Mean arterial pressure, mm Hg CI, L/min/m2 SVI, mL/m2 CVP, mm Hg Hematocrit, % Plasma glucose, mmol/L† IDVGI, L/m2 ITBVI, L/m2
78 ⫾ 17 (50–120) 122 ⫾ 28 (87–174) 90 ⫾ 19 (64–121) 2.6 ⫾ 0.4 (2.0–3.4) 35 ⫾ 9 (17–48) 8 ⫾ 2 (4–12) 30.9 ⫾ 4.6 (24.3–39.1) 8.3 ⫾ 1.4 (5.6–10.1) 3.6 ⫾ 0.4 (3.1–4.2) 0.90 ⫾ 0.1 (0.67–1.12)
92 ⫾ 17 (75–125)‡ 88 ⫾ 9 (76–109)储 63 ⫾ 8 (50–76)储 2.5 ⫾ 0.5 (1.8–3.7) 29 ⫾ 8 (21–44)§ 7 ⫾ 3 (2–12)§ 31.2 ⫾ 5.0 (24.9–41.8) 8.1 ⫾ 1.8 (5.3–11.0) 3.4 ⫾ 0.5 (2.7–4.3) 0.81 ⫾ 0.15 (0.61-1.14)§
85 ⫾ 14 (58–106)¶ 105 ⫾ 16 (82–139)‡# 72 ⫾ 10 (53–91)储 3.5 ⫾ 0.5 (2.9–4.6)储** 42 ⫾ 9 (31–60)§** 9 ⫾ 3 (5–15)§** 25.9 ⫾ 4.3 (20.4–34.9)储** 8.7 ⫾ 1.9 (5.8–11.7) 4.1 ⫾ 0.5 (3.4–4.8)储** 0.95 ⫾ 0.14 (0.75–1.17)**
*Data are presented as mean ⫾ SD (range). †Immediately before glucose injection. ‡p ⬍ 0.05 compared with immediately after ICU admission. §p ⬍ 0.01 compared with immediately after ICU admission. 储p ⬍ 0.001 compared with immediately after ICU admission. ¶p ⬍ 0.05 compared with during hypotension. #p ⬍ 0.01 compared with during hypotension. **p ⬍ 0.001 compared with during hypotension. 1716
Clinical Investigations in Critical Care
Figure 3. Systolic arterial pressure immediately after admission to the ICU, during hypotension, and after fluid volume loading in each individual patient.
would support a decrease in cardiac preload or hypovolemic hypotension, whereas a decrease in CVP, ITBVI, and SVI, but not CI, would support a simultaneous decrease in cardiac preload and afterload. Changes in heart rate and systemic inflammatory response syndrome after esophagectomy17 would also play a role in determining CI at least partly. Presumably, there may be several contributory factors to affect CI in each individual patient of this study. Thus, it would be impossible to draw any definitive conclusion to the different responses of CI during the hypotensive episode in this study. Considering that IDVGI was correlated only slightly with either SVI or ITBVI in this study, IDVGI is not being a relevant surrogate measure of cardiac preload during unstable hemodynamic states. This is contrast to our experimental study13 in which IDVG following hemorrhage and fluid loading had a moderate correlation with ITBV (r2 ⫽ 0.76). Gabbanelli et al12 also demonstrated a moderate
linear correlation between IDVGI and ITBVI in critically ill patients (r2 ⫽ 0.79). In these two studies12,13 both fluid volumes were evaluated during a relatively steady hemodynamic state following experimental or therapeutic interventions. Presumably, the different measurement timing as well as underlying pathophysiology would yield different relationship between the two volumes in this study, even though the relationship between the changed volumes in this study were comparable with these two previous studies. In contrast to ITBVI, both CI and IDVGI were not consistently decreased when hypotension developed. However, the relationship between IDVGI and CI in this study (r2 ⫽ 0.67) was comparable with that of our previous daily based studies after esophagectomy (r2 ⫽ 0.50)10 and early after major burn (r2 ⫽ 0.69).18 We have also observed a linear correlation between IDVG and CO after an infusion of phentolamine that yields fluid redistribution from the central to the peripheral compartment.19 Furthermore, 22 plots of 24 changes in IDVGI and CI moved together to the same direction in this study, as shown in Figure 5. These findings would allow speculation that the IDVGI/CI relationship is consistent even during hypotension. Based on capillary membrane permeability of glucose, the rate at which glucose molecules diffuse through the capillary membrane is approximately 50 times greater than the rate at which plasma itself flows linearly along the capillary,20 suggesting that glucose can rapidly distribute throughout the central ECF compartment even in a low CO state. This is supported by the fact that a relatively large IDVG was observed even in the presence of a small CO in patients with congestive heart failure.4 Accordingly, IDVG may be variable depending on the actual size of the central ECF
Figure 4. Relationship between IDVGI and ITBVI. Shown are actual values (left, A): Y ⫽ 0.14X ⫹ 0.38 (r2 ⫽ 0.23, n ⫽ 36, p ⫽ 0.003); and changed values (right, B): Y ⫽ 0.2X – 0.035 (r2 ⫽ 0.69, n ⫽ 24, p ⬍ 0.0001). www.chestjournal.org
CHEST / 128 / 3 / SEPTEMBER, 2005
1717
Figure 5. Relationship between IDVGI, SVI, and CI. Shown are actual values between IDVGI and SVI (top left, A): Y ⫽ 12X – 8.3 (r2 ⫽ 0.39, n ⫽ 36, p ⬍ 0.0001); changed values between IDVG and SVI (top right, B): Y ⫽ 17X ⫺ 1.3 (r2 ⫽ 0.72, n ⫽ 24, p ⬍ 0.0001); actual values between IDVGI and CI (bottom left, C): Y ⫽ 0.94X – 0.59 (r2 ⫽ 0.61, n ⫽ 36, p ⬍ 0.0001); and changed values between IDVGI and CI (bottom right, D): Y ⫽ X ⫹ 0.14 (r2 ⫽ 0.82, n ⫽ 24, p ⬍ 0.0001).
volume independently of CO states. As IDVG is apparently greater than ITBV, IDVG would reflect not only cardiac preload or ITBV but also rapidly exchanging extravascular fluid volume or the central ECF volume. Nevertheless, a good correlation has been consistently observed between IDVG and CO in critically ill patients without congestive heart failure.4,10,12,14,18 Considering the result of the present study, IDVG measurement cannot be used equally as a measure of cardiac preload. However, it may be of benefit in decision making even immediately after hypotension developed whether volume loading or administration of vasopressors is required for its treatment, since IDVGI has a closer correlation with CI as compared with either ITBVI or SVI in this study. Although we administered fluid volume loading to all patients when hypotension developed in this study, it would have been more appropriate to administer vasoactive drugs such as norepinephrine, instead of fluid volume loading, in some patients associated with an increase in IDVGI or CI. In contrast to ITBV measurements, a considerable 1718
time interval should be allowed between repeated IDVG measurements. We found that the bias of IDVG measurement at a 30-min interval was 0.08 ⫾ 0.32 L in 25 hemodynamically stable patients,8 indicating that IDVG determination can be reliably repeated within a minimum 30 min. Moreover, we reported that IDVG can be approximated simply and rapidly with only two plasma samples: immediately before glucose challenge and 3 min after injection.21 By using the formula we reported,21 IDVG can be approximated even using a conventional glucose analyzer.6 According to our experience of ⬎ 3,700 IDVG determinations in the ICU, it can be performed during the routine fluid management even without taking the greatest care to stabilize plasma glucose levels. We observed a continuous decline in plasma glucose concentration over 60 min after injection, although plasma glucose concentrations at 60 min after injection remained slightly elevated as compared with the preinjection value.8 Hence, IDVG measurement will not induce a continued hyperglycemic state, even in critically ill Clinical Investigations in Critical Care
patients. Diaz-Parejo et al22 suggested that transient hyperglycemia had no adverse effect on outcome in patients with severe traumatic brain lesions and stroke. Therefore, we should be more concerned about normalization in basal plasma glucose concentration than about transient hyperglycemia in these patients. Conclusion Results demonstrate that IDVGI is correlated only slightly with ITBVI or SVI but moderately with CI during hemodynamically unstable states after esophagectomy. Results suggest that IDVGI should not equally be used as an alternative measure of ITBVI but may be useful as a measure of the central fluid volume that may affect CI even during hemodynamically unstable states. ACKNOWLEDGMENT: The authors thank Emeritus Prof. Matsuki A (Hakodate, Japan), Prof. A. H. Giesecke Jr (Dallas, TX), and Dr. P. Hollister (Misawa, Japan) for valuable comments and continued support of the study.
8
9
10
11
12
13
14
References 1 Lichtwarck-Aschoff M, Zeravik J, Pfeiffer UJ. Intrathoracic blood volume accurately reflects circulatory volume status in critically ill patients with mechanical ventilation. Intensive Care Med 1992; 18:142–147 2 Boussat S, Jacques B, Levy B, et al. Intravascular volume monitoring and extravascular lung water in septic patients with pulmonary edema. Intensive Care Med 2002; 28:712– 718 3 Iwakawa T, Ishihara H, Takamura K, et al. Measurements of extracellular fluid volume in highly perfused organs and lung water in hypo- and hypervolaemic dogs. Eur J Anaesthesiol 1998; 15:414 – 421 4 Ishihara H, Takamura K, Koh H, et al. Does the initial distribution volume of glucose reflect the central extracellular fluid volume status in critically ill patients? Infusionsther Transfusionsmed 1996; 23:196 –201 5 Ferrannini E, Smith JD, Cobelli C. Effect of insulin on the distribution and disposition of glucose in man. J Clin Invest 1985; 76:357–364 6 Ishihara H, Nakamura H, Okawa H, et al. Initial distribution volume of glucose can be approximated using a conventional glucose analyzer in the intensive care unit. Crit Care 2005; 9:R144 –R149 7 Mi W, Ishihara H, Sakai T, et al. Possible overestimation of
www.chestjournal.org
15 16 17 18 19 20 21
22
indocyanine green-derived plasma volume early after induction of anesthesia with propofol/fentanyl. Anesth Analg 2003; 97:1421–1427 Rose BO, Ishihara H, Okawa H, et al. Repeatability of measurements of the initial distribution volume of glucose in haemodynamically stable patients. J Clin Pharm Ther 2004; 29:317–323 Ishihara H, Matsuno S, Taguchi S, et al. Evaluation of fluid volume status with a glucose challenge test in a patient with acute adrenal insufficiency. J Anesth 1996; 10:69 –71 Ishihara H, Suzuki A, Okawa H, et al. The initial distribution volume of glucose rather than indocyanine green derived plasma volume is correlated with cardiac output following major surgery. Intensive Care Med 2000; 26:1441–1448 Ishihara H, Suzuki A, Okawa H, et al. Comparison of initial distribution volume of glucose and plasma volume in thoracic fluid-accumulated patients. Crit Care Med 2001; 29:1532– 1538 Gabbanelli V, Pntanetti S, Donati A, et al. Initial distribution volume of glucose as noninvasive indicator of cardiac preload: comparison with intrathoracic blood volume. Intensive Care Med 2004; 30:2067–2073 Nakamura H, Ishihara H, Okawa H, et al. Initial distribution volume of glucose is correlated with intrathoracic blood volume in hypovolemia and following volume loading in dogs. Eur J Anaesth 2005; 22:202–208 Suzuki A, Ishihara H, Okawa H, et al. Can initial distribution volume of glucose predict hypovolemic hypotension after radical surgery for esophageal cancer? Anesth Analg 2001; 92:1146 –1151 Sakka SG, Ru¨hl CC, Pfeiffer UJ, et al. Assessment of cardiac preload and extravascular lung water by single transpulmonary thermodilution. Intensive Care Med 2000; 26:180 –187 Akaike H. A new look at the statistical model identification. IEEE Trans Automatic Control 1974; AC-19:716 –723 Nakanishi K, Takeda S, Terajima K, et al. Myocardial dysfunction associated with proinflammatory cytokines after esophageal resection. Anesth Analg 2000; 91:270 –275 Ishihara H, Otomo N, Suzuki A, et al. Detection of capillary protein leakage by glucose and indocyanine green dilutions during the early post-burn period. Burns 1998; 24:525–531 Matsui A, Ishihara H, Suzuki A, et al. Glucose dilution can detect fluid redistribution following phentolamine infusion in dogs. Intensive Care Med 2000; 26:1131–1138 Guyton AC. The microcirculation and the lymphatic system. In: Guyton AC. Textbook of medical physiology. 10th ed. Philadelphia, PA: WB Saunders, 2000; 162–183 Hirota K, Ishihara H, Tsubo T, et al. Estimation of the initial distribution volume of glucose by an incremental plasma glucose level at 3 min after i.v. glucose in humans. Br J Clin Pharmacol 1999; 47:361–364 Diaz-Parejo P, Stahl N, Xu W, et al. Cerebral energy metabolism during transient hyperglycaemia in patients with severe brain trauma. Intensive Care Med 2003; 29:544 –550
CHEST / 128 / 3 / SEPTEMBER, 2005
1719