The hemostatic consequences of hemodilution have been the focus of in vitro (1–11) and in vivo (11–16) investigations for several years. In particular, the notion that hemodilution resulting from blood loss and asanguineous fluid administration results in hypercoagulability has been of particular interest (11–16), given the significant morbidity and mortality associated with perioperative thrombophilia (17,18). The diagnosis of hypercoagulability can be made based on clinical signs of thrombotic disease, based on changes in circulating procoagulant or anticoagulant activity, or based on changes in variables of hemostatic monitors such as the thrombelastograph® (TEG®) (1–6,8–15). Specifically, a decrease in the time to clot initiation (reaction time, R), an increase in the speed of clot propagation (angle, α), or an increase in clot strength (amplitude, A; or shear elastic modulus, G) indicates an enhancement of hemostasis or hypercoagulability as defined with TEG®. In the setting of hemodilution, in vitro (1–6,8–11) and in vivo (11–15) studies have demonstrated a decrease in R and increase in α values, whereas A or G decrease, connoting a clot with faster initiation and propagation, but with decreased strength. However, the degree of hemodilution, fluid used to dilute, platelet activity, and systemic response (e.g., secretion of procoagulant or anticoagulant) to specific fluids have been observed to result in hypocoagulability, no hemostatic change, or hypercoagulability (1–15). Although some investigations have determined putative mechanisms for in vitro (4) and in vivo (11) dilution-associated hypercoagulability determined with TEG®, the precise role of specific procoagulant and anticoagulants remained unclear.
The purpose of this study was to determine the effects of dilution on coagulation kinetics and clot strength determined with TEG® in a plasma-based system. This goal was achieved by addition of buffered saline to control citrated plasma with known procoagulant and anticoagulant activities. Furthermore, plasma deficient in antithrombin (AT) was also diluted to determine the role of this anticoagulant in dilution-associated hypercoagulability. Finally, diluted control plasma coagulation was activated with celite to determine whether dilution-mediated changes would be discernable in the presence of enhanced procoagulant activity.
Plasma-buffer samples to be subsequently described (330 μL) were placed with 10 μL of 0.9% NaCl or 1% celite (Sigma Chemicals, St. Louis, MO) in 0.9% NaCl and 20 μL of 200 mM CaCl2 into a disposable cup in a computer-controlled TEG® model 5000 (Haemoscope Corp., Niles, IL). The data were collected and analyzed with version 4.0 TEG® software (Haemoscope). Data were collected until stable amplitude was observed for 3 min as determined by software algorithms. The following standard variables were determined at 37°C: R time (R, min); α, degrees; A, mm; and G, dyne/cm2. G is a parametric measure of clot strength expressed in metric units calculated from A as follows: G = (5000 × A)/(100 − A). In addition to these variables, two new variables were determined: maximum thrombus generation (MTG, dynes · cm−2 · s−1) and total thrombus generation (TTG, dyne/cm2). MTG represents the first derivative of the velocity of the increase in clot strength, beginning as G begins to increase and ending after clot strength stabilizes. TTG is the area under the velocity curve, representing the total change in elastic resistance until clot strength stabilization occurs. In concept, MTG and TTG are, respectively, similar to maximum velocity and area under the curve determined with DyCoDerivaAu™ software with roTEG methodology (19). However, unlike maximum velocity and area under the curve, which are expressed as multiples of A (mm), MTG and TTG are expressed using metric units of elastic resistance that accurately describe changes in clot strength. These variables are depicted in Figure 1.
Pooled control plasma (Trinity Biotech, Ventura, CA) and AT-deficient plasma (American Diagnostica, Inc., Greenwich, CT) were reconstituted with deionized water and subsequently diluted with imidazole buffered saline (IBS, Trinity Biotech). IBS was chosen because it is a fluid often used in clinical hematology laboratories and does not interfere with clot-based or chromogenic assays of procoagulant and anticoagulant activity. The AT activity, coagulation factor activity, prothrombin time, and activated partial thromboplastin time values of these plasmas are depicted in Table 1. Control plasma was diluted 0%, 10%, 20%, 30%, 40%, and 50% with IBS before TEG® analyses. The control plasma was further subdivided into 2 categories, 1 exposed to 10 μL of 0.9% NaCl and the other activated with 1% celite as indicated previously. AT-deficient plasma was diluted 0%, 10%, 20%, and 30% with IBS before TEG® analyses without celite activation. All conditions were represented with n = 6 experiments because this number of experiments is required to obtain a β ≥ 0.8 with an α < 0.05 for most TEG® variables as demonstrated in previous in vitro studies with plasma (20,21).
Use of citrated blood in TEG® analyses is controversial to some authors because storage of blood in citrate followed by addition of exogenous Ca+2 results in TEG® data that may be different from fresh whole blood (22). To determine the Ca+2 concentration present in the aforementioned experiments, control plasma samples diluted as noted above had heparin added (10 U/mL) before Ca+2 addition at a ratio of 20 μL of 200 mM CaCl2 to 330 μL of diluted plasma and 10 μL of IBS. In another series of experiments, control plasma with added heparin had CaCl2 added in a ratio of 20 μL of CaCl2 to 330 μL of undiluted plasma with 10 μL of IBS before further dilution with IBS. Subsequently, Ca+2 concentration from diluted samples (n = 2) from both series of experiments was determined by using an analyzer (model 1306; Instrumentation Laboratory, Lexington, MA).
Variables were expressed as mean ± sd. Analyses of the effects of dilution with IBS on TEG® variables were conducted with one-way analysis of variance with the Holm-Sidak post hoc test for multiple comparisons or Student’s t-test as appropriate. The correlation of changes in dilution with changes in TEG® variables was determined with commercially available software (SigmaStat 3.0; SPSS Inc., Chicago, IL). Graphical representation and modeling of data were generated with commercially available software (Origin 7.0; OriginLab Corp., Northampton, MA). A P value < 0.05 was considered significant.
Dilution of control plasma resulted in a significant decrease in R values in samples diluted 10%, 20%, and 30% compared with samples diluted 0% as depicted in Table 2. However, R values of samples diluted 40% and 50% were not significantly different from undiluted samples. With regard to α values, samples diluted 40% and 50% had values significantly less than all other dilutions. A, G, and TTG values significantly decreased in a dilution-dependent manner, whereas changes in MTG did not reach statistical significance.
Because decreased R values only occurred after 10%–30% dilution in control plasma, AT-deficient plasma was diluted 0%–30%; the resultant TEG® variable values are displayed in Table 2. Unlike control plasma, AT-deficient plasma demonstrated no significant change in R values in response to dilution. Furthermore, undiluted AT-deficient plasma R values were significantly smaller (40%) than control plasma values. Neither α nor MTG significantly changed with dilution of AT-deficient plasma, but A, G, and TTG significantly decreased in a dilution-dependent manner.
Celite activation of undiluted control plasma resulted in R values significantly smaller (>75%) than control plasma without celite activation as depicted in Table 3. Similarly, α and MTG values of undiluted, celite-activated control plasma were significantly more than plasma not exposed to celite. However, A, G, and TTG were significantly smaller in undiluted control plasma activated with celite than in plasma without celite. Thus, whereas clot initiation and propagation were enhanced by celite activation, clot strength was decreased significantly.
Celite-activated plasma R values significantly increased with dilution, whereas α, A, G, MTG, and TTG all significantly decreased in a dilution-dependent manner. Because all α, A, G, MTG, and TTG values were different among dilutions, linear regression analyses were performed. Changes in dilution significantly correlated with changes in α (R2 = 0.88, P < 0.001), A (R2 = 0.96, P < 0.001), G (R2 = 0.96, P < 0.001), MTG (R2 = 0.93, P < 0.001), and TTG (R2 = 0.96, P < 0.001).
Addition of a fixed amount of CaCl2 to progressively diluted plasma resulted in an exponential increase in sample Ca+2 concentration as displayed in Figure 2. In contrast, progressive dilution of plasma and CaCl2 mixtures resulted in a linear decrease in sample Ca+2 concentration.
As with whole blood diluted in vitro (1–11), our plasma-based data demonstrate that moderate dilution (30%) is associated with a decreased time to clot initiation (R) that has been interpreted to denote hypercoagulability. A 40%–50% dilution resulted in an increase in R, probably secondary to dilution of not just AT but also key procoagulants. However, within 0%–30% dilution, no dilution-associated increase in clot propagation (α or MTG) was observed in our system, unlike in previous in vitro whole blood studies of hemodilution and TEG® (1–11). A possible explanation for this disparity may be that platelets are responsible for hemodilution-associated increases in clot propagation in whole blood, amplifying the effects of enhanced thrombin formation more than diluted plasma proteins alone. Indeed, platelet count per se does not markedly affect platelet-mediated effects on variables >100,000 per mm3 (23), and whole blood samples in healthy volunteers are not likely to have platelet counts ≤100,000 per mm3 after 10%–30% dilution.
As to the mechanism responsible for the decrease of time to clot initiation in plasma-based systems, our data derived from AT-deficient plasma indirectly demonstrate that hemodilution-associated changes in AT activity are responsible for changes in R values. R values derived from AT-deficient plasma did not change throughout the range of dilution (0%–30%) associated with decreased R values in control plasma. If a deficiency of another endogenous anticoagulant (e.g., protein C) significantly affected R values, dilution of AT-deficient plasma would have resulted in some dilution-associated decreases in R values. Furthermore, R values of undiluted AT-deficient plasma are 40% smaller than control plasma, demonstrating the importance of AT in modulating thrombin generation. Our findings are consistent with those of Ruttmann et al. (4) who demonstrated an increase in R values in diluted whole blood after addition of AT. Of interest, although these authors found that addition of AT completely normalized R values after dilution, α values did not decrease to undiluted values with AT addition (4). When combined, our data and those of our colleagues (4) support the concept that hemodilution-associated decrease in clot initiation is mediated by decreases in AT inhibition of coagulation factors, whereas hemodilution-associated changes in α values in whole blood are likely platelet-mediated in vitro.
Celite activation, a method often used to obtain TEG® data rapidly in clinical and laboratory settings (12,20,21), significantly modified the hemostatic response of plasma to in vitro dilution compared with plasma not exposed to celite. First, celite activation eliminated the AT-mediated changes in R by increasing thrombin generation and overwhelming AT-mediated inhibition. Furthermore, celite-mediated increases in thrombin generation in celite-activated plasma were reflected by an increase in clot propagation (α, MTG) but paradoxically with a concurrent decrease in clot strength (A, G, TTG) compared with plasma without celite. A possible explanation for this celite-mediated phenomenon may be that rapid thrombin generation and fibrin polymerization may result in faster compartmentalization of the forming clot, preventing diffusion of thrombin and factor XIII-mediated fibrin crosslinking, resulting in a fast-forming but weaker clot. Finally, by enhancing thrombin formation, celite activation seems to permit quantification of in vitro dilution, as evidenced by the strong linear relationships between changes in dilution and several TEG® variables (α, A, G, MTG, and TTG).
To mechanistically answer questions about dilution and the role played by changes in procoagulant and anticoagulant activity with TEG®, one must use plasma to avoid confounding effects of platelet activity, which varies from individual to individual. Furthermore, many types of congenitally deficient or immunodepleted plasma are available to mechanistically determine the role of specific hemostatic modulators in TEG®-based or other monitoring systems. However, there are differences between TEG®-derived variables obtained from citrated blood versus fresh whole blood after hemodilution with a balanced electrolyte, hydroxyethyl starch solution (19). One important difference between a diluted citrated blood sample mixed with CaCl2 compared with a whole blood sample that is only diluted may be Ca+2 concentration. In our experiments with diluted plasma, Ca+2 concentrations exponentially increased with dilution in response to exposure to a fixed quantity of CaCl2, a phenomenon likely caused by decreased concentrations of calcium-binding proteins (Fig. 2). In contrast, simple dilution of plasma and CaCl2 already combined resulted in a linear decrease in concentration (Fig. 2). Importantly, thrombin activity is not Ca+2-dependent at Ca+2 concentrations larger than 0.5 mM, and thereafter does not increase at concentrations as large as 6.5 mM (24). In terms of thrombin-dependent clot kinetics, our experimental approach with fixed addition of CaCl2 to diluted plasma always resulted in Ca+2 concentrations within the optimal range, whereas a 50% dilution of plasma and CaCl2 already combined was near the lower limit of optimal Ca+2 concentration. Thus, given the limitations of plasma-based systems required to mechanistically answer hemostatic questions with TEG®, we would recommend fixed addition of CaCl2 to experimental plasmas anticoagulated with sodium citrate.
It is important to note that interpretation of our data should be within the limitations of in vitro investigation. Although such investigations of the effects of dilution on hemostasis can determine the role of specific procoagulants/anticoagulants and the effects of particular solutions used to dilute blood, they cannot predict the hemostatic effects of hemodilution in vivo. For example, rabbits hemodiluted with 5% human albumin demonstrated hypercoagulability by TEG® variables not just because of a decrease in AT activity but secondary to an increase in factor VIII complex (11) an hour after hemodilution. In contrast, rabbits hemodiluted with lactated Ringer’s solution demonstrated hypocoagulability by TEG® variables because there was a significantly larger decrease in procoagulant activity than anticoagulant activity three hours after dilution (7). Humans tend to exhibit hypercoagulability by TEG® variables after hemodilution in the perioperative period (8,9,14,15), but the precise mechanism behind this phenomenon remains unclear. In sum, although dilution and diluent, per se, directly affect hemostasis in vitro, the systemic response to hemodilution may significantly overshadow these in vitro effects in vivo.
In conclusion, dilution-mediated decreases in AT activity were responsible for the decrease in time to clot initiation in our plasma-based system. Furthermore, celite activation prevented detection of dilution-mediated changes in AT activity but instead potentially allowed quantification of dilution by changes in TEG® variables such as α, A, G, MTG, and TTG. Thus, for future in vitro TEG®-based investigations, researchers could choose to use either unmodified or celite-activated plasma to determine the effects of pharmacological interventions (e.g., plasma expanders, antithrombotics) on AT or procoagulant function, respectively.
1. Ekseth K, Abildgaard L, Vegfors M, et al. The in vitro
effects of crystalloids and colloids on coagulation. Anaesthesia 2002;57:1102–8.
2. Fries D, Innerhofer P, Klingler A, et al. The effect of the combined administration of colloids and lactated Ringer’s solution on the coagulation system: an in vitro
study using thrombelastograph® coagulation analysis (ROTEG®). Anesth Analg 2002;94:1280–7.
3. Roche AM, James MFM, Grocott MPW, Mythen MG. Coagulation effects of in vitro
serial haemodilution with a balanced electrolyte hetastarch solution compared with a saline-based hetastarch solution and lactated Ringer’s solution. Anaesthesia 2002;57:950–5.
4. Ruttmann TG, James MFM, Lombard EH. Haemodilution-induced enhancement of coagulation is attenuated in vitro
by restoring antithrombin III to pre-dilution concentrations. Anaesth Intensive Care 2001;29:489–93.
5. Tobias MD, Wambold D, Pilla MA, Greer F. Differential effects of serial hemodilution with hydroxyethyl starch, albumin, and 0.9% saline on whole blood coagulation. J Clin Anesth 1998;10:366–71.
6. Ruttmann TG, James MFM, Viljoen JF. Haemodilution induces a hypercoagulable state. Br J Anaesth 1996;76:412–4.
7. Nielsen VG. Hemodilution with lactated Ringer’s solution causes hypocoagulability in rabbits. Blood Coagul Fibrinolysis 2004;15:55–9.
8. Ng KF, Lam CC, Chan LC. In vivo
effect of haemodilution with saline on coagulation: a randomized controlled trial. Br J Anaesth 2002;88:475–80.
9. Ruttmann TG, James MFM, Finlayson J. Effects on coagulation of intravenous crystalloid or colloid in patients undergoing peripheral vascular surgery. Br J Anaesth 2002;89:226–30.
10. Nielsen VG, Baird MS. Extreme hemodilution in rabbits: an in vitro
and in vivo
thrombelastographic analysis. Anesth Analg 2000;90:541–5.
11. McCammon AT, Wright JP, Figueroa M, Nielsen VG. Hemodilution with albumin, but not Hextend, results in hypercoagulability as assessed by thrombelastography in rabbits: role of heparin-dependent serpins and factor VIII complex. Anesth Analg 2002;95:844–50.
12. Mahla E, Lang T, Vicenzi MN, et al. Thromboelastography for monitoring prolonged hypercoagulability after major abdominal surgery. Anesth Analg 2001;92:572–7.
13. Karoutsos S, Nathan N, Lahrini A, et al. Thrombelastogram reveals hypercoagulability after administration of gelatin solution. Br J Anaesth 1999;82:175–7.
14. Ruttmann TG, James MF, Aronson I. In vivo
investigation into the effects of haemodilution with hydroxyethyl starch (200/0.5) and normal saline on coagulation. Br J Anaesth 1998;80:612–6.
15. Ng KFJ, Lo JWR. The development of hypercoagulability state, as measured by thrombelastography, associated with intraoperative surgical blood loss. Anaesth Intensive Care 1996;24:20–5.
16. Tuman KJ, Spiess BD, Mcarthy RJ, Ivankovich AD. Effects of progressive blood loss on coagulation as measured by thrombelastography. Anesth Analg 1987;66:856–63.
17. Caprini JA, Arcelus JI, Hoffman K, et al. Prevention of venous thromboembolism in North America: results of a survey among general surgeons. J Vasc Surg 1994;20:751–8.
18. Zahn HR, Skinner JA, Porteous MJ. The preoperative prevalence of deep vein thrombosis in patients with femoral neck fractures and delayed operation. Injury 1999;30:605–7.
19. Sørensen B, Johansen P, Christiansen K, et al. Whole blood coagulation thrombelastographic profiles employing minimal tissue factor activation. J Thromb Haemost 2003;1:551–8.
20. Nielsen VG, Crow JP. Peroxynitrite decreases rabbit tissue factor activity in vitro.
Anesth Analg 2004;98:668–71.
21. Nielsen VG, Crow JP, Mogal A, et al. Peroxynitrite decreases hemostasis in human plasma. Anesth Analg 2004;99:21–6.
22. Roche AM, James MFM, Grocott MPW, Mythen MG. Citrated blood does not reliably reflect fresh whole blood coagulability in trials of in vitro
hemodilution. Anesth Analg 2003;96:58–61.
23. Khurana S, Mattson JC, Westley S, et al. Monitoring platelet glycoprotein IIb/IIIa-fibrin interaction with tissue factor-activated thromboelastography. J Lab Clin Med 1997;130:401–11.
24. Ataullakhanov FI, Pohilko AV, Sinauridze EI, Volkova RI. Calcium threshold in human plasma clotting kinetics. Thromb Res 1994;75:383–94.