Perioperative IV administration of fluids is part of the routine management of surgical patients. Maintaining normovolemia, thus preventing hypoperfusion and ischemia, protects the patient from developing reperfusion injury. However, reports on increased mortality rates in aggressively resuscitated patients with torso injuries, as well as metaanalyses showing inferior outcomes of patients receiving colloids as compared with crystalloids, perpetuate the discussion on the optimal management of IV fluid resuscitation (1–3). One of the presumed mechanisms responsible for a poorer outcome of patients receiving colloid solutions is the influence of these solutions on the coagulation system (3). Although several in vitr. and a few in vivo studies have investigated the effects of colloids and normal saline solution on the hemostatic system, results remain inconclusive (4–12). Furthermore, clinical practice, as opposed to in vitr. studies, seldom uses colloids exclusively, and the consequences of an exclusively crystalloid volume regimen using isovolemic amounts of lactated Ringer’s (RL) solution are not known. To explore the effects of IV colloids and crystalloids on the hemostatic system, we investigated orthopedic patients during primary knee replacement surgery and randomized the volume supply with medium molecular weight medium-substituted hydroxyethyl starch or modified gelatin solution in addition to a basis infusion of lactated Ringer’s solution. Patients exclusively receiving lactated Ringer’s solution served as control. We investigated the influence of these fluids on primary platelet-mediated hemostasis and quality of clot formation by using the newly developed platelet function analyzer and ROTEG® analysis. In addition to measurements of standard coagulation variables and global coagulation tests, fibronectin concentrations were determined, as were concentrations of Factor VIII (FVIII), von Willebrand factor (vWF), and vWF ristocetin cofactor activity (vWFRCo). Methods The protocol for this study was approved by the local University Ethics Committee. Informed written consent was obtained from 60 consecutive patients undergoing primary knee replacement surgery with the tourniquet technique. Inclusion criteria were ASA physical status I–III, age <80 yr, and informed consent for study-related arterial blood sampling and the administration of combined spinal/epidural anesthesia. Exclusion criteria were contraindications for regional anesthesia and puncture of the radial artery, any known allergies, or primary or secondary hemostatic disorders (preoperative coagulation abnormalities, renal and liver dysfunction, or intake of aspirin or other platelet aggregation inhibitors). All patients received regional anesthesia (combined spinal/epidural) with plain bupivacaine 0.5% (spinal anesthesia 2.5–3 mL) and 0.25% (epidural anesthesia 8–10 mL) during and 2 h after surgery, respectively. Patients were actively warmed with fluid warmers and a convective warming system. All patients received 40 mg of enoxaparin subcutaneously (Lovenox®) 12 h before surgery and a second-generation cephalosporin during surgery. By use of a computer-generated randomization list, patients were assigned to receive medium-molecular-weight medium-substituted HES (6% Isohäs 200/0.5®, Fresenius, Pharma Austria GmbH, Graz, Austria; HES group, n = 20) or modified GEL solution (4% Gelofusin®, B. Braun, Maria Enzersdorf, Austria; GEL group, n = 20) in addition to a basis infusion of RL, or exclusively RL (Fresenius, Pharma Austria GmbH; RL group, n = 20) throughout the study period. Before spinal anesthesia was administered, all patients received 500 mL of RL to prevent hypotension associated with neuronal block. Subsequently, all patients received RL 5 mL · kg−1 · h−1 to correct the IV volume deficit resulting from the starving period and basal requirements. In addition to and according to randomization, patients in the HES, GEL, and RL groups intraoperatively received HES 3 mL · kg−1 · h−1, GEL 4 mL · kg−1 · h−1, or RL 10 mL · kg−1 · h−1 to maintain normovolemia. Furthermore, the blood loss occurring after tourniquet release was compensated for with group-specific IV fluids in the blood loss/fluids ratio of 1:1, 1:1.3, and 1:3 in the HES, GEL, and RL groups, respectively. In the event of suspected hypovolemia (mean arterial pressure [MAP] <20% of baseline, tachycardia, insufficient urine output), a group-specific fluid bolus was administered (HES, 2.3 mL · kg−1 · h−1; GEL, 3 mL · kg−1 · h−1; and RL, 7 mL · kg−1 · h−1). A hemoglobin value of <8.0 g/dL or physiological signs of anemia served as triggers for transfusion of leukocyte-filtered red cells. After surgery, the administered amounts of basis RL were reduced to 4 mL · kg−1 · h−1 at the observation ward, and blood loss was compensated for by group-specific fluid administration as during surgery. Arterial blood samples were obtained immediately after combined spinal/epidural anesthesia had been successfully administered, immediately before tourniquet release, at the end of the surgical procedure (arrival at the observation ward), and 2 h later. The initial steps of platelet-mediated hemostasis were investigated with the platelet function analyzer, PFA-100® (Dade/Behring, Marburg, Germany). This system has the advantage of simplicity, and measurements of clotting times (CT) have been proven superior to measurements of bleeding time (13). The PFA-100® has been described in detail elsewhere (14). In brief, the system measures the time needed to close a 150-μm aperture in a collagen membrane coated with epinephrine or adenosine diphosphate (disposable cartridges). For measurements, 800 μL of whole blood was drawn into 3.8% citrate-containing Vacutainers, and closure times were determined after a recommended resting time of 30 min by using both collagen/epinephrine (EPI-CT; normal range, 85–165 s) and collagen/ADP cartridges (ADP-CT; normal range, 62–100 s). Concentrations of vWF were measured with a chromogenic latex immunoassay, FVIII was determined coagulometrically, and vWFRCo was measured manually with an agglutination test. Quality of clot formation was analyzed by using modified ROTEG® coagulation analysis (ROTEG®; Pentapharm GmbH, Munich, Germany), which is based on the Thrombelastograph® system (TEG®) after Hartert. Modified ROTEG® coagulation analysis is a valuable tool in investigating the coagulation system (15,16). In addition, the automatic pipetting system of the used ROTEG® analyzer has the advantage of ease of use, and the specific power transduction system permits measurements that are not disturbed by movement or vibration (17). According to the manufacturer’s instructions and by using the automatic pipetting system, citrated whole blood was mixed with buffered CaCl2 (0.2 M) and tissue thromboplastin (rabbit brain extract) for monitoring of the extrinsic system. In a similar manner partial thromboplastin (rabbit brain) was used to monitor the intrinsic coagulation system. Finally, extrinsically activated tests were also performed in the presence of abciximab (Reo Pro®). Thus, measurements rely on the fibrinogen component of the formed clot. ROTEG® tracing (Fig. 1) permits evaluation of initiation of coagulation (CT), speed of clot formation (clot formation time; CFT), and quality of the formed clot (maximum clot firmness; MCF). Thus, as compared with conventional coagulation analysis, CT refers to reaction time, CFT to coagulation time, and MCF to maximal amplitude. Because MCF readings depend on measurement time (maximal value of clot firmness reached), for comparison of all tests we used the amplitude reached at 20 min (A20) instead of MCF. In addition, the calculated maximal clot elasticity [(100 × MCF)/(100 − MCF)] describes mechanical properties of the clot.Figure 1: ROTEG® tracing permits evaluation of initiation of coagulation (coagulation time; CT), speed of clot formation (clot formation time; CFT), and quality of the formed clot (maximum clot firmness; MCF), as well as reduction of the clot as a result of lysis (maximum lysis; ML). Thus, as compared with conventional thrombelastography, CT refers to reaction time (r), CFT to coagulation time (k), and MCF to maximal amplitude (MA). Measurement of CT (normal value: extrinsically activated, <50 s; intrinsically activated, <160 s) describes the beginning of the coagulation process (time from sample placement until a clot firmness of 2 mm) and is prolonged by deficient concentrations of coagulation factors. CFT (normal value, <180 s) illustrates the dynamics of the coagulation process (time interval from the end of the initial coagulation [CT] until a clot firmness of 20 mm) and depends on concentrations of coagulation factors, fibrinogen, and platelets. The MCF value (normal value, 53–74 mm) indicates the maximal reached clot firmness (diameter of the tracing), which depends on the interaction of platelets and fibrinogen. MCF tracing in the presence of the platelet antagonist abciximab (normal value, 6–16 mm) permits the fibrinogen part of the formed clot to be determined.By using routine laboratory methods, prothrombin time (PT; normal range, 70%–120%), activated partial thromboplastin time (normal range, 23–40 s), concentrations of antithrombin III (AT III; normal range, 80%–120%), fibrinogen (normal range, 190–380 mg/dL), hemoglobin, and platelet count were determined. Concentrations of fibronectin were measured with an immunological test (Nephelometer Analyzer II; Behring, Germany; normal range, 25–40 mg/dL). Data recorded included: patient demographics, time until tourniquet release, total time of surgery, estimated blood loss (intraoperative amount of blood in the suction device; postoperative loss into drains), MAP, heart rate, urinary output, and transfusion of leukocyte-filtered red cells. Data in tables are given as mean (sd) or median (interquartile range). Differences in baseline values were analyzed with the Kruskal-Wallis test. The area under the curve (AUCA–D) was calculated after subtracting the baseline value and was analyzed with the Kruskal-Wallis test and post hoc Wilcoxon’s ranked sum tests for comparison of between-group differences in the intraoperative response profile. To investigate time dependency, all groups were combined, and a repeated-measures analysis of variance (ANOVA) with time point as the repeated factor and group as the fixed factor was applied;P values are shown for the overall time factor independently of group differences. All variables were log-transformed in this analysis to compensate for the probable nonnormal nature of these laboratory values. The frequency of variables outside the normal range was analyzed with the Cochran-Mantel-Haenszel test. Correlations of EPI-CT, ADP-CT, vWFRCo, vWF, and FVIII were analyzed by Spearman rank correlation. A P value <0.05 was considered statistically significant. Results Patients in the HES, GEL, and RL groups were comparable regarding demographic and perioperative data (Table 1). One patient in the HES group intraoperatively received 1 U of red cells, and two patients in the GEL and RL groups each needed transfusion of 2 U of red cells. Hemodynamic variables and urinary output were similar in all groups at the end of the observation period, indicating that the intended isovolemic volume support was realized. Baseline measurements of investigated variables did not differ among groups, with the only exception that PT was higher in the GEL as compared with the HES and RL groups. This difference was statistically significant (P = 0.0369) but without clinical relevance, because all baseline PT measurements were within the normal range.Table 1: Descriptive and Perioperative Data of 60 Orthopedic Patients Undergoing Knee Replacement Surgery and Receiving Hydroxyethyl Starch 6% (HES) 200/0.5, Modified Gelatin 4% (GEL), or Lactated Ringer’s Solution Exclusively (RL)Measurements of primary platelet-mediated hemostasis: EPI-CT, ADP-CT and concentrations of FVIII, vWF, and vWFRCo showed a significant time dependency (repeated-measures ANOVA), indicating that impaired platelet-mediated hemostasis occurred with IV administration of fluids (Table 2). Comparison of the intraoperative response profile (AUCA–D) showed no statistically significant differences among groups (Table 2). EPI-CT correlated significantly inversely with vWFRCo (P = 0.0019), as did ADP-CT with vWFRCo, FVIII, and vWF concentrations (vWFRCo, P = 0.0001; FVIII, P = 0.0029; vWF, P = 0.0010).Table 2: Measurements of Factors Associated with Primary Platelet-Mediated Hemostasis During Knee Replacement Surgery and Randomized Volume Supply with Hydroxyethyl Starch 6% (HES) 200/0.5, Modified Gelatin 4% (GEL), or Lactated Ringer’s Solution (RL)ROTEG® analysis: for all variables, except CT (extrinsically activated) and CFT (extrinsically and intrinsically activated), a significant time dependency during the observation period was demonstrated, indicating that impaired clot strength occurred with IV administration of fluids (Tables 3 and 4).Table 3: Extrinsically (Ex) Activated ROTEG® Coagulation Analysis Measurements During Knee Replacement Surgery and Randomized Volume Supply with Hydroxyethyl Starch 6% (HES) 200/0.5, Modified Gelatin 4% (GEL), or Lactated Ringer’s Solution (RL)Table 4: Intrinsically (In) Activated ROTEG® Coagulation Analysis Measurements During Knee Replacement Surgery and Randomized Volume Supply with Hydroxyethyl Starch 6% (HES) 200/0.5, Modified Gelatin 4% (GEL), or Lactated Ringer’s Solution (RL)To detect differences among groups, the intraoperative response profile of coagulation analysis variables (AUCA–D) was compared. Generally, the least effects were seen when RL was used exclusively. Extrinsically activated initiation of coagulation was shortened in the RL group but prolonged in the colloid groups, whereas total clot firmness decreased comparably in all groups. However, the fibrin component of the clot and its maximal clot elasticity decreased to a significantly greater extent in the colloid groups, particularly in the HES group, as compared with the RL group (AUCA–D;Table 3). Intrinsically activated initiation of coagulation was shortened in the RL and GEL groups but lengthened in the HES group. Intrinsic total clot firmness decreased significantly more in the two colloid groups as compared with the RL group (AUCA–D;Table 4). Table 5). Measurements of routine coagulation tests and fibronectin showed a significant time dependency. Comparison of the intraoperative response profile showed that PT and AT III concentrations decreased significantly more in both colloid groups as compared with RL, whereas changes in activated partial thromboplastin time and platelet count were similar in all groups. The decrease in fibronectin concentrations was significantly larger when HES was used as compared with GEL. Concentrations of fibrinogen showed a trend toward a larger decrease in the two colloid groups (P = 0.0861), as compared with the RL group (AUCA–D; Measurements of vWFRCo below normal were more frequently observed in the GEL group (P = 0.0004). With colloids, total clot strength was more frequently below the normal range as compared with RL (intrinsic total clot firmness, P = 0.0001). Fibronectin concentrations were more frequently below the normal range in the RL and HES groups as compared with the GEL group (P = 0.0486).Table 5: Measurements of Routine Coagulation Tests and Fibronectin Concentrations During Knee Replacement Surgery and Randomized Volume Supply with Hydroxyethyl Starch 6% (HES) 200/0.5, Modified Gelatin 4% (GEL), or Lactated Ringer’s Solution (RL)Discussion This is the first in vivo study to investigate the influence of combined crystalloid/colloid IV administration on the coagulation system as compared with exclusively crystalloids. To exclude possible confounding factors, such as transfusion of red cells, fresh frozen plasma, and unequal depth of anesthesia, we investigated patients during knee replacement surgery with the tourniquet technique in regional anesthesia. Although blood loss was only moderate, IV administration of crystalloid/colloid as well as exclusively crystalloid fluids caused a significant decrease in the coagulation factors involved in platelet adhesion, a measurable impairment of platelet function, and changes in the speed and quality of clot formation. Generally, the least effects were observed in the group receiving RL exclusively. Changes in primary platelet-mediated hemostasis were comparable in all groups, whereas ROTEG® tracings were significantly more impaired with both colloids as compared with RL. AT III concentrations and PT measurements decreased significantly more with both colloids and fibronectin concentrations decreased least with GEL. Statistical analysis of fibrinogen concentrations showed only a weak trend toward a larger decrease with colloids; however, measurements also showed a high variability. Platelet adhesion to the injured endothelium represents the initial response of the hemostatic system to control bleeding and initiates a complex cascade that finally leads to formation of a stable fibrin clot. Impairment of platelet function, as seen with the von Willebrand syndrome or various platelet-active drugs, is associated with increased spontaneous and surgical bleeding (18). Thus, the presumed effect of colloids on platelet function is of clinical relevance because colloids are used mainly to bridge blood loss, especially in trauma patients. Stögermüller et al. (9) and de Jonge et al. (10) investigated platelet function by using the platelet function analyzer and flow cytometry, or aggregatometry after infusion of either 1000 mL of HES 200/0.62 or GEL, respectively, as compared with 1000 mL of NaCl. In these studies, primary platelet-mediated hemostasis was significantly impaired with colloids as compared with NaCl. Our finding that moderate amounts of colloids combined with crystalloid fluids, as used in clinical routine, affect primary platelet-mediated hemostasis in a manner comparable with RL given solely is in concordance with the findings of Hüttner et al. (11). These authors investigated patients undergoing major abdominal surgery and receiving HES 200/0.5, HES 70/0.5, or GEL in addition to a basal infusion of RL. The differences between the results of studies by Stögermüller et al. (9) and de Jonge et al. (10) and those of Hüttner et al.’s (11) and this study are most likely due to the different study designs. In studies reporting significant effects of colloids on platelet adhesion colloids or crystalloids were administered exclusively and in equal amounts, contrarily no significant effects were observed when crystalloids and colloids were combined and administered in isovolemic amounts. Thus, the administration of crystal- loid or crystalloid/colloid IV fluids does not specifically affect primary platelet-mediated hemostasis. According to the manufacturer’s information and our own clinical experience, prolongation of ROTEG® coagulation time reaching >200 s and clot formation time >300 s, reduction in clot firmness to <45 mm, and a fibrinogen component of <10–12 mm are associated with clinically relevant impairment of hemostasis. Because blood loss and amounts of administered IV fluids were moderate and comparable to our clinical routine, we did not expect to detect clinically significant impairment of hemostasis. However, some patients reached values of clot firmness below the lower normal range of 53 mm, and they reached this level more frequently with colloids. Furthermore, comparison of the intraoperative response profile (AUCA–D) showed statistically significant differences among groups. Most in vitr. studies confirm our in vivo data and show a greater impairment of clot strength when using colloids as compared with crystalloids (4–7). In addition, our own in vitr. investigations, including the combination of crystalloids and colloids, thus imitating clinical practice, showed comparable results. In this ROTEG® analysis study, clot firmness was least impaired when blood was diluted with RL alone, followed by the combination of RL and the various colloids tested (GEL, HES 130/0.4, and HES 200/0.5) (19). One of the mechanisms presumed to be responsible for colloid-reduced clot strength is decreased fibronectin concentration (6). In the presence of FXIIIa, fibronectin becomes incorporated in fibrin, thereby increasing fibrin fiber size and density (20). It is interesting to note that in this study, fibronectin concentrations were similar at baseline but decreased least with GEL and were more frequently below the normal range in the HES and RL solution groups. At the same time, clot strength was more reduced with both colloids as compared with RL. Thus, the significance of change in fibronectin concentration for clot strength remains unclear. In this study, the reduction in total clot strength seen with colloids was accompanied by a reduction in the fibrinogen component of the clot. In parallel, with both colloids, a trend toward a larger decrease in fibrinogen concentrations was observed. Because baseline measurements of ROTEG® tracings and fibrinogen concentrations were similar in all groups, as was blood loss and change in platelet count, our results suggest that with colloids, impaired fibrin polymerization, finally resulting in decreased clot strength, occurs parallel with increased fibrinogen consumption. The fact that deficiency of fibrinogen first appears when blood loss is compensated for with colloids and red blood cell components was previously reported by Hiippala et al. (21), who investigated 60 patients experiencing considerable blood loss during major urologic or abdominal surgery. Patients received albumin, dextran, or HES 120/0.7 until transfusion triggers (10%–20% loss of calculated blood volume) were reached. Unfortunately, the final colloid requirement was not mentioned. In addition, fresh frozen plasma was administered when abnormal bleeding occurred or 100% of the calculated blood volume was lost. Under these conditions and in the presence of large initial fibrinogen concentrations, as seen in our orthopedic patients, the critical threshold for fibrinogen concentration, namely, 100 mg/dL, was reached when blood loss exceeded 142% of the circulating blood volume. It was our intention to investigate the influence of IV fluids on the hemostatic system in patients usually exhibiting only moderate blood loss. Results therefore were not influenced by the need for transfusion of red blood cells and fresh frozen plasma. As expected, in this study, blood loss was in the range of only 10%, red cells were rarely transfused (5 of 60 patients received 1–2 U), and no fresh frozen plasma was administered. Under these conditions, fibrinogen concentrations decreased in the colloid groups by 100 mg/dL on average during the observation period, indicating that with IV administration of colloids, critically reduced fibrinogen concentrations might be reached earlier than expected. Besides the fact that clot strength was reduced slightly in the group receiving RL only, the initiation and propagation of coagulation were enhanced in the groups exclusively receiving RL or GEL. This hypercoagulability after moderate dilution with GEL and saline has also been described in other in vitr. and in vivo studies (7,8,12). Imbalances in thrombin generation and decreased AT III activity were discussed as a possible underlying mechanism. The results of this study cannot confirm this assumption, because AT III concentrations decreased least in the crystalloid group. In conclusion, the effects of colloids and crystalloids on primary platelet-mediated hemostasis were comparable between groups. However, colloids interfered more with fibrin polymerization and resulting total clot strength than did RL. Of the colloids, GEL exhibited a somewhat different pattern of action on the coagulation system as compared with HES 200/0.5. Because the side effects of HES depend on molecular weight and the degree of substitution, we expect that the differences between HES and GEL observed in our study could disappear with the use of the newly developed HES 130/0.4 formulation (19,22). Our results are not intended to encourage restrictive perioperative use of colloids, because colloids are essential for quick and effective treatment of hypovolemia. We assume that the observed effects of colloids on the hemostatic system could gain clinical relevance in the treatment of patients already exhibiting borderline fibrinogen levels at baseline and experiencing greater blood loss (23). We speculate that when continuing blood loss is bridged mainly by colloid infusion, fibrinogen concentrations can decrease critically, even before transfusion of red cells becomes necessary (24). Because fibrinogen plays an essential role in the hemostatic system, further blood loss might be reduced during deliberate colloidal volume supply by timely control of fibrinogen concentrations and by substituting the necessary fibrinogen. The authors thank Mirijam Schnapka, MD (Central Laboratory Institute, Innsbruck University Hospital), for kind technical assistance.