Anaesthesia 2015, 70, 549–554
doi:10.1111/anae.12990
Original Article The effect of hyperglycaemia on haemostasis testing – a volunteer study G. Lippi,1 R. Buonocore,2 R. Musa,3 L. Ippolito,3 A. Picanza2 and E. J. Favaloro4 1 Director and Professor, 2 Postdoctoral Fellow, 3 Senior Assistant, Laboratory of Clinical Chemistry and Hematology, Academic Hospital of Parma, Parma, Italy 4 Principal Scientist, Department of Haematology, Institute of Clinical Pathology and Medical Research, Pathology West, Westmead Hospital, Westmead, NSW, Australia
Summary We investigated whether the contamination of samples with glucose subsequently tested for haemostasis affected the results, including prothrombin time, activated partial thromboplastin time and fibrinogen concentration. Venous blood was collected from 12 healthy subjects and divided into four aliquots, which were subjected to different degrees of contamination with standard glucose solution (0%, 5%, 10%, 20%). With increasing glucose contamination, prothrombin time increased from mean (SD) 11.0 (0.7) s to 11.2 (0.7) s, 11.5 (0.7) s and 12.2 (0.8) s, all p < 0.001. Activated partial thromboplastin time decreased from 32.3 (0.9) s to 30.9 (0.8) s, 30.8 (0.8) s, and 29.7 (0.7) s, all p < 0.001. Fibrinogen concentration decreased from 3.8 (0.7) g.l 1 to 3.7 (0.6) g.l 1, 3.6 (0.6) g.l 1, and 3.4 (0.6) g.l 1, all p < 0.001. Bias was clinically meaningful from 5% contamination for activated partial thromboplastin time, 10% contamination for prothrombin time and 20% contamination for fibrinogen concentration. We conclude that if glucose contamination of haemostasis samples is suspected or has occurred, the specimens should not be analysed. .................................................................................................................................................................
Correspondence to: G. Lippi Email: [email protected] Accepted: 27 November 2014
Introduction The collection of blood through vascular access devices is commonplace in clinical practice, especially in intensive care units, emergency rooms, oncology or paediatric departments, as well as in patients requiring continuous therapy or fluid infusions [1]. Among the various types of potential errors that may occur during blood collection from vascular devices, blood sample contamination with glucose solutions has been described as an important cause of spurious hyperglycaemia [2, 3], which may occasionally lead to serious harm due to unwarranted prescription of insulin and consequent iatrogenic hypoglycaemia [4, 5]. © 2014 The Association of Anaesthetists of Great Britain and Ireland
Contamination with a glucose-containing solution was also found to have a profound impact on the measured values of other clinical chemistry parameters [6]. Blood contamination by exogenous glucose is a typical preanalytical error attributable to a variety of causes, including withdrawal of an insufficient volume of blood to clear the line before blood sampling, the use of glucose-containing solutions rather than saline to flush the line, or the collection of a blood sample from a site proximal to an intravenous line used for administering a glucose-containing solution. Although the adverse consequences in blood samples contaminated with glucose solutions with respect 549
Anaesthesia 2015, 70, 549–554
Lippi et al. | In-vitro hyperglycaemia and routine haemostasis testing
to the measurement of glucose and other clinical chemistry parameters seem to be widely acknowledged, there is little information regarding the effect on haemostasis testing. Interestingly, the current Clinical Laboratory Standards Institute (CLSI) document contains a specific issue (number 11.5) about sample contamination from exogenous fluids [7]; it is recommended that blood should be sampled from the opposite arm to that in which intravenous fluids are being administered. It is also stated that blood should not be sampled from the same limb unless there is no alternative and the tests are critical to patient care. This implicitly means that, under certain circumstances, collection of specimens from the same limb where intravenous fluids are being administered may still be considered. This interpretation may have come about following the paper by Read et al. [8], which showed that the collection of blood specimens proximal to an intravenous infusion, 3 min after stopping the infusion, produced a clinically meaningless dilutional effect, despite the fact that contaminant substances were detectable at a relatively high concentration in the infused solution. It is also noteworthy that the tests included glucose, sodium, chloride, and red blood cell count; however, no coagulation testing was performed. Additional and useful practice recommendations were recently released to prevent spurious hyperglycaemia during arterial line blood sampling [9], but they do not contain details about samples to be collected for haemostasis testing. In another document about collection, transport, and processing of blood specimens for plasma-based coagulation assays, there is only a vague conclusion that dilution of the specimen should be considered when blood is drawn from an intravenous line [10]. Accordingly, we performed a study to investigate whether contamination of haemostasis samples with a standard glucose solution might generate spurious results for the most commonly performed routine haemostasis tests: prothrombin time (PT); activated partial thromboplastin time (aPTT) and fibrinogen concentration.
Methods Our study was approved by the Institutional Review Board. The study population consisted of 12 fit and healthy volunteers, all of whom provided written 550
informed consent. None of our subjects had been taking any medication known to affect haemostasis testing, including non-steroidal anti-inflammatory drugs. Three venous blood specimens were collected at fasting from each subject in tubes containing 3.2% buffered sodium citrate (13 9 75 mm, 2.7-ml BD Vacutainerâ Plus plastic tubes; Becton Dickinson Italia S.p.A, Milan, Italy). The blood from each subject was pooled, gently mixed, and then divided into four aliquots of 2.0 ml each. The first whole blood aliquot was left untreated (i.e. no contamination), whereas a scalar amount of 100, 200 and 400 ll of a standard 5% glucose solution (25 g glucose monohydrate in 500 ml water, 278 mOsm.l 1; Baxter SPA, Rome, Italy) was added to the other three whole blood aliquots, to generate a final contamination with standard glucose solution of 5%, 10% and 20%, which is representative of the majority of glucose contaminated specimens received in our clinical laboratory [6]. The blood aliquots were left capped and in an upright position for exactly 1 h at room temperature, and citrate plasma was then obtained by conventional centrifugation (i.e. 1500 g for 15 min at room temperature). Routine haemostasis testing (i.e. PT, aPTT and fibrinogen concentration) was then performed using the ACL TOP 700 (Instrumentation Laboratory, Bedford, MA, USA). Specifically, aPTT was assayed using SynthASil (Instrumentation Laboratory), PT using RecombiPlasTin (Instrumentation Laboratory), and fibrinogen concentration using Fibrinogen-C XL (Instrumentation Laboratory). The analytical performance of these tests is reported elsewhere [11]. All measurements were completed within 2 h of collection of blood samples. After haemostasis testing was completed, glucose (hexokinase method) was also measured in citrated plasma using a Beckman DxC (Beckman Coulter Inc., Brea, CA, USA). Normal distribution of data was verified using the Kolmogorov–Smirnov test. We used the Mann–Whitney U-test and Pearson’s correlation coefficient to compare values, using Analyse-it (Analyse-it Software Ltd, Leeds, UK). The proportional bias of test results obtained in the contaminated aliquots was compared against the desirable quality specifications for bias [12], which reliably define the limits of clinically significant variations. © 2014 The Association of Anaesthetists of Great Britain and Ireland
Lippi et al. | In-vitro hyperglycaemia and routine haemostasis testing
Results
Anaesthesia 2015, 70, 549–554
quences of generating unreliable haemostasis results can be as detrimental for patient safety as accidental (spurious) hyperglycaemia [13]. The results of this study clearly show that whole blood contamination with as little as 5% of a standard glucose solution may be a source of significant bias, both analytical and clinical, in routine haemostasis testing. The values of fibrinogen concentration in our tests were progressively reduced to 10% in samples contaminated with 20% standard glucose solution. The proportionally lower reduction of values compared with the expected dilution factor is explainable by the fact that active permeation of glucose into blood cells is always accompanied by water influx and cell swelling [14, 15], thus ultimately reducing the dilutional effect in blood of non-diffusible elements such as fibrinogen. The prolongation of PT clotting times, to +11%, and the strong correlation with the degree of glucose solution contamination, suggest that the mechanism underlying the variation of this test may be similar to that of fibrinogen, and therefore mainly attributable to a dilutional effect of clotting factors of the extrinsic pathway, thus mimicking the prolongation observed with factor VII deficiency.
Measured plasma glucose concentration increased with increasing glucose contamination (Table 1). With increasing glucose contamination, measured PT increased, and aPTT and fibrinogen concentration decreased (Table 1). When test results were reported as percentage variation from the baseline uncontaminated specimen, the bias exceeded the quality specifications from 5% standard glucose solution contamination for aPTT, from 10% standard glucose solution contamination for PT and from 20% standard glucose solution contamination for fibrinogen concentration (Table 2). Interestingly, a significant correlation was observed between standard glucose solution contamination in the specimens and both PT (r = 0.999; p = 0.0005) and fibrinogen (r = 0.996; p = 0.0044) values, whereas the correlation between glucose contamination and APTT values did not reach statistical significance (r = 0.941; p = 0.0593) (Fig. 1).
Discussion In critically ill patients, as well as in those undergoing pre-operative laboratory testing, the potential conse-
Table 1 Results of prothrombin time (PT), activated partial thromboplastin time (aPTT) and fibrinogen concentration in samples with different degrees of contamination with a standard glucose solution. Values are mean (SD). Glucose solution contamination
PT; s aPTT; s Fibrinogen; g.l 1 Glucose; mmol.l 1
Uncontaminated sample
5%
11.0 32.3 3.8 4.4
11.2 30.9 3.7 19.2
(0.7) (0.9) (0.7) (0.4)
(0.7) (0.8) (0.6) (0.6)
p value
10%
< 0.0001 0.0026 < 0.0001 < 0.0001
11.5 30.8 3.6 33.2
(0.7) (0.8) (0.6) (1.0)
p value
20%
< 0.0001 0.0004 < 0.0001 < 0.0001
12.2 29.7 3.4 62.1
p value (0.8) (0.7) (0.6) (1.9)
< 0.0001 0.0002 < 0.0001 < 0.0001
p values – compared with baseline, uncontaminated plasma samples.
Table 2 Percentage variation of prothrombin time (PT), activated partial thromboplastin time (aPTT) and fibrinogen concentration in samples with different degree of contamination with a standard glucose solution. Values are mean (SD). Glucose solution contamination Quality specifications PT; % aPTT; % Fibrinogen; %
2.0 2.3 4.8
© 2014 The Association of Anaesthetists of Great Britain and Ireland
5% 2.5 (1.2) 4.3 (4.2) 1.9 (0.8)
10% 5.3 (1.6) 4.6 (3.6) 4.4 (0.9)
20% 11.0 (3.3) 7.6 (6.0) 10.4 (5.4)
551
Anaesthesia 2015, 70, 549–554
Lippi et al. | In-vitro hyperglycaemia and routine haemostasis testing
Figure 1 Mean percentage variations of prothrombin time (PT), activated partial thromboplastin time (aPTT) and fibrinogen concentration in blood samples with different degree of contamination with a standard glucose solution. Error bars indicate SD. At variance with PT, acute in-vitro hyperglycaemia produced an opposite effect on aPTT, generating a remarkable shortening of clotting times. It is noteworthy that three independent epidemiological studies recently showed that impaired glucose homoeostasis and frank hyperglycaemia are associated with shorter aPTT values [16–18], and this finding may have clinical implication since a shortened aPTT is now regarded as a prothrombotic risk factor [19]. This finding has been interpreted as the result of a multifactorial process, involving endothelial abnormalities with enhanced activation of platelets and increased concentration of clotting factors in the intrinsic pathway in patients with chronic hyperglycaemia. Although our data do not contradict these hypotheses, it seems reasonable to suggest that other mechanisms may play a role, since aPTT was substantially shortened as a result of acute hyperglycaemia in vitro, and hence an effective role of endothelial cells or dilution of intrinsic pathway clotting factors could be ruled out. The existence of a complex and as yet unexplained mechanism is also supported by the relationship between glucose and aPTT values, which could not be adequately described by a linear fit (p = 0.0593) (Fig. 1). A potential explanation for this finding is the theoretical effect of glucose on prekallikrein, a multifunctional serine protease involved in the activation phase of the intrinsic pathway of blood coagulation [20]. Animal studies have recently shown that activation of prekallikrein to kallikrein effectively occurs during hyperglycaemia [21], which would then be followed by downstream activation of the intrinsic pathway and, consequently, by shortening of the aPTT. This evidence has been confirmed in human studies, in which prekallikrein activity was found to be substantially enhanced in diabetic subjects compared with controls [22, 23], and is also consistent with the evidence that platelet-, erythrocyte- and monocyte-derived microparticles are actively generated in diabetics, and may then be
552
© 2014 The Association of Anaesthetists of Great Britain and Ireland
Lippi et al. | In-vitro hyperglycaemia and routine haemostasis testing
responsible for the activation of coagulation factor XII in the presence of prekallikrein [24]. Regardless of the underlying sources of variation in routine haemostasis tests attributable to acute hyperglycaemia, our results may have some practical implications. First, we clearly showed that minor contamination of coagulation samples with even 5% of a standard glucose-containing solution (i.e. approximately 0.1 ml in a conventional 2.7-ml blood tube), already causes a clinically meaningful bias in aPTT test results, whereas a clinically significant impact on PT and fibrinogen concentration was evident at 10% and 20% contamination. This is noteworthy since the 5%, 10% and 20% glucose contaminations used in our experimental study correspond to glucose levels of 19.2 mmol.l 1 (346 mg.dl 1), 33.2 mmol.l 1 (598 mg.dl 1) and 62.1 mmol.l 1 (1118 mg.dl 1) respectively, and hence are representative of the largest number of contaminated samples received in a conventional clinical laboratory [6], as well as of invivo hyperglycaemia frequently found in critical patients. Despite being universally recognised that blood contamination from exogenous glucose solutions has a substantial impact on the quality and reliability of glucose and clinical chemistry testing [6], these results suggest that the concept should be broadened to embrace routine haemostasis testing. As the prekallikrein/kallikrein system also has many other functional effects, including non-haemostasis based [25], one could hypothesise additional effects on laboratorybased tests resulting from hyperglycaemia-based activation. In particular, additional studies should be planned to establish whether this interference may also affect thromboelastography and thromboelastometry [26]. In fact, the reagents used for activating blood coagulation are the same as for aPTT (INTEM activates the contact phase of haemostasis using ellagic acid), and PT (EXTEM activates haemostasis via the physiological activator tissue factor), whereas FIBTEM is an EXTEM-based assay for assessing the contribution of fibrinogen to clot formation [27]. It is hence predictable that spurious hyperglycaemia may generate a type of interference in thrombo elastography/ thrombo-elastometry that is similar to that observed in plasma based clotting tests. Therefore, whenever glucose contamination of a haemostasis sample is © 2014 The Association of Anaesthetists of Great Britain and Ireland
Anaesthesia 2015, 70, 549–554
suspected or clearly identified, the specimen should not be analysed since the data would be unreliable and may affect clinical decision-making, thus ultimately jeopardising patient safety. As such, we suggest that current recommendations should be strengthened and that blood samples should never be taken from the same limb where intravenous glucose is administered. Another clinical implication of this study is the direct effect of acute hyperglycaemia on aPTT. The fact that an enhanced glucose concentration in blood may trigger per se a shortening of the clotting time of this test in vitro provides a further contribution to the debate as to whether a shortened aPTT may be a cause or a consequence of diabetes [19]. The acute effect of hyperglycaemia on aPTT that we have observed in our investigation, along with the evidence that prekallikrein-mediated activation of the contact phase occurs in the hyperglycaemic state [24], seems to support the hypothesis that acute hyperglycaemia may indeed activate the contact phase of blood coagulation, and hence provides another reasonable explanation in support of the epidemiological observation that aPTT is reduced in diabetics. Finally, intuitively translating these results into clinical practice, it is conceivable that the bias observed in our experimental conditions may also be expected in vivo. Accordingly, a substantial enhancement of blood glucose level may be regarded as a potential source of variability in haemostasis testing.
Competing interests No external funding and no competing interests declared.
References
lova D, et al. Preanalytical 1. Lippi G, Becan-McBride K, Behu quality improvement: in quality we trust. Clinical Chemistry and Laboratory Medicine 2013; 51: 229–41. 2. O’Hare D, Chilvers RJ. Arterial blood sampling practices in intensive care units in England and Wales. Anaesthesia 2001; 56: 568–71. 3. Leslie RA, Gouldson S, Habib N, et al. Management of arterial lines and blood sampling in intensive care: a threat to patient safety. Anaesthesia 2013; 68: 1114–9. 4. Gupta KJ, Cook TM. Accidental hypoglycaemia caused by an arterial flush drug error: a case report and contributory causes analysis. Anaesthesia 2013; 68: 1179–87. 5. Thirugnanam M, French J. Accidental hypoglycaemia caused by an arterial flush drug error. Anaesthesia 2014; 69: 524–5. 553
Anaesthesia 2015, 70, 549–554
Lippi et al. | In-vitro hyperglycaemia and routine haemostasis testing
6. Lippi G, Avanzini R, Sandei F, Aloe R, Cervellin G. Blood sample contamination by glucose-containing solutions: effects and identification. British Journal of Biomedical Sciences 2013; 70: 180–3. 7. Clinical Laboratory Standards Institute. Procedures for Collection of Diagnostic Blood Specimens by Venipuncture; Approved Guideline – 6th Ed. CLSI document H3-A6. Wayne, PA: Clinical and Laboratory Standards Institute, 2007. 8. Read DC, Viera H, Arkin C. Effect of drawing blood specimens proximal to an in-place but discontinued intravenous solution. Can blood be drawn above the site of a shut-off i.v.? American Journal of Clinical Pathology 1988; 90: 702–6. 9. Association of Anaesthetists of Great Britain and Ireland. Arterial line blood sampling: preventing hypoglycaemic brain injury 2014. Anaesthesia 2014; 69: 380–5. 10. Adcock DM, Hoefner DM, Kottke-Marchant K, Marlar RA, Szamosi DI, Warunek DJ. Collection, Transport, and Processing of Blood Specimens for Testing Plasma-based Coagulation Assays and Molecular Hemostasis Assays: Approved Guideline-fifth Edition. CLSI document H21-A5. Wayne, PA: Clinical Laboratory Standards Institute, 2008. 11. Milos M, Herak D, Kuric L, Horvat I, Zadro R. Evaluation and performance characteristics of the coagulation system: ACL TOP analyzer – HemosIL reagents. International Journal of Laboratory Hematology 2009; 31: 26–35. 12. Ricos C, Alvarez V, Cava F, Garcia-Lario JV, et al. Current databases on biologic variation: pros, cons and progress. Scandinavian Journal of Clinical and Laboratory Investigation 1999; 59: 491–500. 13. Plebani M, Favaloro EJ, Lippi G. Patient safety and quality in laboratory and hemostasis testing: a renewed loop? Seminars in Thrombosis and Hemostasis 2012; 38: 553–8. 14. Vander AJ, Sherman JH, Luciano DS. Human Physiology – The Mechanisms of Body Function, 8th edn. New York: McGraw Hill, 2001: 115–41. ISBN: 0071183825. 15. Gallagher PG. Disorders of red cell volume regulation. Current Opinion in Hematology 2013; 20: 201–7. 16. Lippi G, Franchini M, Targher G, et al. Epidemiological association between fasting plasma glucose and shortened APTT. Clinical Biochemistry 2009; 42: 118–20.
554
17. Zhao Y, Zhang J, Zhang J, Wu J. Diabetes mellitus is associated with shortened activated partial thromboplastin time and increased fibrinogen values. PLoS One 2011; 6: e16470. 18. Sapkota B, Shrestha SK, Poudel S. Association of activated partial thromboplastin time and fibrinogen level in patients with type II diabetes mellitus. BMC Research Notes 2013; 6: 485. 19. Lippi G, Salvagno GL, Ippolito L, Franchini M, Favaloro EJ. Shortened activated partial thromboplastin time: causes and management. Blood Coagulation and Fibrinolysis 2010; 21: 459–63. 20. Girolami A, Vidal J, Salagh M, et al. The old and the new in prekallikrein deficiency: historical context and a family from Argentina with PK deficiency due to a new mutation (Arg541Gln) in exon 14 associated with a common polymorphysm (Asn124Ser) in exon 5. Seminars in Thrombosis and Hemostasis 2014; 40: 592–9. 21. Liu J, Gao BB, Clermont AC, et al. Hyperglycemia-induced cerebral hematoma expansion is mediated by plasma kallikrein. Nature Medicine 2011; 17: 206–10. 22. Patrassi GM, Vettor R, Padovan D, Girolami A. Contact phase of blood coagulation in diabetes mellitus. European Journal of Clinical Investigation 1982; 12: 307–11. 23. Federspil G, Vettor R, De Palo E, Padovan D, Sicolo N, Scandellari C. Plasma kallikrein activity in human diabetes mellitus. Metabolism 1983; 32: 540–2. 24. Feener EP, Zhou Q, Fickweiler W. Role of plasma kallikrein in diabetes and metabolism. Thrombosis and Haemostasis 2013; 110: 434–41. 25. de Maat S, de Groot PG, Maas C. Contact system activation on endothelial cells. Seminars in Thrombosis and Hemostasis 2014; 40: 887–94. 26. Quarterman C, Shaw M, Johnson I, Agarwal S. Intra- and intercentre standardisation of thromboelastography (TEGâ). Anaesthesia 2014; 69: 883–90. €rnberg J, Asmis LM, Seifert B, Spahn DR. 27. Theusinger OM, Nu Rotation thromboelastometry (ROTEM) stability and reproducibility over time. European Journal of Cardiothoracic Surgery 2010; 37: 677–83.
© 2014 The Association of Anaesthetists of Great Britain and Ireland
doi:10.1111/anae.12990
Original Article The effect of hyperglycaemia on haemostasis testing – a volunteer study G. Lippi,1 R. Buonocore,2 R. Musa,3 L. Ippolito,3 A. Picanza2 and E. J. Favaloro4 1 Director and Professor, 2 Postdoctoral Fellow, 3 Senior Assistant, Laboratory of Clinical Chemistry and Hematology, Academic Hospital of Parma, Parma, Italy 4 Principal Scientist, Department of Haematology, Institute of Clinical Pathology and Medical Research, Pathology West, Westmead Hospital, Westmead, NSW, Australia
Summary We investigated whether the contamination of samples with glucose subsequently tested for haemostasis affected the results, including prothrombin time, activated partial thromboplastin time and fibrinogen concentration. Venous blood was collected from 12 healthy subjects and divided into four aliquots, which were subjected to different degrees of contamination with standard glucose solution (0%, 5%, 10%, 20%). With increasing glucose contamination, prothrombin time increased from mean (SD) 11.0 (0.7) s to 11.2 (0.7) s, 11.5 (0.7) s and 12.2 (0.8) s, all p < 0.001. Activated partial thromboplastin time decreased from 32.3 (0.9) s to 30.9 (0.8) s, 30.8 (0.8) s, and 29.7 (0.7) s, all p < 0.001. Fibrinogen concentration decreased from 3.8 (0.7) g.l 1 to 3.7 (0.6) g.l 1, 3.6 (0.6) g.l 1, and 3.4 (0.6) g.l 1, all p < 0.001. Bias was clinically meaningful from 5% contamination for activated partial thromboplastin time, 10% contamination for prothrombin time and 20% contamination for fibrinogen concentration. We conclude that if glucose contamination of haemostasis samples is suspected or has occurred, the specimens should not be analysed. .................................................................................................................................................................
Correspondence to: G. Lippi Email: [email protected] Accepted: 27 November 2014
Introduction The collection of blood through vascular access devices is commonplace in clinical practice, especially in intensive care units, emergency rooms, oncology or paediatric departments, as well as in patients requiring continuous therapy or fluid infusions [1]. Among the various types of potential errors that may occur during blood collection from vascular devices, blood sample contamination with glucose solutions has been described as an important cause of spurious hyperglycaemia [2, 3], which may occasionally lead to serious harm due to unwarranted prescription of insulin and consequent iatrogenic hypoglycaemia [4, 5]. © 2014 The Association of Anaesthetists of Great Britain and Ireland
Contamination with a glucose-containing solution was also found to have a profound impact on the measured values of other clinical chemistry parameters [6]. Blood contamination by exogenous glucose is a typical preanalytical error attributable to a variety of causes, including withdrawal of an insufficient volume of blood to clear the line before blood sampling, the use of glucose-containing solutions rather than saline to flush the line, or the collection of a blood sample from a site proximal to an intravenous line used for administering a glucose-containing solution. Although the adverse consequences in blood samples contaminated with glucose solutions with respect 549
Anaesthesia 2015, 70, 549–554
Lippi et al. | In-vitro hyperglycaemia and routine haemostasis testing
to the measurement of glucose and other clinical chemistry parameters seem to be widely acknowledged, there is little information regarding the effect on haemostasis testing. Interestingly, the current Clinical Laboratory Standards Institute (CLSI) document contains a specific issue (number 11.5) about sample contamination from exogenous fluids [7]; it is recommended that blood should be sampled from the opposite arm to that in which intravenous fluids are being administered. It is also stated that blood should not be sampled from the same limb unless there is no alternative and the tests are critical to patient care. This implicitly means that, under certain circumstances, collection of specimens from the same limb where intravenous fluids are being administered may still be considered. This interpretation may have come about following the paper by Read et al. [8], which showed that the collection of blood specimens proximal to an intravenous infusion, 3 min after stopping the infusion, produced a clinically meaningless dilutional effect, despite the fact that contaminant substances were detectable at a relatively high concentration in the infused solution. It is also noteworthy that the tests included glucose, sodium, chloride, and red blood cell count; however, no coagulation testing was performed. Additional and useful practice recommendations were recently released to prevent spurious hyperglycaemia during arterial line blood sampling [9], but they do not contain details about samples to be collected for haemostasis testing. In another document about collection, transport, and processing of blood specimens for plasma-based coagulation assays, there is only a vague conclusion that dilution of the specimen should be considered when blood is drawn from an intravenous line [10]. Accordingly, we performed a study to investigate whether contamination of haemostasis samples with a standard glucose solution might generate spurious results for the most commonly performed routine haemostasis tests: prothrombin time (PT); activated partial thromboplastin time (aPTT) and fibrinogen concentration.
Methods Our study was approved by the Institutional Review Board. The study population consisted of 12 fit and healthy volunteers, all of whom provided written 550
informed consent. None of our subjects had been taking any medication known to affect haemostasis testing, including non-steroidal anti-inflammatory drugs. Three venous blood specimens were collected at fasting from each subject in tubes containing 3.2% buffered sodium citrate (13 9 75 mm, 2.7-ml BD Vacutainerâ Plus plastic tubes; Becton Dickinson Italia S.p.A, Milan, Italy). The blood from each subject was pooled, gently mixed, and then divided into four aliquots of 2.0 ml each. The first whole blood aliquot was left untreated (i.e. no contamination), whereas a scalar amount of 100, 200 and 400 ll of a standard 5% glucose solution (25 g glucose monohydrate in 500 ml water, 278 mOsm.l 1; Baxter SPA, Rome, Italy) was added to the other three whole blood aliquots, to generate a final contamination with standard glucose solution of 5%, 10% and 20%, which is representative of the majority of glucose contaminated specimens received in our clinical laboratory [6]. The blood aliquots were left capped and in an upright position for exactly 1 h at room temperature, and citrate plasma was then obtained by conventional centrifugation (i.e. 1500 g for 15 min at room temperature). Routine haemostasis testing (i.e. PT, aPTT and fibrinogen concentration) was then performed using the ACL TOP 700 (Instrumentation Laboratory, Bedford, MA, USA). Specifically, aPTT was assayed using SynthASil (Instrumentation Laboratory), PT using RecombiPlasTin (Instrumentation Laboratory), and fibrinogen concentration using Fibrinogen-C XL (Instrumentation Laboratory). The analytical performance of these tests is reported elsewhere [11]. All measurements were completed within 2 h of collection of blood samples. After haemostasis testing was completed, glucose (hexokinase method) was also measured in citrated plasma using a Beckman DxC (Beckman Coulter Inc., Brea, CA, USA). Normal distribution of data was verified using the Kolmogorov–Smirnov test. We used the Mann–Whitney U-test and Pearson’s correlation coefficient to compare values, using Analyse-it (Analyse-it Software Ltd, Leeds, UK). The proportional bias of test results obtained in the contaminated aliquots was compared against the desirable quality specifications for bias [12], which reliably define the limits of clinically significant variations. © 2014 The Association of Anaesthetists of Great Britain and Ireland
Lippi et al. | In-vitro hyperglycaemia and routine haemostasis testing
Results
Anaesthesia 2015, 70, 549–554
quences of generating unreliable haemostasis results can be as detrimental for patient safety as accidental (spurious) hyperglycaemia [13]. The results of this study clearly show that whole blood contamination with as little as 5% of a standard glucose solution may be a source of significant bias, both analytical and clinical, in routine haemostasis testing. The values of fibrinogen concentration in our tests were progressively reduced to 10% in samples contaminated with 20% standard glucose solution. The proportionally lower reduction of values compared with the expected dilution factor is explainable by the fact that active permeation of glucose into blood cells is always accompanied by water influx and cell swelling [14, 15], thus ultimately reducing the dilutional effect in blood of non-diffusible elements such as fibrinogen. The prolongation of PT clotting times, to +11%, and the strong correlation with the degree of glucose solution contamination, suggest that the mechanism underlying the variation of this test may be similar to that of fibrinogen, and therefore mainly attributable to a dilutional effect of clotting factors of the extrinsic pathway, thus mimicking the prolongation observed with factor VII deficiency.
Measured plasma glucose concentration increased with increasing glucose contamination (Table 1). With increasing glucose contamination, measured PT increased, and aPTT and fibrinogen concentration decreased (Table 1). When test results were reported as percentage variation from the baseline uncontaminated specimen, the bias exceeded the quality specifications from 5% standard glucose solution contamination for aPTT, from 10% standard glucose solution contamination for PT and from 20% standard glucose solution contamination for fibrinogen concentration (Table 2). Interestingly, a significant correlation was observed between standard glucose solution contamination in the specimens and both PT (r = 0.999; p = 0.0005) and fibrinogen (r = 0.996; p = 0.0044) values, whereas the correlation between glucose contamination and APTT values did not reach statistical significance (r = 0.941; p = 0.0593) (Fig. 1).
Discussion In critically ill patients, as well as in those undergoing pre-operative laboratory testing, the potential conse-
Table 1 Results of prothrombin time (PT), activated partial thromboplastin time (aPTT) and fibrinogen concentration in samples with different degrees of contamination with a standard glucose solution. Values are mean (SD). Glucose solution contamination
PT; s aPTT; s Fibrinogen; g.l 1 Glucose; mmol.l 1
Uncontaminated sample
5%
11.0 32.3 3.8 4.4
11.2 30.9 3.7 19.2
(0.7) (0.9) (0.7) (0.4)
(0.7) (0.8) (0.6) (0.6)
p value
10%
< 0.0001 0.0026 < 0.0001 < 0.0001
11.5 30.8 3.6 33.2
(0.7) (0.8) (0.6) (1.0)
p value
20%
< 0.0001 0.0004 < 0.0001 < 0.0001
12.2 29.7 3.4 62.1
p value (0.8) (0.7) (0.6) (1.9)
< 0.0001 0.0002 < 0.0001 < 0.0001
p values – compared with baseline, uncontaminated plasma samples.
Table 2 Percentage variation of prothrombin time (PT), activated partial thromboplastin time (aPTT) and fibrinogen concentration in samples with different degree of contamination with a standard glucose solution. Values are mean (SD). Glucose solution contamination Quality specifications PT; % aPTT; % Fibrinogen; %
2.0 2.3 4.8
© 2014 The Association of Anaesthetists of Great Britain and Ireland
5% 2.5 (1.2) 4.3 (4.2) 1.9 (0.8)
10% 5.3 (1.6) 4.6 (3.6) 4.4 (0.9)
20% 11.0 (3.3) 7.6 (6.0) 10.4 (5.4)
551
Anaesthesia 2015, 70, 549–554
Lippi et al. | In-vitro hyperglycaemia and routine haemostasis testing
Figure 1 Mean percentage variations of prothrombin time (PT), activated partial thromboplastin time (aPTT) and fibrinogen concentration in blood samples with different degree of contamination with a standard glucose solution. Error bars indicate SD. At variance with PT, acute in-vitro hyperglycaemia produced an opposite effect on aPTT, generating a remarkable shortening of clotting times. It is noteworthy that three independent epidemiological studies recently showed that impaired glucose homoeostasis and frank hyperglycaemia are associated with shorter aPTT values [16–18], and this finding may have clinical implication since a shortened aPTT is now regarded as a prothrombotic risk factor [19]. This finding has been interpreted as the result of a multifactorial process, involving endothelial abnormalities with enhanced activation of platelets and increased concentration of clotting factors in the intrinsic pathway in patients with chronic hyperglycaemia. Although our data do not contradict these hypotheses, it seems reasonable to suggest that other mechanisms may play a role, since aPTT was substantially shortened as a result of acute hyperglycaemia in vitro, and hence an effective role of endothelial cells or dilution of intrinsic pathway clotting factors could be ruled out. The existence of a complex and as yet unexplained mechanism is also supported by the relationship between glucose and aPTT values, which could not be adequately described by a linear fit (p = 0.0593) (Fig. 1). A potential explanation for this finding is the theoretical effect of glucose on prekallikrein, a multifunctional serine protease involved in the activation phase of the intrinsic pathway of blood coagulation [20]. Animal studies have recently shown that activation of prekallikrein to kallikrein effectively occurs during hyperglycaemia [21], which would then be followed by downstream activation of the intrinsic pathway and, consequently, by shortening of the aPTT. This evidence has been confirmed in human studies, in which prekallikrein activity was found to be substantially enhanced in diabetic subjects compared with controls [22, 23], and is also consistent with the evidence that platelet-, erythrocyte- and monocyte-derived microparticles are actively generated in diabetics, and may then be
552
© 2014 The Association of Anaesthetists of Great Britain and Ireland
Lippi et al. | In-vitro hyperglycaemia and routine haemostasis testing
responsible for the activation of coagulation factor XII in the presence of prekallikrein [24]. Regardless of the underlying sources of variation in routine haemostasis tests attributable to acute hyperglycaemia, our results may have some practical implications. First, we clearly showed that minor contamination of coagulation samples with even 5% of a standard glucose-containing solution (i.e. approximately 0.1 ml in a conventional 2.7-ml blood tube), already causes a clinically meaningful bias in aPTT test results, whereas a clinically significant impact on PT and fibrinogen concentration was evident at 10% and 20% contamination. This is noteworthy since the 5%, 10% and 20% glucose contaminations used in our experimental study correspond to glucose levels of 19.2 mmol.l 1 (346 mg.dl 1), 33.2 mmol.l 1 (598 mg.dl 1) and 62.1 mmol.l 1 (1118 mg.dl 1) respectively, and hence are representative of the largest number of contaminated samples received in a conventional clinical laboratory [6], as well as of invivo hyperglycaemia frequently found in critical patients. Despite being universally recognised that blood contamination from exogenous glucose solutions has a substantial impact on the quality and reliability of glucose and clinical chemistry testing [6], these results suggest that the concept should be broadened to embrace routine haemostasis testing. As the prekallikrein/kallikrein system also has many other functional effects, including non-haemostasis based [25], one could hypothesise additional effects on laboratorybased tests resulting from hyperglycaemia-based activation. In particular, additional studies should be planned to establish whether this interference may also affect thromboelastography and thromboelastometry [26]. In fact, the reagents used for activating blood coagulation are the same as for aPTT (INTEM activates the contact phase of haemostasis using ellagic acid), and PT (EXTEM activates haemostasis via the physiological activator tissue factor), whereas FIBTEM is an EXTEM-based assay for assessing the contribution of fibrinogen to clot formation [27]. It is hence predictable that spurious hyperglycaemia may generate a type of interference in thrombo elastography/ thrombo-elastometry that is similar to that observed in plasma based clotting tests. Therefore, whenever glucose contamination of a haemostasis sample is © 2014 The Association of Anaesthetists of Great Britain and Ireland
Anaesthesia 2015, 70, 549–554
suspected or clearly identified, the specimen should not be analysed since the data would be unreliable and may affect clinical decision-making, thus ultimately jeopardising patient safety. As such, we suggest that current recommendations should be strengthened and that blood samples should never be taken from the same limb where intravenous glucose is administered. Another clinical implication of this study is the direct effect of acute hyperglycaemia on aPTT. The fact that an enhanced glucose concentration in blood may trigger per se a shortening of the clotting time of this test in vitro provides a further contribution to the debate as to whether a shortened aPTT may be a cause or a consequence of diabetes [19]. The acute effect of hyperglycaemia on aPTT that we have observed in our investigation, along with the evidence that prekallikrein-mediated activation of the contact phase occurs in the hyperglycaemic state [24], seems to support the hypothesis that acute hyperglycaemia may indeed activate the contact phase of blood coagulation, and hence provides another reasonable explanation in support of the epidemiological observation that aPTT is reduced in diabetics. Finally, intuitively translating these results into clinical practice, it is conceivable that the bias observed in our experimental conditions may also be expected in vivo. Accordingly, a substantial enhancement of blood glucose level may be regarded as a potential source of variability in haemostasis testing.
Competing interests No external funding and no competing interests declared.
References
lova D, et al. Preanalytical 1. Lippi G, Becan-McBride K, Behu quality improvement: in quality we trust. Clinical Chemistry and Laboratory Medicine 2013; 51: 229–41. 2. O’Hare D, Chilvers RJ. Arterial blood sampling practices in intensive care units in England and Wales. Anaesthesia 2001; 56: 568–71. 3. Leslie RA, Gouldson S, Habib N, et al. Management of arterial lines and blood sampling in intensive care: a threat to patient safety. Anaesthesia 2013; 68: 1114–9. 4. Gupta KJ, Cook TM. Accidental hypoglycaemia caused by an arterial flush drug error: a case report and contributory causes analysis. Anaesthesia 2013; 68: 1179–87. 5. Thirugnanam M, French J. Accidental hypoglycaemia caused by an arterial flush drug error. Anaesthesia 2014; 69: 524–5. 553
Anaesthesia 2015, 70, 549–554
Lippi et al. | In-vitro hyperglycaemia and routine haemostasis testing
6. Lippi G, Avanzini R, Sandei F, Aloe R, Cervellin G. Blood sample contamination by glucose-containing solutions: effects and identification. British Journal of Biomedical Sciences 2013; 70: 180–3. 7. Clinical Laboratory Standards Institute. Procedures for Collection of Diagnostic Blood Specimens by Venipuncture; Approved Guideline – 6th Ed. CLSI document H3-A6. Wayne, PA: Clinical and Laboratory Standards Institute, 2007. 8. Read DC, Viera H, Arkin C. Effect of drawing blood specimens proximal to an in-place but discontinued intravenous solution. Can blood be drawn above the site of a shut-off i.v.? American Journal of Clinical Pathology 1988; 90: 702–6. 9. Association of Anaesthetists of Great Britain and Ireland. Arterial line blood sampling: preventing hypoglycaemic brain injury 2014. Anaesthesia 2014; 69: 380–5. 10. Adcock DM, Hoefner DM, Kottke-Marchant K, Marlar RA, Szamosi DI, Warunek DJ. Collection, Transport, and Processing of Blood Specimens for Testing Plasma-based Coagulation Assays and Molecular Hemostasis Assays: Approved Guideline-fifth Edition. CLSI document H21-A5. Wayne, PA: Clinical Laboratory Standards Institute, 2008. 11. Milos M, Herak D, Kuric L, Horvat I, Zadro R. Evaluation and performance characteristics of the coagulation system: ACL TOP analyzer – HemosIL reagents. International Journal of Laboratory Hematology 2009; 31: 26–35. 12. Ricos C, Alvarez V, Cava F, Garcia-Lario JV, et al. Current databases on biologic variation: pros, cons and progress. Scandinavian Journal of Clinical and Laboratory Investigation 1999; 59: 491–500. 13. Plebani M, Favaloro EJ, Lippi G. Patient safety and quality in laboratory and hemostasis testing: a renewed loop? Seminars in Thrombosis and Hemostasis 2012; 38: 553–8. 14. Vander AJ, Sherman JH, Luciano DS. Human Physiology – The Mechanisms of Body Function, 8th edn. New York: McGraw Hill, 2001: 115–41. ISBN: 0071183825. 15. Gallagher PG. Disorders of red cell volume regulation. Current Opinion in Hematology 2013; 20: 201–7. 16. Lippi G, Franchini M, Targher G, et al. Epidemiological association between fasting plasma glucose and shortened APTT. Clinical Biochemistry 2009; 42: 118–20.
554
17. Zhao Y, Zhang J, Zhang J, Wu J. Diabetes mellitus is associated with shortened activated partial thromboplastin time and increased fibrinogen values. PLoS One 2011; 6: e16470. 18. Sapkota B, Shrestha SK, Poudel S. Association of activated partial thromboplastin time and fibrinogen level in patients with type II diabetes mellitus. BMC Research Notes 2013; 6: 485. 19. Lippi G, Salvagno GL, Ippolito L, Franchini M, Favaloro EJ. Shortened activated partial thromboplastin time: causes and management. Blood Coagulation and Fibrinolysis 2010; 21: 459–63. 20. Girolami A, Vidal J, Salagh M, et al. The old and the new in prekallikrein deficiency: historical context and a family from Argentina with PK deficiency due to a new mutation (Arg541Gln) in exon 14 associated with a common polymorphysm (Asn124Ser) in exon 5. Seminars in Thrombosis and Hemostasis 2014; 40: 592–9. 21. Liu J, Gao BB, Clermont AC, et al. Hyperglycemia-induced cerebral hematoma expansion is mediated by plasma kallikrein. Nature Medicine 2011; 17: 206–10. 22. Patrassi GM, Vettor R, Padovan D, Girolami A. Contact phase of blood coagulation in diabetes mellitus. European Journal of Clinical Investigation 1982; 12: 307–11. 23. Federspil G, Vettor R, De Palo E, Padovan D, Sicolo N, Scandellari C. Plasma kallikrein activity in human diabetes mellitus. Metabolism 1983; 32: 540–2. 24. Feener EP, Zhou Q, Fickweiler W. Role of plasma kallikrein in diabetes and metabolism. Thrombosis and Haemostasis 2013; 110: 434–41. 25. de Maat S, de Groot PG, Maas C. Contact system activation on endothelial cells. Seminars in Thrombosis and Hemostasis 2014; 40: 887–94. 26. Quarterman C, Shaw M, Johnson I, Agarwal S. Intra- and intercentre standardisation of thromboelastography (TEGâ). Anaesthesia 2014; 69: 883–90. €rnberg J, Asmis LM, Seifert B, Spahn DR. 27. Theusinger OM, Nu Rotation thromboelastometry (ROTEM) stability and reproducibility over time. European Journal of Cardiothoracic Surgery 2010; 37: 677–83.
© 2014 The Association of Anaesthetists of Great Britain and Ireland
