N-acetylcysteine (NAC) is the treatment of choice to prevent hepatotoxicity after paracetamol intoxication^1–3 which is the most common cause of acute liver failure in Denmark, the UK, and the USA.^4–6 It has also been used to treat other critically ill patients such as those with non-paracetamol induced acute liver failure, sepsis, or acute respiratory distress syndrome, but the therapeutic effect of NAC in these illnesses are equivocal.^7–9 The rationale of using NAC in critically ill patients is its antioxidant function and its modulating effect on the proinflammatory cytokine response. It has a direct antioxidant effect and it stimulates synthesis of glutathione, an important endogenous intracellular antioxidant.^10,11

Acute liver failure is associated with severe derangement of haemostatic function as the liver is the main site of synthesis of coagulation factors and anticoagulant proteins.^12,13 Plasma levels of the vitamin K dependent coagulation factors II, VII, and X decrease at an early stage of hepatocellular damage. Therefore, it is widely accepted that the reduction in coagulation factor II+VII+X is a prognostic marker for the severity of liver damage.^14–16

Intravenous infusion of therapeutic doses of NAC to healthy subjects was reported to depress the plasma activity of coagulation factor II+VII+X as well as the activities of the single factors II, VII, and X.¹⁷ NAC also decreased the plasma activity of factor II+VII +X in vitro. This effect of NAC may be misinterpreted as a sign of liver failure and complicate the management of patients with paracetamol poisoning.¹⁸ The aim of the present study was to investigate in more detail the effect of NAC on haemostatic plasma parameters after intravenous infusion in healthy subjects. The in vivo data were in all instances compared with those obtained in vitro in order to address the mechanism of action of NAC. Our study expands previously cited work¹⁷ by including data on the effect of NAC on factors V, VIII, and IX, von Willebrand factor (vWf), and the anticoagulant proteins antithrombin, protein C, and free protein S. Approximately 30% of protein S in human plasma is free while the remainder is bound to the complement regulatory protein C4 binding protein.¹⁹ Only free protein S has anticoagulant function.

METHODS

Subjects

Ten healthy subjects (six males and four females) participated in the study. Their age and weight (median (range)) were 29 (21–32) years and 68 (52–93) kg. Before the start of the study, all subjects were judged as healthy by interview, physical examination, and routine laboratory investigations. Three additional normal subjects donated blood for the in vivo study (males aged 25–30 years). Written consent was obtained from each subject participating in the study and the study protocol was in accordance with the ethical guidelines of the 1975 Declaration of Helsinki and was approved by the regional ethics committee (protocol No KF 01-177/00) and the Danish Medicines Agency.

Analytical techniques

Blood for determination of haemostatic parameters was stabilised by 0.1 vol 0.129 mmol/l trisodium citrate. Plasma was separated and stored at −80°C (for assay of factor V and VIII) or −20°C (for assay of the other parameters) until analysed. If not otherwise stated, haemostatic parameters were analysed on ACL Futura or ACL 3000 (both Instrumentation Laboratories, Milan, Italy). Activated partial thromboplastin time (APTT) was analysed by Platelin Plus (Organon Teknika, Durham, North Carolina, USA). The activity of coagulation factor II+VII+X was analysed by Nycotest PT (Nycomed, Oslo, Norway) and expressed as a fraction of normal coagulation activity (arbitrary U/l), which was derived from prothrombin time measurements according to the manufacturer’s instruction. The activities of coagulation factors II, V, VII, and X were analysed by the one stage clotting assay using Dade Thromboplastin-IS and Dade Coagulation Factor Deficient Plasma II, V, VII, and X (Dade Behring, Marburg, Germany). The activities of coagulation factors VIII and IX were analysed by the one stage clotting assay using Automated APTT (Organon Teknika) and Dade Coagulation Factor VIII and IX Deficient Plasma (Dade Behring). The activities of antithrombin and protein C, and the reactivity of free protein S, were analysed by Coamatic Antithrombin, Coamatic Protein C (both Chromogenix-Instrumentation Laboratory, Milan, Italy) and IL Test Free Protein S (Instrumentation Laboratories), respectively.

The antigen concentration of vWf was measured by an enzyme linked immunosorbent assay using rabbit immunoglobulin against human vWf as the catching antibody and peroxidase conjugated rabbit immunoglobulin against human vWf as the detecting antibody (code No. p 0082 and p 0226; Dako, Glostrup, Denmark). The interassay coefficient of variations at normal levels of the haemostatic parameters were 3.2% (APTT), 5.1% (factor II+VII+X activity), 4.7% (factor II activity), 6.0% (factor V activity), 7.3% (factor VII activity), 6.6% (factor VIII activity), 4.0% (factor IX activity), 7.3% (factor X activity), 4.9% (antithrombin activity), 2.9% (protein C activity), 6.3% (free protein S reactivity), and 6.3% (vWf antigen).

Haemoglobin and platelet count in EDTA stabilised blood were analysed on an Advia 120 Hematology System (Bayer Diagnostics, Tarrytown, New York, USA). Alanine aminotransferase, alkaline phosphatase, creatinine, and lactate dehydrogenase in serum were analysed on a Vitros 950 Chemistry Analyzer (Ortho-Clinical Diagnostics, Rochester, New York, USA). Albumin and C reactive protein (CRP) in trisodium citrate stabilised plasma were analysed by Alb/BCP (bromcresol purple) and Tina-quant a CRP, respectively, using Modular Analytics-P (Roche Diagnostics, Mannheim, Germany). NAC up to 10 mmol/l did not influence assay of albumin or CRP when examined as described below in the in vitro protocol for haemostatic parameters.

In vivo protocol

Subjects were given NAC 400 mg/kg bodyweight (N-acetylcysteine 200 mg/ml; SAD, Copenhagen, Denmark) intravenously over 36 hours, distributed as a 150 mg/kg bolus over 15 min, 50 mg/kg over four hours, and 200 mg/kg over 32 hours. This corresponds to the regimen recommended by the Danish Association for the Study of the Liver and the Danish National Board of Health for the treatment of paracetamol poisoning. Blood samples were collected by separate vein punctures before and at 1, 3, 6, 8, 12, 16, 24, 32, 38, 48, 72, and 120 hours after start of the NAC infusion.

In vitro protocol

The influence of NAC on haemostatic parameters was examined after incubation of 0.8 vol trisodium citrate stabilised plasma for one hour at 37°C with 0.2 vol NAC in citrate-Tris buffer (1 vol trisodium citrate 129 mmol/l+9 vol Tris HCl 100 mmol/l, NaCl 50 mmol/l, pH 7.40 (37°C), adjusted with NaOH), yielding final concentrations of the drug ranging from 0.02 to 10 mmol/l.

Data treatment and statistical analysis

Haemostatic parameters were normalised to the same scale and presented as fractions of baseline values obtained before the start of the NAC infusion (in vivo study) or values obtained without NAC (in vitro study). Statistical analysis was performed using GraphPad Prism version 4.00 for Windows (GraphPad Software, San Diego, California, USA; www.graphpad.com). A p value of <0.05 was considered significant. Two groups were analysed by the Student’s t test and three or more groups were analysed by one way or two way ANOVA of data before normalisation to their baseline values. Descriptive normalised data are given as mean (95% confidence interval (CI)) and in the in vivo study n = 10 at each time point, if not otherwise indicated. In the case of factor VII, one subject was excluded due to preanalytical or analytical mistakes in the samples obtained at baseline and at other time points.

RESULTS

In vivo study

Basic laboratory investigation and clinical course

Haemoglobin and platelet count were measured at baseline and at 120 hours after the start of the NAC infusion, while albumin, creatinine, alanine aminotransferase, alkaline phosphatase, and lactate dehydrogenase were measured at one and three hours (albumin) and at 12, 32, and 120 hours (all components) after the start of infusion of the drug. All values were within the normal range (not shown). Normalised mean albumin concentration ranged between 0.92 and 1.06 of baseline values at 1, 3, 12, 32, and 120 hours. These deviations from baseline may reflect plasma volume changes. Correction for albumin concentrations did not alter the pattern of changes observed with the haemostatic parameters during NAC infusion. For this reason only uncorrected data for haemostatic parameters are presented. The haemostatic parameters at baseline are shown in table 1. They did not deviate significantly from the 95% reference interval of our laboratory.

Five of the 10 subjects developed moderate adverse reactions characterised by both cutaneous and systemic manifestations in relation to the initial bolus infusion (table 2). Three of these subjects received antihistamines. None of the subjects had any signs of adverse reactions after the first hour and all received NAC as planned.

Effect of NAC on global coagulation tests

Infusion of NAC depressed factor II+VII+X activity to 0.73 (0.69–0.76) of baseline values three hours after start of the infusion, and activity remained at this level until 38 hours (two hours after cessation of the infusion) (fig 1). Activity gradually reached baseline levels. NAC infusion had no overall significant effect on APTT (p = 0.93).

Effect of NAC on vitamin K dependent single coagulation factors II, VII, IX, and X

The activities of the single factors II, VII, IX, and X decreased abruptly at one hour after the start of the NAC infusion to 0.73 (0.67–0.79), 0.66 (0.58–0.73), 0.81 (0.73–0.90), and 0.64 (0.57–0.70) of their baseline values, respectively (figs 1, 2). In all instances this was followed by an increase in their activity level. Factor II and IX activities returned to baseline by six hours after the start of the NAC infusion while factor VII activity gradually returned to baseline at 38 hours. Factor X activity increased from one to six hours. Thereafter, the mean level was sustained at 0.7–0.8 of baseline values until 72 hours after the start of the infusion before it reached baseline at 120 hours.

Effect of NAC on anticoagulant proteins

NAC also depressed levels of the vitamin K dependent factors protein C and protein S (fig 2). Protein C activity and free protein S reactivity decreased abruptly at one hour to 0.74 (0.65–0.82) and 0.61 (0.54–0.67) of baseline values, respectively. Thereafter, protein C activity gradually increased until it reached baseline values at 38–48 hours after the start of the NAC infusion. The reactivity of free protein S increased to 0.88 (0.85–0.92) of baseline values at six hours, and this concentration was sustained (range of means 0.76–0.91) until the end of the observation period.

The overall effect of NAC on antithrombin activity was not significant when analysed at 1, 3, 12, and 32 hours (n = 10) and at 120 hours (n = 8) after the start of the infusion (p = 0.22) (data not shown).

Effect of NAC on coagulation factor V and VIII, and on vWf

NAC infusion induced concordant changes in levels of factor VIII activity and vWf antigen. The groups with and without moderate adverse reactions (n = 5 in each group) responded differently to NAC. Four of the subjects in the group with moderate adverse reactions showed a marked change in levels of factor VIII activity and vWf antigen within one hour, leading to an increase in the levels of the two parameters in this group to 1.85 (1.08–2.62) and 1.77 (0.83–2.71) of baseline values, respectively (tables 2, 3). Thereafter, levels slowly decreased towards baseline values (table 3). The overall effect of NAC on levels of factor VIII activity and vWf antigen in the group without adverse reactions was insignificant (p = 0.35 and 0.83, respectively).

The sensitive marker of the acute phase reaction, CRP, was measured before, and eight and 16 hours after the start of the NAC infusion in four subjects with a rapid marked response of factor VIII activity and vWf antigen to NAC. In all instances levels of CRP were ⩽3 mg/l (reference interval <10 mg/l).

The overall effect of NAC infusion on the level of factor V activity was not significant (p = 0.78) when analysed in all subjects at the same time points as in table 3 (not shown).

In vitro study

Incubation of plasma for one hour at 37°C in vitro with increasing concentrations of NAC resulted in a progressive decrease in levels of the vitamin K dependent proteins (fig 3). Concentrations of NAC that reduced levels of the respective proteins to 0.70 of their baseline values (comparable with the decreases in vivo) were approximately 3.5 (factor II+VII+X activity), 10 (factor II activity), 3.5 (factor VII activity), 5 (factor IX activity), 0.9 (factor X activity), 1.5 (protein C activity), and 0.3 mmol/l (free protein S reactivity). NAC from 0.5 to 10 mmol/l resulted in a progressive small decrease in coagulation factor V activity (mean (SEM)) to 0.81 (0.018) but it had no effect on factor VIII activity, vWf antigen, or antithrombin activity (not shown).

DISCUSSION

Apart from decreasing the activity levels of factors II, VII, and X,¹⁷ the present study shows that intravenous infusion of NAC to healthy subjects also depresses plasma levels of other vitamin K dependent proteins participating in the coagulation cascade (factor IX activity, protein C activity, and free protein S reactivity). The maximal effect of NAC on vitamin K dependent proteins occurred within one hour in our study and after several hours in a previously reported study,¹⁷ probably because we used a threefold higher bolus dose of NAC. Data on the effect of NAC on vitamin K dependent proteins in healthy subjects can most likely be extended to include patients treated with NAC, as a comparable decrease in factor II+VII+X activity as observed in the present study after initiation of NAC infusion was seen in patients with paracetamol poisoning but without any sign of hepatocellular damage.¹⁸

In agreement with the just cited study,¹⁷ we found that NAC infusion did not affect APTT, which is in agreement with the notion that APTT is rather insensitive to changes in activities of coagulation factors, contributing to the outcome of the test.²⁰

Depressed levels of vitamin K dependent proteins are most likely caused by modification of their structure/function by NAC. This is supported by the following evidence. (A) Concentrations of NAC affecting levels of the respective proteins in vivo and in vitro were comparable. The dose regimen of NAC used in the present study has been reported to give rise to a total plasma mean concentration of the drug of 3.4 mmol/l after the initial loading dose.²¹ Then, the mean concentration fell to a steady state level of 0.21 mmol/l after approximately 12 hours. After discontinuation of NAC it disappeared from plasma with a half life of approximately six hours. In vitro, we observed that NAC in the range 0.2–10 mmol/l induced a progressive decrease in levels of vitamin K dependent proteins. (B) In all instances the maximal fall in vitamin K dependent protein levels occurred within one hour after the initial loading dose of NAC, irrespective of their plasma half lives, which range from about six hours (factor VII) to approximately 70 hours (factor II).^22–24 (C) Factor II and IX activities reached their baseline values by six hours after the start of the NAC infusion, despite the fact that their half lives are approximately 24 and 70 hours, respectively. As these two factors are among the least sensitive to the effect of NAC in vitro, the relatively rapid normalisation of activity levels could be explained by a decline in the concentration of NAC after the initial loading dose, leading to dissociation of NAC from the two proteins and restoration of their normal function. (D) Depression of factor X activity and free protein S reactivity was sustained several hours after cessation of the NAC infusion. This sustained effect is in agreement with the observation that the activities of these two proteins are the most sensitive to the effect of NAC in vitro and therefore are also affected by the drug at the low concentrations achieved after the initial loading dose. The sustained effect of NAC on factor X activity is most likely responsible for the sustained effect of the drug on factor II+VII+X activity.

The fall in free protein S reactivity in response to NAC could be due to a modifying effect of the drug, leading to a decrease in the affinity of one or both of the targets for the protein S detecting reagents in the assay. The targets are the sex hormone binding globulin-like domain and the N terminal Gla domain, which are recognised by C4 binding protein and monoclonal antibody coated latex particles, respectively.²⁵

Vitamin K dependent proteins are homologous multidomain proteins that share a unique Gla domain of pivotal significance for their function,²⁶ indicating that NAC affects the structure/function of these proteins by a common mechanism. One mechanism could be reduction of exposed sensitive intramolecular disulphide bonds required for maintenance of their structure/function. By using thiol reactive agents, evidence has been provided that NAC can reversibly reduce disulphide bonds in the insulin receptor, matrix proteins, and intracellular signal transducers at concentrations similar to those impairing the vitamin K dependent proteins in the present study.^27–29 Another mechanism of impairment could be that the vitamin K dependent proteins analogous to other plasma proteins (primarily albumin) are nitrosylated,³⁰ that this nitrosylation enhances their activity, and that NAC induces denitrosylation resulting in a decrease in activity. This mechanism seems less likely as preliminary experiments in vitro showed that the nitrosylating agent DETA NONOate at 1–10 mmol/l had no effect on factor II or X activity or on free protein S reactivity in plasma incubated for one or two hours at 37°C (unpublished).

VWf serves as a stabilising carrier of factor VIII and any change in the plasma level of vWf is associated with a concordant change in the level of factor VIII.^31,32 In the present work, four of five subjects with moderate adverse reactions in response to NAC showed concordant and rapid marked increases in factor VIII activity and vWf antigen followed by progressive slow decreases in their levels. This is compatible with the immediate release of vWf from storage sites (Weibel-Palade bodies) in endothelial cells and the subsequent elimination of the factor VIII-vWf complex from plasma with the reported half life of approximately 14 hours.^23,32 Histamine may be the responsible agonist for the release of vWf as NAC liberates histamine in cell cultures^33,34 and histamine liberates vWf from its storage pools in vivo.³⁵ The proposed mechanism agrees with the observation that a controlled insect sting challenge resulted in a rapid and marked parallel increase in plasma levels of histamine and vWf in subjects who developed anaphylactic reactions and hypotension.³⁶ Interestingly, the one subject with systemic adverse reactions, but without an increase in levels of factor VIII activity and vWf antigen, was among the three subjects who received antihistamines. Both factor VIII and vWf are acute phase reactants.³⁷ It is unlikely that the NAC induced increases in the levels of these two factors in the four subjects with systemic adverse reactions were mediated by an acute phase reaction, as the level of the sensitive acute phase reactant CRP was unaffected. It is also unlikely that a gain in function mechanism was involved, as NAC did not affect plasma levels of factor VIII activity and vWf antigen in subjects without adverse reactions or in vitro.

The frequency of systemic adverse reactions in response to intravenous NAC infusion in patients with paracetamol intoxication varies from 3% to 48%.^38–40 This great variation in frequency could be caused by underestimation of adverse reactions in some studies as paracetamol can produce nausea and vomiting, making it difficult to distinguish these symptoms from those caused by NAC itself. Another explanation could be that adverse reactions may be underreported. In the present study, the frequency of systemic adverse reactions was 50%. The nature of these reactions and their early manifestation in relation to bolus infusion of NAC are in agreement with results obtained by others.^38–40 In a similar study to ours in healthy subjects, no systemic reactions occurred.¹⁷ This may be explained by the three times lower bolus dose used in this study.

There is no general consensus on when to stop NAC administration in patients being treated for paracetamol intoxication but the activity of coagulation factor II+VII+X remains essential in the monitoring of these patients.^15,41–46 It remains to be determined whether derangements of the haemostatic system observed here will be of future clinical relevance for decision making in the treatment of paracetamol intoxication. It is notable that NAC infusion has little effect on factor V activity and may increase factor VIII activity, as levels of both of these factors have been reported to be of prognostic value in fulminant hepatic failure.^47,48

The bioavailability of NAC after oral administration is much lower (approximately 9%) than after intravenous administration.⁴⁹ This means that the conclusions of the present study cannot be directly transferred to patients receiving NAC orally.

In conclusion, we have demonstrated that NAC infusion in doses used to treat paracetamol intoxication causes small to moderate derangements of the coagulation system, which affect both vitamin K dependent coagulant and anticoagulant proteins, some of which are used as early predictors of hepatic damage. We also found that adverse reactions can be associated with increases in the levels of factor VIII and its carrier protein vWf. Further studies are needed to elucidate the biochemical and pharmacokinetic bases for the observed changes in haemostatic parameters.