- Email: [email protected]
Effect of freeze-drying, freezing and frozen storage of blood plasma on fibrin network characteristics
Thrombosis Research 107 (2002) 263 – 269
Regular Article
Effect of freeze-drying, freezing and frozen storage of blood plasma on fibrin network characteristics Marlien Pieters a,*, Johann C. Jerling a, John W. Weisel b a
Potchefstroom Institute of Nutrition, Potchefstroom University for Christian Higher Education, Private Bag X6001, Potchefstroom 2520, South Africa b Department of Cell and Developmental Biology, University of Pennsylvania School of Medicine, Philadelphia, PA, USA Received 31 May 2002; received in revised form 24 October 2002; accepted 24 October 2002
Abstract Introduction: We investigated the effect of freezing, freeze-drying and the duration of frozen storage of blood plasma on fibrin network characteristics of clots subsequently produced. Materials and methods: Fibrin network characteristics of clots made from freeze-dried and frozen plasma were compared to those made from fresh plasma. Freeze-dried pooled plasma was reconstituted and frozen each month for 4 months to describe the differences in fibrin networks that occur as a result of storage of the plasma over this period. Results: Compared to freezing, freeze-drying of plasma had fewer undesirable effects on the fibrin network characteristics measured. Only the permeability of the clots from freeze-dried plasma was significantly less compared to the values of clots from the fresh plasma ( p = 0.005). Fibrinogen activity and mass – length ratio, compaction and fibrin content of the clots made from frozen plasma were, however, all significantly affected by freezing. Mass – length ratio and compaction showed a linear decrease and fibrin content a linear increase over a 4-month frozen storage period, thereby indicating that these variables were probably not stable. Large variation found in the data from each month indicates that there may be other factors, apart from storage time, that have a larger influence on these fibrin network characteristics, than frozen storage of plasma for 4 months. Storage of plasma in the freeze-dried form for 4 months resulted in a significant increase in fibrinogen ( p = 0.0004) but significant decrease in fibrin content ( p = 0.0002). Conclusions: Although the process of freeze-drying had fewer undesirable effects on the measured fibrin network characteristics compared to freezing, storage in both forms resulted in altered activity upon rehydration and thawing. D 2002 Elsevier Science Ltd. All rights reserved. Keywords: Stability; Fibrin networks; Frozen storage; Freeze-drying
1. Introduction During intervention and epidemiological studies, blood sampling is done over a period of time. It is preferred that all analyses are done in one batch to minimise between-day analytical variation, since the same batch of reagents, controls and calibration material can be used throughout. Samples are usually stored frozen for varying lengths of time, risking alterations in plasma protein concentration and activity [1,2]. Samples that have been frozen for different lengths of time (as is the current situation in many laboratories), might therefore not be comparable. Several studies
Abbreviations: CD, critical difference; CVA, analytical variation; CVI, within-subject variation; CVP, preanalytical variation. * Corresponding author. Tel.: +31-71-518-1218; fax: +31-71-5181904. E-mail address: [email protected] (M. Pieters).
have been done measuring the stability of coagulation proteins during frozen storage [3– 6], but the effect on fibrin network characteristics (permeability, mass – length ratio of fibres, compaction and fibrin content) has not been described. The characterisation of fibrin networks is still a developing research area. Although several studies have been done using some techniques to examine clot properties, there are still several unanswered fundamental questions, for instance, the effect of freezing and freeze-drying on clot structure. Another challenge in the study of fibrin networks is that many of the measurements have large coefficients of variation, arising either from the methodologies currently used or from sample differences. Several studies acknowledge these large variations by reporting them [7 – 10], while in other studies the coefficients of variation are not even reported [11,12]. In none of these studies, however, has the coefficient of variation been considered or incorporated into the interpretation of the results. One of the aims of this study was indeed to report
0049-3848/02/$ - see front matter D 2002 Elsevier Science Ltd. All rights reserved. PII: S 0 0 4 9 - 3 8 4 8 ( 0 2 ) 0 0 3 4 4 - 4
264
M. Pieters et al. / Thrombosis Research 107 (2002) 263–269
the coefficients of variation to establish the usefulness and limitations of these methodologies. The aims of this study were therefore to determine:
the between-day variation of the techniques used, offering possible solutions for distinguishing between analytical variation and treatment effects in the interpretation of the results; how the processes of freeze-drying and freezing of plasma affect fibrin network characteristics (permeability, mass – length ratio of fibres, compaction, fibrin content and fibrinogen) compared to those of fresh samples; the effect of different durations (1 –4 months) of frozen storage ( 83 jC) of plasma on these fibrin network characteristics; and the effect of storage of freeze-dried plasma on these fibrin network characteristics.
2. Materials and methods 2.1. Subjects Thirty employees of the Potchefstroom University in South Africa were randomly selected (after signing informed consent) to provide one large pooled plasma sample.
2.2. Blood sampling Blood was drawn, with minimal stasis, between 07:00 and 10:00. For the determination of the fibrin network characteristics, citrated (3.8%) blood with added aprotinin (Trasylol 35 Al/10 ml) was drawn and centrifuged twice at 3660 g for 15 min to provide platelet poor plasma. For the determination of fibrinogen, citrated blood was used and centrifuged to provide plasma. The two types of plasma were immediately pooled into two separate pools and then distributed in 2 ml aliquots into 5 ml containers. The plasma samples were then freeze-dried, using standardised procedures, except for small volumes necessary to provide 12-h fresh and 12-h frozen ( 83 jC) samples to be analysed with the freeze-dried samples. 2.3. Study design This experiment consisted of three sets of analyses of which Analyses 1 and 2 are depicted in the study design (Fig. 1). For each variable and each treatment, 10 samples were analysed and compared. For the determination of the effect of frozen storage on fibrin network characteristics, all the samples had to be analysed on the same day to exclude between-day analytical variation. The freeze-dried plasma therefore had to be reconstituted and frozen on different days to ensure different
Fig. 1. Study design.
M. Pieters et al. / Thrombosis Research 107 (2002) 263–269
lengths of frozen storage. It was also imperative that the blood samples were drawn on the same day, and pooled, to exclude between-subject variation. As a possible solution for this problem of drawing the blood samples in 1 day and doing all the analyses in 1 day, but having different durations of frozen storage in between, the plasma was freeze-dried after collection and then reconstituted and frozen on different days to ensure different durations of frozen storage. We decided on this specific storage period of no longer than 4 months, since this is the maximum period of time samples are currently being stored in our laboratory. After distributing the plasma samples into 2 ml aliquots, they were frozen at 83 jC and then freeze-dried in a Dura-Dry IIk with Dura-Portk Manifold System. Samples were put under vacuum and sealed under sterile conditions before they were stored at 4 jC until reconstitution and clot preparation. In Analysis 1, clots prepared from 12-h freeze-dried samples were compared with clots prepared from 12-h fresh plasma (4 jC) and 12-h frozen ( 83 jC) plasma. To determine the effect of the duration of frozen storage, clots prepared from plasma stored frozen for 1– 4 months were then analysed on another day (Analysis 2) and compared. Before analysis and clot preparation, plasma samples were thawed in a water bath at 25 jC. To determine the effect of storage time of freeze-dried plasma on fibrin network characteristics, clots prepared from reconstituted freeze-dried plasma analysed in Analysis 1 were compared to those analysed in Analysis 2. For the determination of fibrin network characteristics, there are, however, no standardised controls available that can be analysed with every run, to control for between-day analytical variation. To solve this problem, a third set of experiments were conducted to determine the between-day variation. In this phase, freeze-dried samples were reconstituted on five consecutive days to provide five sets of experiments for the determination of between-day analytical variation. 2.4. Analytical methods Measurements done on the fibrin networks included: fibrin content of the formed clots using the method of Ratnoff and Menzie [13]—this method measures the protein content of a washed clot. Clots are left overnight for complete polymerisation, they are then washed and centrifuged several times to get rid of all other proteins before they are hydrolysed. Folin and Ciocalteus’ reagent is then added to measure the protein content (i.e. fibrin) of the lysate; permeability was done in a special polystyrene cuvette provided with a silk net attached to a percolation device at the upper aspect of the gel and an outlet at the bottom, the cuvettes were sealed with a synthetic rubber stopper, essentially according to the design described previously [11], the Darcy constant (KS) was then calculated [7]; compaction was measured using a modified method of
265
Dhall et al. [14]—plasma clots were prepared in microcentrifuge tubes, centrifuged and the expelled supernatant was measured and expressed as percentage of the total volume. Mass –length ratio was determined turbidimetrically using the method described by Nair et al. [8], turbidity was recorded continuously using unclotted citrated plasma in the reference cell at wavelengths of 600– 800 nm with 2nm intervals using a Shimadzu UV 2100 spectrophotometer (Shimadzu, Kyoto, Japan); and fibrinogen was measured using the Clauss method modified by Instrumentation Laboratories for use on the Automated Coagulation Laboratory (ACL) analyser. This study was approved by the Ethics Committee of the Potchefstroom University. 2.5. Statistics Since data were not normally distributed and could not be normalised by logarithmic transformation, data are presented as medians with 25th and 75th percentiles. Outliers have been excluded from the data set since pooled plasma was used and abnormal values would therefore probably be due to some analytical error. Including outliers would therefore not reflect the true nature of the results. Mann – Whitney U-tests with Bonferroni adjustments were done to determine statistical significance ( p < 0.05) between fresh, frozen and freeze-dried samples (Analysis 1). To assess the degree of linearity of the relationship between frozen storage and the fibrin network characteristics (Analysis 2), Spearman rank order correlations were done. Mann –Whitney U-tests were also used to compare the freeze-dried samples from Analysis 1 with those of Analysis 2. 2.6. Critical difference (CD) To take into consideration the relatively high betweenday variation of the analysis, the critical differences of the five variables were calculated. This concept can be applied to determine whether the significant differences observed between two analyses done on separate days were due to: (a) between-day variation or (b) the effect of the treatment [15,16]. If the difference (as percentage change) between the two values, measured on separate days, was statistically significant but smaller than the critical difference, then the difference was only due to analytical variation. If the percentage change was larger than the critical difference, the difference can be ascribed to the effect of the specific treatment and not to analytical variation. CD is calculated as CD ¼
pffiffiffi 1 2 Z ðCV2A þ CV2I þ CV2P Þ 2
where CVA represents between-run(-day) analytical variation (which was calculated using data from the third set of experiments); CVI represents biological within-subject variation (which was neglected since a pooled sample was used); CVP represents preanalytical variation (which was also neglected since samples were collected and prepared
266
M. Pieters et al. / Thrombosis Research 107 (2002) 263–269
only under strictly standardised and controlled conditions) [15]. For a 5% significance concentration in any direction, Z = 1.96 [16]. This concept was applied for the interpretation of the results when variables measured in Analysis 1 were compared to variables measured in Analysis 2.
3. Results 3.1. Analysis 1: effect of freezing and freeze-drying on fibrin network characteristics There were no significant differences between the clots prepared from fresh and freeze-dried plasma for mass – length ratio, compaction and fibrin content. The mass – length ratio of the fibres and compaction of the clots prepared from the frozen samples, however, were significantly higher than that of the clots prepared from the fresh samples (mass– length ratio: 49.44 vs. 30.2 dalton/cm 10 12, p = 0.001; compaction: 45.5% vs. 37%, p = 0.001). The fibrin content values of the clots prepared from frozen plasma, however, were significantly lower than the values obtained from clots prepared from fresh plasma (2.03 vs. 2.87 g/l, p = 0.002). Medians, 25th and 75th percentiles, and significant differences are given in Table 1. The permeability of the clots prepared from freeze-dried plasma was significantly lower (4.73 vs. 7.35 cm2 10 9, p = 0.005) than the permeability of the clots prepared from fresh plasma (Table 1). The fibrinogen concentration of the frozen plasma was significantly higher than that of the fresh plasma (3.11 vs. 3.02 g/l, p = 0.04). 3.2. Effect of storage in the freeze-dried form Table 2 presents the between-run(-day) analytical variation (CVA) as well as the critical difference values for all the variables. Table 3 presents the median and 25th and 75th percentile values of the freeze-dried samples measured during Analysis 1 and Analysis 2, together with the percentage difference between the two. There were significant differences between the permeability and fibrin content of the clots prepared from freeze-
Table 2 Between-day variation coefficients and the critical difference values Variable
Between-day variation coefficient (CVA) (%)
Critical difference
Mass – length ratio (dalton/cm 10 12) Permeability (cm2 10 9) Compaction (%) Fibrinogen (Clauss method) (g/l) Fibrin content (g/l)
14.45
48.37
14.73
40.83
6.51 2.28
18.04 6.32
8.65
23.98
dried plasma, stored for 12 h (Analysis 1) and freeze-dried plasma that was stored for 4 months (Analysis 2). There were also significant differences between the fibrinogen concentration of these two sets of plasma. The percentage change for permeability was, however, smaller than the critical difference ( 23% vs. 40.83), indicating that the difference was probably not due to storage time, but only due to between-day analytical variation. For fibrinogen and fibrin content, the percentage change was, however, larger than the critical difference ( + 9.8% vs. 6.32 for fibrinogen and 30% vs. 23.98 for fibrin content) indicating that the duration of storage of plasma in the freeze-dried form does have an effect on these two variables and that the change was not only caused by analytical variation. 3.3. Analysis 2: effect of frozen storage on fibrin network characteristics Fig. 2 presents the data over the 4-month period and also shows the R- and p-values derived from the Spearman correlation. According to the p-values derived from the Spearman correlation, the mass – length ratio of the fibres and compaction showed a significant time-dependent decrease over the 4-month period (mass – length ratio: p < 0.001 and compaction: p = 0.011) while there was a statistically significant positive correlation between storage time and fibrin content ( p = 0.049). Frozen storage over the 4-month period had no significant effect on permeability or fibrinogen values. The highest R-value was 0.66 (for
Table 1 Comparison of freeze-dried and frozen samples with fresh samples Variable
Mass – length ratio (dalton/cm 10 12) Permeability (cm2 10 9) Compaction (%) Fibrinogen (Clauss method) (g/l) Fibrin content (g/l)
Fresh
Frozen
Freeze-dried
Median
25th and 75th percentiles
Median
25th and 75th percentiles
Median
25th and 75th percentiles
32.8a
30.2 – 35.5
49.4a,b
45.1 – 52.8
27.45b
22.29 – 34.39
7.35c
6.94 – 7.63
6.94d
6.65 – 7.48
37.0e 3.02g
33.5 – 42 2.96 – 3.11
48.25e,f 3.11g,h
45.5 – 49.5 3.06 – 3.18
37.75f 3.06h
36.0 – 41.5 2.99 – 3.13
2.87i
2.85 – 2.9
2.16i,j
2.03 – 2.49
2.58j
2.4 – 2.77
4.73c,d
Medians with the same letter (a, b, c, d, e, f, g, h, i, j) differ significantly ( p < 0.05). n = 10 for each variable and treatment.
4.27 – 5.26
M. Pieters et al. / Thrombosis Research 107 (2002) 263–269
267
Table 3 Effect of storage of plasma in freeze-dried form on fibrin network characteristics Variable
Mass – length ratio (dalton/cm 10 12) Permeability (cm2 10 9) Compaction (%) Fibrinogen (Clauss method) (g/l) Fibrin content (g/l)
Analysis 1
Analysis 2
Median
25th and 75th percentiles
Median
25th and 75th percentiles
27.45
22.29 – 34.39
35.18
29.61 – 39.37
4.73*
4.27 – 5.26
3.64*
3.53 – 3.93
37.75 3.06*
36 – 41.5 2.99 – 3.13
37 3.36*
36.5 – 38.5 3.21 – 3.39
2.58*
2.40 – 2.77
1.96*
1.87 – 2.01
Percentage difference (%) + 28.16 23a 1.99 + 9.8b 30b
n = 10 for each variable and treatment. * Significant difference between Analyses 1 and 2 ( p < 0.05). a Percentage change smaller than critical difference—change due to analytical variation. b Percentage change larger than critical difference—change due to effect of storage time.
mass – length ratio). As seen in Fig. 2, the data show relatively large variations for each specific month, which might explain the relatively low R-values. Only mass – length ratio, compaction and fibrin content showed any significant trend to be associated with the duration of frozen storage.
4. Discussion Many clinicians use frozen plasma for identification of disease states or other characterisation of patient health. However, there is very little information available on the effects of freezing on plasma samples. Until
Fig. 2. Correlation between fibrin network characteristics and duration of frozen storage of plasma.
268
M. Pieters et al. / Thrombosis Research 107 (2002) 263–269
now, most studies have examined this question using only standard haematological assays. The techniques for the analysis of the fibrin network characteristics examined in the present study provide new insight on the effect of these processes on the structures of fibrin clots. 4.1. Effect of freezing and freeze-drying on fibrin network characteristics When plasma is freeze-dried, it undergoes both freezing and dehydration. Both processes have been shown to cause the unfolding of proteins and consequent aggregation, which leads to possible loss of activity upon rehydration [1,2]. In this study there was, however, no difference between the mass –length ratio of the fibres, compaction and fibrin content values of the clots made from freeze-dried and fresh plasma, nor was there any difference between the fibrinogen concentration of the freeze-dried and fresh samples. Only the permeability of the clots prepared from freeze-dried plasma was significantly lower that that of the fresh samples. Redissolving of freeze-dried samples can lead to increased protein – protein interactions. If there are more fibrinogen – fibrinogen interactions even before cleavage of the fibrinopeptides, there will initially be more small oligomers. This could lead to more protofibrils and more fibres, thus yielding smaller pores and therefore lower permeability [17]. The mass –length ratio and compaction of the clots made from frozen plasma were all significantly higher than the values for clots made from fresh plasma, while fibrin content values were significantly lower. These results are mostly internally consistent. Thicker fibres are consistent with lower fibrin content and fewer branch points, both of which possibly imply lower clottability, resulting in higher compaction because of the fewer branch points [8]. From these results it appears as though freezing alone had a more undesirable effect on the fibrin network characteristics than the freeze-drying process, since most of the fibrin network characteristics of the clots prepared from frozen plasma differed from those prepared from fresh plasma. This result might be ascribed to the fact that freeze-dried plasma does not undergo the thawing process that frozen plasma does. It has been shown that the thawing process itself can cause further aggregation [18] and destruction of protein integrity [19]. Further research is encouraged to help explain these results. SDS-PAGE could be used to investigate the structure and composition of the fibrinogen-rich fraction of the plasma. Fibrinogen could also be determined using immunoassays to see whether there is any actual change in the fibrinogen concentration, since both the Clauss method and the method of Ratnoff and Menzie [13] are based on fibrinogen activity instead of concentration.
4.2. Effect of storage in the freeze-dried form If the concept of the critical difference is interpreted correctly, it enables researchers to distinguish between differences due to analytical variation and true differences. Mass –length ratio of the fibres and compaction were not affected by storage of plasma in the freeze-dried form for up to 4 months. The permeability of the samples decreased significantly after being stored for 4 months, but since the percentage change was smaller than the critical difference value, this difference is due to between-day analytical variation and not to changes that occurred during the storage period. This presents a problem since no standards and normal controls are available to adjust the results obtained in different batches according to the same reference value (the control). By using the critical difference concept, it is at least possible, when two samples are analysed on different days, in different batches, to calculate whether statistical differences are due to the effect of the specific treatment or only due to analytical variation. Fibrin content of the clot and plasma fibrinogen were the only two variables that were significantly affected by the storage period (their percentage changes were larger than their critical difference values). Another group, however, found that fibrinogen (also determined by the Clauss method) was stable for up to 59 months when stored in the freeze-dried form [20]. It is, however, known that changes in protein structure as well as aggregation with other molecules in the plasma, which occur with freezedrying, may influence the activity of the protein upon rehydration. Further research, investigating these structural changes, is therefore needed to help explain and identify the mechanisms behind these changes that occur during the frozen storage. 4.3. Effect of frozen storage The plasma samples measured in this section of the experiment were frozen for different durations (1 – 4 months) before clots were prepared for analysis (Analysis 2). There was a significant downward trend in compaction and mass – length ratio of the fibres in clots made from samples stored over the 4-month period (Fig. 2). The lower amount of compaction is consistent with the presence of thinner fibres, which leads to increased branching and stiffer clots. Fibrin content, on the other hand, showed a significant increase, which is also consistent with thinner fibres [8]. This indicates that these variables were probably not stable even over the short period of 4 months. An increasing degree of protein – protein interactions on thawing of frozen samples after longer periods of time could possibly account for these observations [17]. No linear association could be found for fibrinogen and permeability over time. These two variables therefore showed no clear downward or upward trend after the 4-month period. Other researchers have also
M. Pieters et al. / Thrombosis Research 107 (2002) 263–269
found that fibrinogen was stable for periods of longer than 1 year [4,6,20]. When considering the relatively low R-values derived from the correlation, it should be noted that the values for each month also show large variation (Fig. 2). Taking this into consideration, a longer storage period is probably needed to give a better indication of what effect the duration of frozen storage ( 83 jC) has on fibrin network characteristics. It is clear from the spread of the data that there are other factors causing these variations that might obscure the effect of frozen storage for a short period (1 –4 months). Due to the time-consuming nature of these analyses, we were limited by the number of samples that could be analysed in 1 day (Analysis 2) and therefore only 10 repetitions were done for each subset. Although the process of freeze-drying had fewer undesirable effects on the measured fibrin network characteristics compared to freezing, storage in both forms resulted in altered activity upon rehydration and thawing. Plasma samples of individuals should therefore probably be stored for the same period of time, before analysis, to ensure accurate, comparative results. Further research to identify the mechanisms behind these changes is therefore advised.
Acknowledgements The authors would like to thank Mr. Peter Rigsby from the NIBSC, Potters Bar, UK, for his input in the statistical analyses. Special thanks to nursing Sister Chrissie Lessing for her professional handling of all the subjects.
References [1] Vemuri S, Yu C-D, Roosdorp N. Effect of cryoprotectants on freezing, lyophilization, and storage of lyophilized recombinant alpha1antitrypsin formulations. PDA J Pharm Sci Technol 1994;48:241 – 6. [2] Allison SD, Randolph TW, Manning MC, Middleton K, Davis A, Carpenter JF. Effects of drying methods and additives on structure and function of actin: mechanisms of dehydration-induced damage and its inhibition. Arch Biochem Biophys 1998;358:171 – 81. [3] Iversen LH, Thorlacius-Ussing O. Short-time stability of markers of coagulation and fibrinolysis in frozen plasma. Thromb Res 1996;81: 253 – 61.
269
[4] McNerlan SE, Crawford VLS, Stout RW. Measurement of fibrinogen in frozen plasma. Thromb Res 1997;88:481 – 4. [5] Grau E, Tenias JM, Angeles Olaso M, Ferrando I, Juan MT, Pastor E, et al. Monitoring oral anticoagulant treatment from plasma stored for up to 48 hours and frozen plasma. Haematologica 1999;84:633 – 6. [6] Woodhams B, Girardot O, Blanco M-J, Colesse G, Gourmelin Y. Stability of coagulation proteins in frozen plasma. Blood Coagul Fibrinolysis 2001;12:229 – 36. [7] Fatah K, Hamsten A, Blomba¨ck B, Blomba¨ck M. Fibrin gel network characteristics and coronary heart disease: relations to plasma fibrinogen concentration, acute phase protein, serum lipoproteins and coronary atherosclerosis. Thromb Haemost 1992;68:130 – 5. [8] Nair CH, Azhar A, Dhall DP. Studies on fibrin network structure in human plasma: Part 1. Methods for clinical application. Thromb Res 1991;64:455 – 76. [9] Malan MM. Fibrin network characteristics of obese African women. PhD thesis, Potchefstroom: PU for CHE; 1999. 236 pp. [10] Nair CH, Shah GA, Dhall DP. Effect of temperature, pH and ionic strength and composition on fibrin network structure and its development. Thromb Res 1986;42:809 – 16. [11] Blomba¨ck B, Carlsson K, Fatah K, Hessel B, Procyk R. Fibrin in human plasma: gel architectures governed by rate and nature of fibrinogen activation. Thromb Res 1994;75:521 – 38. [12] Van Gelder JM, Nair CH, Dhall DP. Colloid determination of fibrin network permeability. Blood Coagul Fibrinolysis 1996;7:747 – 60. [13] Ratnoff OD, Menzie CAB. A new method for the determination of fibrinogen in small samples of plasma. J Lab Clin Med 1951;37: 316 – 20. [14] Dhall TZ, Bryce WAJ, Dhall DP. Effects of dextran on the molecular structure and tensile behaviour of human fibrin. Thromb Haemost 1976;35:737 – 45. [15] Fraser CG, Harris EK. Generation and application of data on biological variation in clinical chemistry. Crit Rev Clin Lab Sci 1989; 27:409 – 37. [16] Fraser CG, Hyltoft Peterson P, Larsen ML. Setting analytical goals for random analytical error in specific clinical monitoring situations. Clin Chem 1990;36:1625 – 8. [17] Weisel JW, Nagaswami C. Computer modelling of fibrin polymerization kinetics correlated with electron microscope and turbidity observations: clot structure and assembly are kinetically controlled. Biophys J 1992;63:111 – 28. [18] Kerwin BA, Heller MC, Levin SH, Randolph TW. Effects of Tween 80 and sucrose on acute short-term stability and long-term storage at 20 degrees C of a recombinant hemoglobin. J Pharm Sci 1998; 87:1062 – 8. [19] Kuznetsova IN. Stability of perfluorocarbon emulsions and their compatibility with blood serum. Artif Cells Blood Substit Immobil Biotechnol 1998;26:181 – 9. [20] Lewis MR, Callas PW, Jenny NS, Tracy RP. Longitudinal stability of coagulation, fibrinolysis, and inflammation factors in stored plasma samples. Thromb Haemost 2001;86:1495 – 500.
Regular Article
Effect of freeze-drying, freezing and frozen storage of blood plasma on fibrin network characteristics Marlien Pieters a,*, Johann C. Jerling a, John W. Weisel b a
Potchefstroom Institute of Nutrition, Potchefstroom University for Christian Higher Education, Private Bag X6001, Potchefstroom 2520, South Africa b Department of Cell and Developmental Biology, University of Pennsylvania School of Medicine, Philadelphia, PA, USA Received 31 May 2002; received in revised form 24 October 2002; accepted 24 October 2002
Abstract Introduction: We investigated the effect of freezing, freeze-drying and the duration of frozen storage of blood plasma on fibrin network characteristics of clots subsequently produced. Materials and methods: Fibrin network characteristics of clots made from freeze-dried and frozen plasma were compared to those made from fresh plasma. Freeze-dried pooled plasma was reconstituted and frozen each month for 4 months to describe the differences in fibrin networks that occur as a result of storage of the plasma over this period. Results: Compared to freezing, freeze-drying of plasma had fewer undesirable effects on the fibrin network characteristics measured. Only the permeability of the clots from freeze-dried plasma was significantly less compared to the values of clots from the fresh plasma ( p = 0.005). Fibrinogen activity and mass – length ratio, compaction and fibrin content of the clots made from frozen plasma were, however, all significantly affected by freezing. Mass – length ratio and compaction showed a linear decrease and fibrin content a linear increase over a 4-month frozen storage period, thereby indicating that these variables were probably not stable. Large variation found in the data from each month indicates that there may be other factors, apart from storage time, that have a larger influence on these fibrin network characteristics, than frozen storage of plasma for 4 months. Storage of plasma in the freeze-dried form for 4 months resulted in a significant increase in fibrinogen ( p = 0.0004) but significant decrease in fibrin content ( p = 0.0002). Conclusions: Although the process of freeze-drying had fewer undesirable effects on the measured fibrin network characteristics compared to freezing, storage in both forms resulted in altered activity upon rehydration and thawing. D 2002 Elsevier Science Ltd. All rights reserved. Keywords: Stability; Fibrin networks; Frozen storage; Freeze-drying
1. Introduction During intervention and epidemiological studies, blood sampling is done over a period of time. It is preferred that all analyses are done in one batch to minimise between-day analytical variation, since the same batch of reagents, controls and calibration material can be used throughout. Samples are usually stored frozen for varying lengths of time, risking alterations in plasma protein concentration and activity [1,2]. Samples that have been frozen for different lengths of time (as is the current situation in many laboratories), might therefore not be comparable. Several studies
Abbreviations: CD, critical difference; CVA, analytical variation; CVI, within-subject variation; CVP, preanalytical variation. * Corresponding author. Tel.: +31-71-518-1218; fax: +31-71-5181904. E-mail address: [email protected] (M. Pieters).
have been done measuring the stability of coagulation proteins during frozen storage [3– 6], but the effect on fibrin network characteristics (permeability, mass – length ratio of fibres, compaction and fibrin content) has not been described. The characterisation of fibrin networks is still a developing research area. Although several studies have been done using some techniques to examine clot properties, there are still several unanswered fundamental questions, for instance, the effect of freezing and freeze-drying on clot structure. Another challenge in the study of fibrin networks is that many of the measurements have large coefficients of variation, arising either from the methodologies currently used or from sample differences. Several studies acknowledge these large variations by reporting them [7 – 10], while in other studies the coefficients of variation are not even reported [11,12]. In none of these studies, however, has the coefficient of variation been considered or incorporated into the interpretation of the results. One of the aims of this study was indeed to report
0049-3848/02/$ - see front matter D 2002 Elsevier Science Ltd. All rights reserved. PII: S 0 0 4 9 - 3 8 4 8 ( 0 2 ) 0 0 3 4 4 - 4
264
M. Pieters et al. / Thrombosis Research 107 (2002) 263–269
the coefficients of variation to establish the usefulness and limitations of these methodologies. The aims of this study were therefore to determine:
the between-day variation of the techniques used, offering possible solutions for distinguishing between analytical variation and treatment effects in the interpretation of the results; how the processes of freeze-drying and freezing of plasma affect fibrin network characteristics (permeability, mass – length ratio of fibres, compaction, fibrin content and fibrinogen) compared to those of fresh samples; the effect of different durations (1 –4 months) of frozen storage ( 83 jC) of plasma on these fibrin network characteristics; and the effect of storage of freeze-dried plasma on these fibrin network characteristics.
2. Materials and methods 2.1. Subjects Thirty employees of the Potchefstroom University in South Africa were randomly selected (after signing informed consent) to provide one large pooled plasma sample.
2.2. Blood sampling Blood was drawn, with minimal stasis, between 07:00 and 10:00. For the determination of the fibrin network characteristics, citrated (3.8%) blood with added aprotinin (Trasylol 35 Al/10 ml) was drawn and centrifuged twice at 3660 g for 15 min to provide platelet poor plasma. For the determination of fibrinogen, citrated blood was used and centrifuged to provide plasma. The two types of plasma were immediately pooled into two separate pools and then distributed in 2 ml aliquots into 5 ml containers. The plasma samples were then freeze-dried, using standardised procedures, except for small volumes necessary to provide 12-h fresh and 12-h frozen ( 83 jC) samples to be analysed with the freeze-dried samples. 2.3. Study design This experiment consisted of three sets of analyses of which Analyses 1 and 2 are depicted in the study design (Fig. 1). For each variable and each treatment, 10 samples were analysed and compared. For the determination of the effect of frozen storage on fibrin network characteristics, all the samples had to be analysed on the same day to exclude between-day analytical variation. The freeze-dried plasma therefore had to be reconstituted and frozen on different days to ensure different
Fig. 1. Study design.
M. Pieters et al. / Thrombosis Research 107 (2002) 263–269
lengths of frozen storage. It was also imperative that the blood samples were drawn on the same day, and pooled, to exclude between-subject variation. As a possible solution for this problem of drawing the blood samples in 1 day and doing all the analyses in 1 day, but having different durations of frozen storage in between, the plasma was freeze-dried after collection and then reconstituted and frozen on different days to ensure different durations of frozen storage. We decided on this specific storage period of no longer than 4 months, since this is the maximum period of time samples are currently being stored in our laboratory. After distributing the plasma samples into 2 ml aliquots, they were frozen at 83 jC and then freeze-dried in a Dura-Dry IIk with Dura-Portk Manifold System. Samples were put under vacuum and sealed under sterile conditions before they were stored at 4 jC until reconstitution and clot preparation. In Analysis 1, clots prepared from 12-h freeze-dried samples were compared with clots prepared from 12-h fresh plasma (4 jC) and 12-h frozen ( 83 jC) plasma. To determine the effect of the duration of frozen storage, clots prepared from plasma stored frozen for 1– 4 months were then analysed on another day (Analysis 2) and compared. Before analysis and clot preparation, plasma samples were thawed in a water bath at 25 jC. To determine the effect of storage time of freeze-dried plasma on fibrin network characteristics, clots prepared from reconstituted freeze-dried plasma analysed in Analysis 1 were compared to those analysed in Analysis 2. For the determination of fibrin network characteristics, there are, however, no standardised controls available that can be analysed with every run, to control for between-day analytical variation. To solve this problem, a third set of experiments were conducted to determine the between-day variation. In this phase, freeze-dried samples were reconstituted on five consecutive days to provide five sets of experiments for the determination of between-day analytical variation. 2.4. Analytical methods Measurements done on the fibrin networks included: fibrin content of the formed clots using the method of Ratnoff and Menzie [13]—this method measures the protein content of a washed clot. Clots are left overnight for complete polymerisation, they are then washed and centrifuged several times to get rid of all other proteins before they are hydrolysed. Folin and Ciocalteus’ reagent is then added to measure the protein content (i.e. fibrin) of the lysate; permeability was done in a special polystyrene cuvette provided with a silk net attached to a percolation device at the upper aspect of the gel and an outlet at the bottom, the cuvettes were sealed with a synthetic rubber stopper, essentially according to the design described previously [11], the Darcy constant (KS) was then calculated [7]; compaction was measured using a modified method of
265
Dhall et al. [14]—plasma clots were prepared in microcentrifuge tubes, centrifuged and the expelled supernatant was measured and expressed as percentage of the total volume. Mass –length ratio was determined turbidimetrically using the method described by Nair et al. [8], turbidity was recorded continuously using unclotted citrated plasma in the reference cell at wavelengths of 600– 800 nm with 2nm intervals using a Shimadzu UV 2100 spectrophotometer (Shimadzu, Kyoto, Japan); and fibrinogen was measured using the Clauss method modified by Instrumentation Laboratories for use on the Automated Coagulation Laboratory (ACL) analyser. This study was approved by the Ethics Committee of the Potchefstroom University. 2.5. Statistics Since data were not normally distributed and could not be normalised by logarithmic transformation, data are presented as medians with 25th and 75th percentiles. Outliers have been excluded from the data set since pooled plasma was used and abnormal values would therefore probably be due to some analytical error. Including outliers would therefore not reflect the true nature of the results. Mann – Whitney U-tests with Bonferroni adjustments were done to determine statistical significance ( p < 0.05) between fresh, frozen and freeze-dried samples (Analysis 1). To assess the degree of linearity of the relationship between frozen storage and the fibrin network characteristics (Analysis 2), Spearman rank order correlations were done. Mann –Whitney U-tests were also used to compare the freeze-dried samples from Analysis 1 with those of Analysis 2. 2.6. Critical difference (CD) To take into consideration the relatively high betweenday variation of the analysis, the critical differences of the five variables were calculated. This concept can be applied to determine whether the significant differences observed between two analyses done on separate days were due to: (a) between-day variation or (b) the effect of the treatment [15,16]. If the difference (as percentage change) between the two values, measured on separate days, was statistically significant but smaller than the critical difference, then the difference was only due to analytical variation. If the percentage change was larger than the critical difference, the difference can be ascribed to the effect of the specific treatment and not to analytical variation. CD is calculated as CD ¼
pffiffiffi 1 2 Z ðCV2A þ CV2I þ CV2P Þ 2
where CVA represents between-run(-day) analytical variation (which was calculated using data from the third set of experiments); CVI represents biological within-subject variation (which was neglected since a pooled sample was used); CVP represents preanalytical variation (which was also neglected since samples were collected and prepared
266
M. Pieters et al. / Thrombosis Research 107 (2002) 263–269
only under strictly standardised and controlled conditions) [15]. For a 5% significance concentration in any direction, Z = 1.96 [16]. This concept was applied for the interpretation of the results when variables measured in Analysis 1 were compared to variables measured in Analysis 2.
3. Results 3.1. Analysis 1: effect of freezing and freeze-drying on fibrin network characteristics There were no significant differences between the clots prepared from fresh and freeze-dried plasma for mass – length ratio, compaction and fibrin content. The mass – length ratio of the fibres and compaction of the clots prepared from the frozen samples, however, were significantly higher than that of the clots prepared from the fresh samples (mass– length ratio: 49.44 vs. 30.2 dalton/cm 10 12, p = 0.001; compaction: 45.5% vs. 37%, p = 0.001). The fibrin content values of the clots prepared from frozen plasma, however, were significantly lower than the values obtained from clots prepared from fresh plasma (2.03 vs. 2.87 g/l, p = 0.002). Medians, 25th and 75th percentiles, and significant differences are given in Table 1. The permeability of the clots prepared from freeze-dried plasma was significantly lower (4.73 vs. 7.35 cm2 10 9, p = 0.005) than the permeability of the clots prepared from fresh plasma (Table 1). The fibrinogen concentration of the frozen plasma was significantly higher than that of the fresh plasma (3.11 vs. 3.02 g/l, p = 0.04). 3.2. Effect of storage in the freeze-dried form Table 2 presents the between-run(-day) analytical variation (CVA) as well as the critical difference values for all the variables. Table 3 presents the median and 25th and 75th percentile values of the freeze-dried samples measured during Analysis 1 and Analysis 2, together with the percentage difference between the two. There were significant differences between the permeability and fibrin content of the clots prepared from freeze-
Table 2 Between-day variation coefficients and the critical difference values Variable
Between-day variation coefficient (CVA) (%)
Critical difference
Mass – length ratio (dalton/cm 10 12) Permeability (cm2 10 9) Compaction (%) Fibrinogen (Clauss method) (g/l) Fibrin content (g/l)
14.45
48.37
14.73
40.83
6.51 2.28
18.04 6.32
8.65
23.98
dried plasma, stored for 12 h (Analysis 1) and freeze-dried plasma that was stored for 4 months (Analysis 2). There were also significant differences between the fibrinogen concentration of these two sets of plasma. The percentage change for permeability was, however, smaller than the critical difference ( 23% vs. 40.83), indicating that the difference was probably not due to storage time, but only due to between-day analytical variation. For fibrinogen and fibrin content, the percentage change was, however, larger than the critical difference ( + 9.8% vs. 6.32 for fibrinogen and 30% vs. 23.98 for fibrin content) indicating that the duration of storage of plasma in the freeze-dried form does have an effect on these two variables and that the change was not only caused by analytical variation. 3.3. Analysis 2: effect of frozen storage on fibrin network characteristics Fig. 2 presents the data over the 4-month period and also shows the R- and p-values derived from the Spearman correlation. According to the p-values derived from the Spearman correlation, the mass – length ratio of the fibres and compaction showed a significant time-dependent decrease over the 4-month period (mass – length ratio: p < 0.001 and compaction: p = 0.011) while there was a statistically significant positive correlation between storage time and fibrin content ( p = 0.049). Frozen storage over the 4-month period had no significant effect on permeability or fibrinogen values. The highest R-value was 0.66 (for
Table 1 Comparison of freeze-dried and frozen samples with fresh samples Variable
Mass – length ratio (dalton/cm 10 12) Permeability (cm2 10 9) Compaction (%) Fibrinogen (Clauss method) (g/l) Fibrin content (g/l)
Fresh
Frozen
Freeze-dried
Median
25th and 75th percentiles
Median
25th and 75th percentiles
Median
25th and 75th percentiles
32.8a
30.2 – 35.5
49.4a,b
45.1 – 52.8
27.45b
22.29 – 34.39
7.35c
6.94 – 7.63
6.94d
6.65 – 7.48
37.0e 3.02g
33.5 – 42 2.96 – 3.11
48.25e,f 3.11g,h
45.5 – 49.5 3.06 – 3.18
37.75f 3.06h
36.0 – 41.5 2.99 – 3.13
2.87i
2.85 – 2.9
2.16i,j
2.03 – 2.49
2.58j
2.4 – 2.77
4.73c,d
Medians with the same letter (a, b, c, d, e, f, g, h, i, j) differ significantly ( p < 0.05). n = 10 for each variable and treatment.
4.27 – 5.26
M. Pieters et al. / Thrombosis Research 107 (2002) 263–269
267
Table 3 Effect of storage of plasma in freeze-dried form on fibrin network characteristics Variable
Mass – length ratio (dalton/cm 10 12) Permeability (cm2 10 9) Compaction (%) Fibrinogen (Clauss method) (g/l) Fibrin content (g/l)
Analysis 1
Analysis 2
Median
25th and 75th percentiles
Median
25th and 75th percentiles
27.45
22.29 – 34.39
35.18
29.61 – 39.37
4.73*
4.27 – 5.26
3.64*
3.53 – 3.93
37.75 3.06*
36 – 41.5 2.99 – 3.13
37 3.36*
36.5 – 38.5 3.21 – 3.39
2.58*
2.40 – 2.77
1.96*
1.87 – 2.01
Percentage difference (%) + 28.16 23a 1.99 + 9.8b 30b
n = 10 for each variable and treatment. * Significant difference between Analyses 1 and 2 ( p < 0.05). a Percentage change smaller than critical difference—change due to analytical variation. b Percentage change larger than critical difference—change due to effect of storage time.
mass – length ratio). As seen in Fig. 2, the data show relatively large variations for each specific month, which might explain the relatively low R-values. Only mass – length ratio, compaction and fibrin content showed any significant trend to be associated with the duration of frozen storage.
4. Discussion Many clinicians use frozen plasma for identification of disease states or other characterisation of patient health. However, there is very little information available on the effects of freezing on plasma samples. Until
Fig. 2. Correlation between fibrin network characteristics and duration of frozen storage of plasma.
268
M. Pieters et al. / Thrombosis Research 107 (2002) 263–269
now, most studies have examined this question using only standard haematological assays. The techniques for the analysis of the fibrin network characteristics examined in the present study provide new insight on the effect of these processes on the structures of fibrin clots. 4.1. Effect of freezing and freeze-drying on fibrin network characteristics When plasma is freeze-dried, it undergoes both freezing and dehydration. Both processes have been shown to cause the unfolding of proteins and consequent aggregation, which leads to possible loss of activity upon rehydration [1,2]. In this study there was, however, no difference between the mass –length ratio of the fibres, compaction and fibrin content values of the clots made from freeze-dried and fresh plasma, nor was there any difference between the fibrinogen concentration of the freeze-dried and fresh samples. Only the permeability of the clots prepared from freeze-dried plasma was significantly lower that that of the fresh samples. Redissolving of freeze-dried samples can lead to increased protein – protein interactions. If there are more fibrinogen – fibrinogen interactions even before cleavage of the fibrinopeptides, there will initially be more small oligomers. This could lead to more protofibrils and more fibres, thus yielding smaller pores and therefore lower permeability [17]. The mass –length ratio and compaction of the clots made from frozen plasma were all significantly higher than the values for clots made from fresh plasma, while fibrin content values were significantly lower. These results are mostly internally consistent. Thicker fibres are consistent with lower fibrin content and fewer branch points, both of which possibly imply lower clottability, resulting in higher compaction because of the fewer branch points [8]. From these results it appears as though freezing alone had a more undesirable effect on the fibrin network characteristics than the freeze-drying process, since most of the fibrin network characteristics of the clots prepared from frozen plasma differed from those prepared from fresh plasma. This result might be ascribed to the fact that freeze-dried plasma does not undergo the thawing process that frozen plasma does. It has been shown that the thawing process itself can cause further aggregation [18] and destruction of protein integrity [19]. Further research is encouraged to help explain these results. SDS-PAGE could be used to investigate the structure and composition of the fibrinogen-rich fraction of the plasma. Fibrinogen could also be determined using immunoassays to see whether there is any actual change in the fibrinogen concentration, since both the Clauss method and the method of Ratnoff and Menzie [13] are based on fibrinogen activity instead of concentration.
4.2. Effect of storage in the freeze-dried form If the concept of the critical difference is interpreted correctly, it enables researchers to distinguish between differences due to analytical variation and true differences. Mass –length ratio of the fibres and compaction were not affected by storage of plasma in the freeze-dried form for up to 4 months. The permeability of the samples decreased significantly after being stored for 4 months, but since the percentage change was smaller than the critical difference value, this difference is due to between-day analytical variation and not to changes that occurred during the storage period. This presents a problem since no standards and normal controls are available to adjust the results obtained in different batches according to the same reference value (the control). By using the critical difference concept, it is at least possible, when two samples are analysed on different days, in different batches, to calculate whether statistical differences are due to the effect of the specific treatment or only due to analytical variation. Fibrin content of the clot and plasma fibrinogen were the only two variables that were significantly affected by the storage period (their percentage changes were larger than their critical difference values). Another group, however, found that fibrinogen (also determined by the Clauss method) was stable for up to 59 months when stored in the freeze-dried form [20]. It is, however, known that changes in protein structure as well as aggregation with other molecules in the plasma, which occur with freezedrying, may influence the activity of the protein upon rehydration. Further research, investigating these structural changes, is therefore needed to help explain and identify the mechanisms behind these changes that occur during the frozen storage. 4.3. Effect of frozen storage The plasma samples measured in this section of the experiment were frozen for different durations (1 – 4 months) before clots were prepared for analysis (Analysis 2). There was a significant downward trend in compaction and mass – length ratio of the fibres in clots made from samples stored over the 4-month period (Fig. 2). The lower amount of compaction is consistent with the presence of thinner fibres, which leads to increased branching and stiffer clots. Fibrin content, on the other hand, showed a significant increase, which is also consistent with thinner fibres [8]. This indicates that these variables were probably not stable even over the short period of 4 months. An increasing degree of protein – protein interactions on thawing of frozen samples after longer periods of time could possibly account for these observations [17]. No linear association could be found for fibrinogen and permeability over time. These two variables therefore showed no clear downward or upward trend after the 4-month period. Other researchers have also
M. Pieters et al. / Thrombosis Research 107 (2002) 263–269
found that fibrinogen was stable for periods of longer than 1 year [4,6,20]. When considering the relatively low R-values derived from the correlation, it should be noted that the values for each month also show large variation (Fig. 2). Taking this into consideration, a longer storage period is probably needed to give a better indication of what effect the duration of frozen storage ( 83 jC) has on fibrin network characteristics. It is clear from the spread of the data that there are other factors causing these variations that might obscure the effect of frozen storage for a short period (1 –4 months). Due to the time-consuming nature of these analyses, we were limited by the number of samples that could be analysed in 1 day (Analysis 2) and therefore only 10 repetitions were done for each subset. Although the process of freeze-drying had fewer undesirable effects on the measured fibrin network characteristics compared to freezing, storage in both forms resulted in altered activity upon rehydration and thawing. Plasma samples of individuals should therefore probably be stored for the same period of time, before analysis, to ensure accurate, comparative results. Further research to identify the mechanisms behind these changes is therefore advised.
Acknowledgements The authors would like to thank Mr. Peter Rigsby from the NIBSC, Potters Bar, UK, for his input in the statistical analyses. Special thanks to nursing Sister Chrissie Lessing for her professional handling of all the subjects.
References [1] Vemuri S, Yu C-D, Roosdorp N. Effect of cryoprotectants on freezing, lyophilization, and storage of lyophilized recombinant alpha1antitrypsin formulations. PDA J Pharm Sci Technol 1994;48:241 – 6. [2] Allison SD, Randolph TW, Manning MC, Middleton K, Davis A, Carpenter JF. Effects of drying methods and additives on structure and function of actin: mechanisms of dehydration-induced damage and its inhibition. Arch Biochem Biophys 1998;358:171 – 81. [3] Iversen LH, Thorlacius-Ussing O. Short-time stability of markers of coagulation and fibrinolysis in frozen plasma. Thromb Res 1996;81: 253 – 61.
269
[4] McNerlan SE, Crawford VLS, Stout RW. Measurement of fibrinogen in frozen plasma. Thromb Res 1997;88:481 – 4. [5] Grau E, Tenias JM, Angeles Olaso M, Ferrando I, Juan MT, Pastor E, et al. Monitoring oral anticoagulant treatment from plasma stored for up to 48 hours and frozen plasma. Haematologica 1999;84:633 – 6. [6] Woodhams B, Girardot O, Blanco M-J, Colesse G, Gourmelin Y. Stability of coagulation proteins in frozen plasma. Blood Coagul Fibrinolysis 2001;12:229 – 36. [7] Fatah K, Hamsten A, Blomba¨ck B, Blomba¨ck M. Fibrin gel network characteristics and coronary heart disease: relations to plasma fibrinogen concentration, acute phase protein, serum lipoproteins and coronary atherosclerosis. Thromb Haemost 1992;68:130 – 5. [8] Nair CH, Azhar A, Dhall DP. Studies on fibrin network structure in human plasma: Part 1. Methods for clinical application. Thromb Res 1991;64:455 – 76. [9] Malan MM. Fibrin network characteristics of obese African women. PhD thesis, Potchefstroom: PU for CHE; 1999. 236 pp. [10] Nair CH, Shah GA, Dhall DP. Effect of temperature, pH and ionic strength and composition on fibrin network structure and its development. Thromb Res 1986;42:809 – 16. [11] Blomba¨ck B, Carlsson K, Fatah K, Hessel B, Procyk R. Fibrin in human plasma: gel architectures governed by rate and nature of fibrinogen activation. Thromb Res 1994;75:521 – 38. [12] Van Gelder JM, Nair CH, Dhall DP. Colloid determination of fibrin network permeability. Blood Coagul Fibrinolysis 1996;7:747 – 60. [13] Ratnoff OD, Menzie CAB. A new method for the determination of fibrinogen in small samples of plasma. J Lab Clin Med 1951;37: 316 – 20. [14] Dhall TZ, Bryce WAJ, Dhall DP. Effects of dextran on the molecular structure and tensile behaviour of human fibrin. Thromb Haemost 1976;35:737 – 45. [15] Fraser CG, Harris EK. Generation and application of data on biological variation in clinical chemistry. Crit Rev Clin Lab Sci 1989; 27:409 – 37. [16] Fraser CG, Hyltoft Peterson P, Larsen ML. Setting analytical goals for random analytical error in specific clinical monitoring situations. Clin Chem 1990;36:1625 – 8. [17] Weisel JW, Nagaswami C. Computer modelling of fibrin polymerization kinetics correlated with electron microscope and turbidity observations: clot structure and assembly are kinetically controlled. Biophys J 1992;63:111 – 28. [18] Kerwin BA, Heller MC, Levin SH, Randolph TW. Effects of Tween 80 and sucrose on acute short-term stability and long-term storage at 20 degrees C of a recombinant hemoglobin. J Pharm Sci 1998; 87:1062 – 8. [19] Kuznetsova IN. Stability of perfluorocarbon emulsions and their compatibility with blood serum. Artif Cells Blood Substit Immobil Biotechnol 1998;26:181 – 9. [20] Lewis MR, Callas PW, Jenny NS, Tracy RP. Longitudinal stability of coagulation, fibrinolysis, and inflammation factors in stored plasma samples. Thromb Haemost 2001;86:1495 – 500.