Vox Sang. 35: 105-110 (1978)
Thrombogenicity of Factor 1X Concentrates: in vitro and in vivo (Rabbit) Studies J . D . Cash, Rosemary Owens, R . G . Dalton and R . 1.Prescott S-E Scotland Regional Blood Transfusion Service, Royal Infirmary; University Department of Surgery, and Medical Computing and Statistics Unit, Medical School, Edinburgh
Abstract. The reproducibility and correlation between the NAPTT, TGt,, in (vitro) tests and the Wessler (rabbit) stasis thrombus (in vivo) model have been studied using 10 different factor IX concentrates. The TGt,, test was more reproducible than the NAPTT and the overall reproducibility of the rabbit model was poor. The low reproducibility of the rabbit model appeared to be largely confined to those factor IX concentrates which showed a poor correlation between the NAPTT and TGt,, iesults. The TGt,, test emerged as the in vitro test which correlated most closely with the in vivo (rabbit) test. It is concluded that the NAPTT and TGt,, test are measuring different thrombogenic moieties in factor IX concentrates and that further studies are required to elucidate this phenomenon.
Introduction Mtnachk et al. [ll], in a report on the use of factor IX concentrates in patients with liver disease, first drew attention to the potential thrombogenicity of this type of blood product; a hazard that was not reemphasised until some 11 years later by Tullis and Breen [18]. Within the last 5 years, several well documented reports of thromboembolic phenomena, some fatal, following the administration of factor IX concentrates have been published [2,4,5, 8, 10, 12, 171. The problem has now been considered to be of sufficient importance to warrant the establishment of a specific investigative Task Force by the International Committee on Haemostasis and Thrombosis. Several groups have attempted to improve and develop methods by which individual batches of factor IX concentrates
could be screened in the laboratory for potential thrombogenicity, prior to issue for clinical use. Such an approach is of importance since the clinical reports of thrombogenicity have revealed not only batch differences, but also that the existing in vitro screening methods (the plasma recalcification time and fibrinogen clotting time) are less than satisfactory. Early studies, performed on dogs in this laboratory, indicated that the development of laboratory and histological evidence of intravascular coagulation might be one approach by which batches of factor IX concentrates could be tested for thrombogenicity [3]. These observations have since been confirmed [7], but the practicability of this method for routine batch screening is doubted. Further relevant studies were reported in 1975; these were designed to establish new in vitro assays for thrombogenicity. King-
106
don el al. [9] introduced a test in which factor IX concentrates were added to a nonactivated partial thromboplastin time system (NAPTT), whereas Sas et al. [16] examined the generation of thrombin after the recalcification of factor IX concentrates (TGt,, test). In both tests the shorter reaction times were assumed to indicate increasing thrombogenicity. Kingdon et al. [9] provided evidence to support this assumption; thus, on testing various factor IX concentrates they observed a good correlation between the degree of shortening of the NAPTT and the occurrence of stasis thrombi in the jugular veins of rabbits after the concentrates were infused into the ear veins. This in vivo test was based on that originally described by Wessler et al. [19]. Preliminary studies in this laboratory demonstrated that although there appeared to be some agreement between the NAPTT and TGt,, tests, major discrepancies were readily demonstrated with many factor IX preparations [151. The following communication describes a series of studies in which this problem has been further explored: 10 factor IX concentrates were selected (partly on the basis of their known discrepant in vitro results), their NAPTT and TGt,, times were rechecked and their in vivo thrombogenicity assessed using the Wessler (rabbit) technique. The primary purpose of this exercise was to examine whether this particular in vivo thrombogenicity model correlated more closely with the in vitro (NAPTT or TGt5,) results.
Materials and Methods In vitro Assays All concentrates were reconstituted according to the manufacturers’ directions as if they were to be administered to patients. The NAPTT test was
Cash/Owens/Dalton/Prescott
performed, as described by Kingdon et al. [9], using a 1/10, M O O and 1/1,000 dilution in Trisbuffered saline at pH 7.4 [6]. The specially prepared non-activated platelet-poor plasma (aliquots stored at -36OC), gave control times of 250-300 sec. All concentrates were tested using the same batch of platelet-poor plasma, Thrombofax and calcium chloride. The TGt,, test was performed according to the method of Sas et al. [15], with the addition that comparisons were made between different volumes of test concentrate (0.05, 0.1 and 0.2 ml). Biological Assay The i n vivo studies were made on New Zealand white rabbits weighing 2.0k0.3 kg. An identical ‘Wessler type’ test procedure to that described by Kingdon et al. [9] was used, with the exception that minimal thrombogenic doses were not calculated, but an arbitrary visual grading (1+ to 4t) made of clot formation in the ligated jugular vein, The factor IX concentrates selected for this study, none of which contained heparin, were obtained from several sources. All concentrates were examined in both in vitro and in vivo tests immediately (within 5 min) after reconstitution. Because of the subjective nature of the biological assay the in vitro and in vivo tests were undertaken by two independent teams and those conducting the in vivo studies were not at the time aware of the in vitro results. Statistical Methods Reproducibility of the tests had to be assessed in an indirect manner as there was no common scale of measurement for the various tests. However, as we had repeated measurements on a number of different specimens we were able to apply a one-way analysis of variance to the results of each test to identify two components of variation. Thus, we were able to estimate the variance corresponding to repeated observations on the same specimen (it was assumed that all specimens showed similar reproducibility), and we were also able to estimate the variance corresponding to differences between specimens. If the reproducibility of a particular test was good then the variance term corresponding to variation within the same specimen should have been small in relation to the between specimen variance. An example of the method is pro-
Thrombogenicity of Factor IX Concentrates
107
j
TableI. Ratio ofestimated betweensamplevariance to within sample variance for the estimation procedures performed Test
Quantity or dilution
TGt,,'
0.05 ml 0.10 ml
1/10
10.7
0.1 14.1
4.5 0.8 0.4
0.8 -
Biological assay
1/100 1/1,O00 -
-
1
58 109 21
1/1,O00
1/10
1/100 NAPTT blank
47 364 15 3.3 1.6 0.1
0.20 ml NAPTT
Ratio of between sample variance to within sample variation original logarithms measurements
6.0
2.0
4.9
One sample gave consistent readings in the TGt,, test of greater than 80 min and has been omitted from these calculations. The effect of inclusion of th!s sample, taking its value as 80 min is to increase the ratios still further.
vided by Armitage [l]. Taking the ratio of the estimate of the between specimen variance to the within specimen variance gave a dimensionless quantity which indicated the relative reproducibility of the various tests considered, with higher values indicating a better test. A detailed comparison of this ratio between different tests is dependent on the assumptions of results following a normal distribution, which may be difficult to justify. However, even if this condition is not fulfilled, the order of magnitude of the ratio gives an indication of the discriminating power of the test. For each of the variables studied we have calculated this ratio, both for the original measurement and for its logarithm. Examination of this ratio under both scales of measurement permitted unambiguous inferences to be drawn about the relative reproducibility of the various tests. Finally, the relationship between the TGt,, test (0.1 ml volume), the NAPTT (1110 dilution) test and the biological assay was examined, using standard methods of regression.
FACTOR IX CONCENTRATES
Fig. 1. Summary of a comparison of the biological, TGt,, and NAPTT tests of the 10 different factor IX concentrates studied.
Results An assessment of the reproducibility of each of the three techniques, when applied to several factor IX cncentrates, revealed that the most reproducible was the TGtro test using a test volume of 0.1 ml. The NAPTT was consistently much less reproducible and the optimum repeatability was achieved when using a dilution of 1/10; this could be further improved by expressing the results as the ratio of testkontrol. These conclusions held whether the analyses of variance were performed using the recorded values or their logarithms (table I). The reproducibility of both in vitro tests compared most favourably with that of the biological assay (fig. 1).
Cash/Owens/Dalton/Prescott
108
/ 0
0
P-53
-
.
I
zz
(1.25 0.3
0.4 05 0.6 NAPTTltestlcontmll(log xolel
08
1.0
Fig. 2. Correlation (r = 0.59) between the mean TGt,, (0.1 ml) and NAP= (1110 dilution) test results obtained from the 10 factor IX concentrates studied. 1.0
0.6
0.4
.
In order to define the correlation between the NAPTT and TGt,, techniques, the most reproducible expression of their individual test results was used, i.e. the NAPTT ratio at a 1/10 dilution and the TGt,, with a 0.1 ml volume. When linear regressions were derived the correlation coefficients were higher using logarithms and the relationship between the NAPTT and TGt,, is shown in figure 2. This analysis revealed that although there was a weak positive correlation between the two techniques, this was largely attributable to the fact that 2 of the 10 concentrates tested had both short NAPTT and TGt,, times. In effect, the NAPTT test showed little ability to discriminate between 8 of the 10 samples. Although we have used a regression line in figure 2 to draw attention to the association, this is by no means an adequate description of the presumably complex association between the two tests. Figure 3 shows the correlation between the biological assay and the NAP'IT and the TGt,, tests, respectively. The TGt,, test appeared to give the best agreement with the mean score obtained in the biological assay.
Discussion
Bdogical Assay1 Ahbunits)
Fig. 3. Correlation between the mean N A P m (1110 dilution) test and biological assay (r = -0.55), and TGt,, (0.1 ml) test and biological assay (r = -0.82) in the 10 factor IX concentrates studied.
The results of this study have confirmed other findings [15], that although some factor IX preparations show excellent correlation between the results of the NAPTT and TGt,, tests, there are outstanding exceptions. Thus, with the preparations selected for this study, it was not possible to differentiate concentrates B, C and D from H, I and J using the NAP", whereas they were consistently differentiated using the TGt5,. These differences could not be ex-
Thrombogenicity of Factor IX Concentrates
plained on the basis of the poorer reproducibility of the NAPTT technique and we believe this indicates that the two techniques are detecting different thrombogenic moieties. Similar conclusions have come from chromatographic separation studies, reported by Pepper et al. [14]. The data obtained from the rabbit studies suggested that when unequivocally activated or non-activated factor IX concentrates, as defined by both the NAPTT and TGt,, techniques, were administered, the effect was reasonably reproducible, and despite its subjective end-point this correlated well with the in vitro assays. However, reproducibility of the in vivo technique seemed to be particularly poor with those concentrates which had significant differences between their NAPTT andTGt,, results. The problem of reproducibility of the in vivo test with certain concentrates led to some doubt on the validity of comparing these results and those of the NAPTT and TGt,, tests. However, when this was done, using mean values for all measurements, the TGt,, appeared to correlate more closely with the in vivo assay. These findings would suggest that the original demonstration by Kingdon et al. [9] of a good correlation (r = 0.77) between the NAPTT and the in vivo assay results was due to the fact that the majority of the concentrates they tested were either significantly activated or outstandingly non-activated. Detailed examination of their results supports this conclusion. It is also of interest to note that two of the concentrates tested by Kingdon et al. [9], with an NAPTT of 182 and 149 sec (at 1/10 dilution), respectively, appeared to be more thrombogenic in vivo than several others with NAPTT values between 25 and 50 sec. One of these two concentrates had NAPTT values which
109
were indistinguishable from another, yet there was a 7.5 times difference in the minimum thrombogenic dose in rabbits between these concentrates. It is our impression that had we selected concentrates showing the particular in vitro characteristics as those studied by Kingdon et al. [9], then we would have come to similar conclusions. At the present time we would not conclude from our investigations that the TGt,, test is a more appropriate in vitro technique than the NAPTT for screening thrombogenic factor IX concentrates. We would, however, caution against the exclusive use of the NAPTT, despite recent attempts to enhance its reproducibility and standardisation [20]. Furthermore, we suggest that the TGt,, technique, which has now been greatly simplified using chromogenic substrates [14] should be included to assay preparations of factor IX concentrates. It should also be emphasised that there is little evidence at present to suggest that the rabbit (Wessler) model is necessarily one which is applicable to the clinical (human) situation. It has already emerged that there are major differences between the rabbit and dog models, for in addition to their localised and systemic natures, the thrombotic event in the rabbit occurs within seconds of the administration of factor IX concentrate, whereas in the dog model it is not evidenced for 2-4 h [3,7]. Moreover, it has been demonstrated that heparin in the factor IX concentrate inhibits thrombosis in the rabbit model [9], whereas in the dog it is without effect [3,7]. These differences are striking and may indicate that different thrombogenic moieties are operative in each animal model. We conclude that the mechanisms underlying the problem of thrombogenicity of fac-
110
tor IX concentrates remain incompletely understood and that the techniques for detecting potentially hazardous products continue to be inadequate. Further research, both in the laboratory and at the bedside, is urgently required.
Acknowledgements This project was supported by a grant from the Scottish Home and Health Department.
References Armitage, P.: Statistical methods in medical research, pp. 198-201 (Blackwell, Oxford 1971). Blatt, P. M.; Lundblad, R. L.; Kingdon, H. S.; McLean, G., and Roberts, H. R.: Thrombogenic materials in prothrombin complex concentrates. Ann. intern. Med. 81: 766-770 (1974). Cash, J.D.:; Dalton, R.G.; Middleton, S., and Smith, J. K.:Studies on the thrombogenicity of Scottish factor IX concentrates in dogs. Thromb. Diath. haemorrh. 33: 632639 (1975). Cederbaum. A. I. and Roberts, H. R.: Complications of the use of prothrombin complex concentrates in liver disease.Clin.Res.21: 92(1973). Davey, R. J.; Shashaty, G. G., and Rath, C. E.: Acute coagulopathy following infusion of prothrombin complex concentrates. Am. J. Med. 60: 719-722 (1976).. Gomori, G.: Buffers in the range of pH6.5 to 9.6. Proc. SOC.exp. Biol. Med. 62: 33-34 (1946). Hedner, U.;Nilsson, I. M., and Bergentz, S. E.: Various prothrombin complex concentrates and their effect on coagulation and fibrinolysis in vivo. Thromb. Haemost. 35: 386-394 (1976). Kasper, C. K.: Postoperative thrombosis in hemophilia B. New. Engl. J. Med. 289: 160 (1973). 9 Kingdon, II. S.; Lundblad, R. L.; Veltkamp, J. J., and Aronson, D. L.: Potentially thrombogenic materials in factor IX concentrates. Thromb. Diath. haemorrh. 33: 617-631 (1975). 10 Marchesi, S . L. and Burney, R.: Prothrombin complex concentrates and thrombosis. New Engl. J. Med. 290: 403-404 (1974).
Cash/Owens/Dal tonPrescott
11 MCnachC, D.; Fauvert, R. et Soulier, J. P.: Utilisation en hepatologie d’une fraction contenant prothrombin, le complex proconvertine-facteur Stuart et le facteur anti-hemophilique B (PPB). Path. Biol., Paris 7: 2515-2523 (1959). 12 MBnachC, D. and Guillin, M. C.: The use of factor IX concentrates for patients with conditions other than factor IX deficiency. Br. J. Haemat. 31: 247-250 (1975). 13 Pepper, D. S.; Banhegyi, D.; Howie, A., and Cash, J. D.: In vitro thrombogenicity tests of factor IX concentrates. Br. J. Haemat. 36: 555-565 (1977). 14 Pepper, D. S.; Banhegyi, D., and Howie, A.: Spectrophotometric determination of factor Xa generation in Factor IX concentrates. Thromb. Haemostat. 37: 535-540 (1977). 15 Prowse, C. V.; Pepper, D. S.; Cash, J. D., and Patterson, M.: Thrombogenicity screening of factor IX concentrates. Thromb. Diath. haemorrh. 38: 728 (1977). 16 Sas, G.; Owens, R. E.; Smith, J. K.; Middleton, S., and Cash, J. D.: In vitro spontaneous thrombin generation in human factor IX concentrates. Br. J. Haemat. 31: 25-35 (1975). 17 Schimpf, K.; Zimmerman, K., and Kompf, B.: DIC and postoperative wound bleeding under factor IX substitution therapy in a case of hemophilia B; successful treatment with heparin. Thromb. Res. 8: 65-70 (1976). 18 Tullis, J. L. and Breen, F.: Christmas factor concentrates; the clinical use of several preparations. Biblthca haemat., vol. 34, pp. 40-51 (Karger, Basel 1970). 19 Wessler, S.; Reimer, S. M., and Sheps, M. C.: Biological assay of a thrombosis-inducing activity in human serum. J. appl. Physiol. 14: 943-946 (I 959). 20 White, G. C.; Roberts, R. H.; Kingdon, H. S., and Lundblad, R. L.: Prothrombin complex concentrates. Potentially thrombogenic materials and clues to the mechanism of thrombosis in vivo. Blood 49: 159-170 (1977).
Received: August 31, 1977 Accepted: November 10, 1977 Dr. J. D. Cash, S-E Scotland Regional Blood Transfusion Service, Royal Infirmary, Edinburgh EH3 9HB (Scotland)
Thrombogenicity of Factor 1X Concentrates: in vitro and in vivo (Rabbit) Studies J . D . Cash, Rosemary Owens, R . G . Dalton and R . 1.Prescott S-E Scotland Regional Blood Transfusion Service, Royal Infirmary; University Department of Surgery, and Medical Computing and Statistics Unit, Medical School, Edinburgh
Abstract. The reproducibility and correlation between the NAPTT, TGt,, in (vitro) tests and the Wessler (rabbit) stasis thrombus (in vivo) model have been studied using 10 different factor IX concentrates. The TGt,, test was more reproducible than the NAPTT and the overall reproducibility of the rabbit model was poor. The low reproducibility of the rabbit model appeared to be largely confined to those factor IX concentrates which showed a poor correlation between the NAPTT and TGt,, iesults. The TGt,, test emerged as the in vitro test which correlated most closely with the in vivo (rabbit) test. It is concluded that the NAPTT and TGt,, test are measuring different thrombogenic moieties in factor IX concentrates and that further studies are required to elucidate this phenomenon.
Introduction Mtnachk et al. [ll], in a report on the use of factor IX concentrates in patients with liver disease, first drew attention to the potential thrombogenicity of this type of blood product; a hazard that was not reemphasised until some 11 years later by Tullis and Breen [18]. Within the last 5 years, several well documented reports of thromboembolic phenomena, some fatal, following the administration of factor IX concentrates have been published [2,4,5, 8, 10, 12, 171. The problem has now been considered to be of sufficient importance to warrant the establishment of a specific investigative Task Force by the International Committee on Haemostasis and Thrombosis. Several groups have attempted to improve and develop methods by which individual batches of factor IX concentrates
could be screened in the laboratory for potential thrombogenicity, prior to issue for clinical use. Such an approach is of importance since the clinical reports of thrombogenicity have revealed not only batch differences, but also that the existing in vitro screening methods (the plasma recalcification time and fibrinogen clotting time) are less than satisfactory. Early studies, performed on dogs in this laboratory, indicated that the development of laboratory and histological evidence of intravascular coagulation might be one approach by which batches of factor IX concentrates could be tested for thrombogenicity [3]. These observations have since been confirmed [7], but the practicability of this method for routine batch screening is doubted. Further relevant studies were reported in 1975; these were designed to establish new in vitro assays for thrombogenicity. King-
106
don el al. [9] introduced a test in which factor IX concentrates were added to a nonactivated partial thromboplastin time system (NAPTT), whereas Sas et al. [16] examined the generation of thrombin after the recalcification of factor IX concentrates (TGt,, test). In both tests the shorter reaction times were assumed to indicate increasing thrombogenicity. Kingdon et al. [9] provided evidence to support this assumption; thus, on testing various factor IX concentrates they observed a good correlation between the degree of shortening of the NAPTT and the occurrence of stasis thrombi in the jugular veins of rabbits after the concentrates were infused into the ear veins. This in vivo test was based on that originally described by Wessler et al. [19]. Preliminary studies in this laboratory demonstrated that although there appeared to be some agreement between the NAPTT and TGt,, tests, major discrepancies were readily demonstrated with many factor IX preparations [151. The following communication describes a series of studies in which this problem has been further explored: 10 factor IX concentrates were selected (partly on the basis of their known discrepant in vitro results), their NAPTT and TGt,, times were rechecked and their in vivo thrombogenicity assessed using the Wessler (rabbit) technique. The primary purpose of this exercise was to examine whether this particular in vivo thrombogenicity model correlated more closely with the in vitro (NAPTT or TGt5,) results.
Materials and Methods In vitro Assays All concentrates were reconstituted according to the manufacturers’ directions as if they were to be administered to patients. The NAPTT test was
Cash/Owens/Dalton/Prescott
performed, as described by Kingdon et al. [9], using a 1/10, M O O and 1/1,000 dilution in Trisbuffered saline at pH 7.4 [6]. The specially prepared non-activated platelet-poor plasma (aliquots stored at -36OC), gave control times of 250-300 sec. All concentrates were tested using the same batch of platelet-poor plasma, Thrombofax and calcium chloride. The TGt,, test was performed according to the method of Sas et al. [15], with the addition that comparisons were made between different volumes of test concentrate (0.05, 0.1 and 0.2 ml). Biological Assay The i n vivo studies were made on New Zealand white rabbits weighing 2.0k0.3 kg. An identical ‘Wessler type’ test procedure to that described by Kingdon et al. [9] was used, with the exception that minimal thrombogenic doses were not calculated, but an arbitrary visual grading (1+ to 4t) made of clot formation in the ligated jugular vein, The factor IX concentrates selected for this study, none of which contained heparin, were obtained from several sources. All concentrates were examined in both in vitro and in vivo tests immediately (within 5 min) after reconstitution. Because of the subjective nature of the biological assay the in vitro and in vivo tests were undertaken by two independent teams and those conducting the in vivo studies were not at the time aware of the in vitro results. Statistical Methods Reproducibility of the tests had to be assessed in an indirect manner as there was no common scale of measurement for the various tests. However, as we had repeated measurements on a number of different specimens we were able to apply a one-way analysis of variance to the results of each test to identify two components of variation. Thus, we were able to estimate the variance corresponding to repeated observations on the same specimen (it was assumed that all specimens showed similar reproducibility), and we were also able to estimate the variance corresponding to differences between specimens. If the reproducibility of a particular test was good then the variance term corresponding to variation within the same specimen should have been small in relation to the between specimen variance. An example of the method is pro-
Thrombogenicity of Factor IX Concentrates
107
j
TableI. Ratio ofestimated betweensamplevariance to within sample variance for the estimation procedures performed Test
Quantity or dilution
TGt,,'
0.05 ml 0.10 ml
1/10
10.7
0.1 14.1
4.5 0.8 0.4
0.8 -
Biological assay
1/100 1/1,O00 -
-
1
58 109 21
1/1,O00
1/10
1/100 NAPTT blank
47 364 15 3.3 1.6 0.1
0.20 ml NAPTT
Ratio of between sample variance to within sample variation original logarithms measurements
6.0
2.0
4.9
One sample gave consistent readings in the TGt,, test of greater than 80 min and has been omitted from these calculations. The effect of inclusion of th!s sample, taking its value as 80 min is to increase the ratios still further.
vided by Armitage [l]. Taking the ratio of the estimate of the between specimen variance to the within specimen variance gave a dimensionless quantity which indicated the relative reproducibility of the various tests considered, with higher values indicating a better test. A detailed comparison of this ratio between different tests is dependent on the assumptions of results following a normal distribution, which may be difficult to justify. However, even if this condition is not fulfilled, the order of magnitude of the ratio gives an indication of the discriminating power of the test. For each of the variables studied we have calculated this ratio, both for the original measurement and for its logarithm. Examination of this ratio under both scales of measurement permitted unambiguous inferences to be drawn about the relative reproducibility of the various tests. Finally, the relationship between the TGt,, test (0.1 ml volume), the NAPTT (1110 dilution) test and the biological assay was examined, using standard methods of regression.
FACTOR IX CONCENTRATES
Fig. 1. Summary of a comparison of the biological, TGt,, and NAPTT tests of the 10 different factor IX concentrates studied.
Results An assessment of the reproducibility of each of the three techniques, when applied to several factor IX cncentrates, revealed that the most reproducible was the TGtro test using a test volume of 0.1 ml. The NAPTT was consistently much less reproducible and the optimum repeatability was achieved when using a dilution of 1/10; this could be further improved by expressing the results as the ratio of testkontrol. These conclusions held whether the analyses of variance were performed using the recorded values or their logarithms (table I). The reproducibility of both in vitro tests compared most favourably with that of the biological assay (fig. 1).
Cash/Owens/Dalton/Prescott
108
/ 0
0
P-53
-
.
I
zz
(1.25 0.3
0.4 05 0.6 NAPTTltestlcontmll(log xolel
08
1.0
Fig. 2. Correlation (r = 0.59) between the mean TGt,, (0.1 ml) and NAP= (1110 dilution) test results obtained from the 10 factor IX concentrates studied. 1.0
0.6
0.4
.
In order to define the correlation between the NAPTT and TGt,, techniques, the most reproducible expression of their individual test results was used, i.e. the NAPTT ratio at a 1/10 dilution and the TGt,, with a 0.1 ml volume. When linear regressions were derived the correlation coefficients were higher using logarithms and the relationship between the NAPTT and TGt,, is shown in figure 2. This analysis revealed that although there was a weak positive correlation between the two techniques, this was largely attributable to the fact that 2 of the 10 concentrates tested had both short NAPTT and TGt,, times. In effect, the NAPTT test showed little ability to discriminate between 8 of the 10 samples. Although we have used a regression line in figure 2 to draw attention to the association, this is by no means an adequate description of the presumably complex association between the two tests. Figure 3 shows the correlation between the biological assay and the NAP'IT and the TGt,, tests, respectively. The TGt,, test appeared to give the best agreement with the mean score obtained in the biological assay.
Discussion
Bdogical Assay1 Ahbunits)
Fig. 3. Correlation between the mean N A P m (1110 dilution) test and biological assay (r = -0.55), and TGt,, (0.1 ml) test and biological assay (r = -0.82) in the 10 factor IX concentrates studied.
The results of this study have confirmed other findings [15], that although some factor IX preparations show excellent correlation between the results of the NAPTT and TGt,, tests, there are outstanding exceptions. Thus, with the preparations selected for this study, it was not possible to differentiate concentrates B, C and D from H, I and J using the NAP", whereas they were consistently differentiated using the TGt5,. These differences could not be ex-
Thrombogenicity of Factor IX Concentrates
plained on the basis of the poorer reproducibility of the NAPTT technique and we believe this indicates that the two techniques are detecting different thrombogenic moieties. Similar conclusions have come from chromatographic separation studies, reported by Pepper et al. [14]. The data obtained from the rabbit studies suggested that when unequivocally activated or non-activated factor IX concentrates, as defined by both the NAPTT and TGt,, techniques, were administered, the effect was reasonably reproducible, and despite its subjective end-point this correlated well with the in vitro assays. However, reproducibility of the in vivo technique seemed to be particularly poor with those concentrates which had significant differences between their NAPTT andTGt,, results. The problem of reproducibility of the in vivo test with certain concentrates led to some doubt on the validity of comparing these results and those of the NAPTT and TGt,, tests. However, when this was done, using mean values for all measurements, the TGt,, appeared to correlate more closely with the in vivo assay. These findings would suggest that the original demonstration by Kingdon et al. [9] of a good correlation (r = 0.77) between the NAPTT and the in vivo assay results was due to the fact that the majority of the concentrates they tested were either significantly activated or outstandingly non-activated. Detailed examination of their results supports this conclusion. It is also of interest to note that two of the concentrates tested by Kingdon et al. [9], with an NAPTT of 182 and 149 sec (at 1/10 dilution), respectively, appeared to be more thrombogenic in vivo than several others with NAPTT values between 25 and 50 sec. One of these two concentrates had NAPTT values which
109
were indistinguishable from another, yet there was a 7.5 times difference in the minimum thrombogenic dose in rabbits between these concentrates. It is our impression that had we selected concentrates showing the particular in vitro characteristics as those studied by Kingdon et al. [9], then we would have come to similar conclusions. At the present time we would not conclude from our investigations that the TGt,, test is a more appropriate in vitro technique than the NAPTT for screening thrombogenic factor IX concentrates. We would, however, caution against the exclusive use of the NAPTT, despite recent attempts to enhance its reproducibility and standardisation [20]. Furthermore, we suggest that the TGt,, technique, which has now been greatly simplified using chromogenic substrates [14] should be included to assay preparations of factor IX concentrates. It should also be emphasised that there is little evidence at present to suggest that the rabbit (Wessler) model is necessarily one which is applicable to the clinical (human) situation. It has already emerged that there are major differences between the rabbit and dog models, for in addition to their localised and systemic natures, the thrombotic event in the rabbit occurs within seconds of the administration of factor IX concentrate, whereas in the dog model it is not evidenced for 2-4 h [3,7]. Moreover, it has been demonstrated that heparin in the factor IX concentrate inhibits thrombosis in the rabbit model [9], whereas in the dog it is without effect [3,7]. These differences are striking and may indicate that different thrombogenic moieties are operative in each animal model. We conclude that the mechanisms underlying the problem of thrombogenicity of fac-
110
tor IX concentrates remain incompletely understood and that the techniques for detecting potentially hazardous products continue to be inadequate. Further research, both in the laboratory and at the bedside, is urgently required.
Acknowledgements This project was supported by a grant from the Scottish Home and Health Department.
References Armitage, P.: Statistical methods in medical research, pp. 198-201 (Blackwell, Oxford 1971). Blatt, P. M.; Lundblad, R. L.; Kingdon, H. S.; McLean, G., and Roberts, H. R.: Thrombogenic materials in prothrombin complex concentrates. Ann. intern. Med. 81: 766-770 (1974). Cash, J.D.:; Dalton, R.G.; Middleton, S., and Smith, J. K.:Studies on the thrombogenicity of Scottish factor IX concentrates in dogs. Thromb. Diath. haemorrh. 33: 632639 (1975). Cederbaum. A. I. and Roberts, H. R.: Complications of the use of prothrombin complex concentrates in liver disease.Clin.Res.21: 92(1973). Davey, R. J.; Shashaty, G. G., and Rath, C. E.: Acute coagulopathy following infusion of prothrombin complex concentrates. Am. J. Med. 60: 719-722 (1976).. Gomori, G.: Buffers in the range of pH6.5 to 9.6. Proc. SOC.exp. Biol. Med. 62: 33-34 (1946). Hedner, U.;Nilsson, I. M., and Bergentz, S. E.: Various prothrombin complex concentrates and their effect on coagulation and fibrinolysis in vivo. Thromb. Haemost. 35: 386-394 (1976). Kasper, C. K.: Postoperative thrombosis in hemophilia B. New. Engl. J. Med. 289: 160 (1973). 9 Kingdon, II. S.; Lundblad, R. L.; Veltkamp, J. J., and Aronson, D. L.: Potentially thrombogenic materials in factor IX concentrates. Thromb. Diath. haemorrh. 33: 617-631 (1975). 10 Marchesi, S . L. and Burney, R.: Prothrombin complex concentrates and thrombosis. New Engl. J. Med. 290: 403-404 (1974).
Cash/Owens/Dal tonPrescott
11 MCnachC, D.; Fauvert, R. et Soulier, J. P.: Utilisation en hepatologie d’une fraction contenant prothrombin, le complex proconvertine-facteur Stuart et le facteur anti-hemophilique B (PPB). Path. Biol., Paris 7: 2515-2523 (1959). 12 MBnachC, D. and Guillin, M. C.: The use of factor IX concentrates for patients with conditions other than factor IX deficiency. Br. J. Haemat. 31: 247-250 (1975). 13 Pepper, D. S.; Banhegyi, D.; Howie, A., and Cash, J. D.: In vitro thrombogenicity tests of factor IX concentrates. Br. J. Haemat. 36: 555-565 (1977). 14 Pepper, D. S.; Banhegyi, D., and Howie, A.: Spectrophotometric determination of factor Xa generation in Factor IX concentrates. Thromb. Haemostat. 37: 535-540 (1977). 15 Prowse, C. V.; Pepper, D. S.; Cash, J. D., and Patterson, M.: Thrombogenicity screening of factor IX concentrates. Thromb. Diath. haemorrh. 38: 728 (1977). 16 Sas, G.; Owens, R. E.; Smith, J. K.; Middleton, S., and Cash, J. D.: In vitro spontaneous thrombin generation in human factor IX concentrates. Br. J. Haemat. 31: 25-35 (1975). 17 Schimpf, K.; Zimmerman, K., and Kompf, B.: DIC and postoperative wound bleeding under factor IX substitution therapy in a case of hemophilia B; successful treatment with heparin. Thromb. Res. 8: 65-70 (1976). 18 Tullis, J. L. and Breen, F.: Christmas factor concentrates; the clinical use of several preparations. Biblthca haemat., vol. 34, pp. 40-51 (Karger, Basel 1970). 19 Wessler, S.; Reimer, S. M., and Sheps, M. C.: Biological assay of a thrombosis-inducing activity in human serum. J. appl. Physiol. 14: 943-946 (I 959). 20 White, G. C.; Roberts, R. H.; Kingdon, H. S., and Lundblad, R. L.: Prothrombin complex concentrates. Potentially thrombogenic materials and clues to the mechanism of thrombosis in vivo. Blood 49: 159-170 (1977).
Received: August 31, 1977 Accepted: November 10, 1977 Dr. J. D. Cash, S-E Scotland Regional Blood Transfusion Service, Royal Infirmary, Edinburgh EH3 9HB (Scotland)
