- Email: [email protected]
A collaborative study to establish a Korea national biological standard for antithrombin concentrate
Thrombosis Research (2006) 117, 591 — 596
intl.elsevierhealth.com/journals/thre
REGULAR ARTICLE
A collaborative study to establish a Korea national biological standard for antithrombin concentrate Hye Na Kang *, Sung Han Lee, Soon Nam Kim, Choong Man Hong, Seok Ho Lee, Seung Hwa Hong Biologics Evaluation Department, Korea Food and Drug Administration, 5 Nokbun-Dong, Eunpyung-Gu, Seoul 122-704, Republic of Korea Received 1 February 2005; received in revised form 8 May 2005; accepted 8 May 2005 Available online 1 July 2005
KEYWORDS National biological standard; Antithrombin concentrate
Abstract Background and objectives: Six laboratories consisting of three manufacturers and three national control laboratories participated in a collaborative study to evaluate the suitability of a candidate material to serve as the first Korean National Standard for Antithrombin (AT) concentrate. Materials and methods: The potency of this candidate preparation was determined using the heparin cofactor chromogenic method. The method is described in the Minimum Requirements for Biological Products in Korea and in the European Pharmacopoeia. The candidate was calibrated against the second International Standard for AT concentrate, coded as 96/520. Results: The participants contributed data from a total of 90 independent assays and the results were accepted as statistically valid when the outcome of the analysis exhibited linear dose—response relationships and intersected at a common point at zero dose in the slope—ratio model. The combined potency estimates were obtained by taking the geometric means of results from all assays at each laboratory, and overall potency estimates were calculated as unweighted geometric means of results from all laboratories. The results were expressed in the form of histograms with 95% confidence intervals. Conclusions: According to the results of the collaborative study, the candidate preparation showed excellent intra- and inter-laboratory correlations and is judged to be suitable to serve as the Korean National Standard for AT concentrate with the following potency: 51.9 IU/vial (95% confidence intervals = 48.24~55.98 IU/vial). D 2005 Elsevier Ltd. All rights reserved.
* Corresponding author. Tel.: +82 2 380 1715; fax: +82 2 380 1349. E-mail address: [email protected] (H.N. Kang). 0049-3848/$ - see front matter D 2005 Elsevier Ltd. All rights reserved. doi:10.1016/j.thromres.2005.05.016
592
Introduction Antithrombin (AT) is a natural inhibitor of the blood coagulation cascade. AT concentrate has been used for hereditary or acquired AT deficiencies [1]. Like other biologics, the establishment of the standard for AT concentrate is necessary for authorities at the national control laboratories and for quality control of manufactures. In 1980, the first International Reference Preparation for AT was established as a plasma preparation, and then the first and the second International Standards for AT concentrate were established in 1990 and in 1999, respectively, to decrease disagreement of inter-laboratory variability when using a plasma standard for concentrates [2,3]. The Korea Food and Drug Administration (KFDA) has been a leader in international collaborative studies for establishing several national standards. The purpose of these studies has been to overcome shortages in supply of international standards and to encourage comparison of results among different laboratories through a collaborative effort [4]. We organized an international collaborative study to calibrate the biological activity of the candidate material AT concentrate against the second International Standard for AT concentrate and to evaluate the suitability of this candidate to serve as the first National Standard for AT Concentrate.
Materials and methods Participants Six participants involving 3 manufacturers and 3 National Control Laboratories contributed to this study (see Appendix). Each laboratory was assigned a randomly generated code number for this study.
Materials The second international standard A freeze-dried concentrate of AT concentrate, coded 96/520 was used throughout the study, with the following potencies: functional potency 4.7 IU/vial; antigenic potency 5.1 IU/vial [3]. The proposed Korean standard According to the World Health Organization (WHO) Guideline for the Preparation of International and Other Standards and Reference Materi-
H.N. Kang et al. als for Biological Substances [5] and the Minimum Requirements for Biological Products in Korea, issued by the KFDA [6], the candidate material was manufactured by Korea Green Cross PBM. During the manufacturing, the candidate material was purified from human plasma supplied by the Plasma Fractionation Center, Republic of Korea Red Cross. The candidate material was formulated with inactive ingredients (10 mg of Glycine, 3 mg of Tris sodium citrate and 8 mg of sodium chloride per vial) and was adjusted to approximately 50 IU/mL before distribution. The potency of the final bulk solution was calculated based on values in the standard of NIBSC (96/520). The material was distributed into 5000 rubber-stoppered vials to a final volume of 1.0 mL. The mean filling weight was 1.014 g (CV 0.71%), and the material in each vial was lyophilized. The vials of AT concentrate for the national standard were stored at 20 8C. Meanwhile, the human pooled plasma was tested and found negative for HBsAg, anti-HIV 1/ 2 and anti-HCV before use. AT was purified through heparin Sepharose gel from the filtrate of Fractions II + III. In order to address the potential inactivation of live viruses, the heat inactivation process was introduced for 10 h at 60 F 10 8C. Further purification was achieved using heparin affinity chromatography. The entire manufacturing process was validated by the heparin-binding fraction test as described in the European Pharmacopoeia [7].
Study design Except for two laboratories in NIBSC and NIID which conducted assays on 8 vials per day, all domestic participants performed assays using 3 vials per day. Per vial, each assay was repeated to estimate intra-assay variation which is meaningful for statistical validity and precision. As in our previous study, independent assays on 6 different days were required to calculate the variability within a single day and between days [8]. The participants returned the raw data, and KFDA conducted statistical analysis of the data.
Assay methods To determine the potency of AT, all participants used the heparin co-factor chromogenic method. The assay was designed for slope—ratio model analysis in which the interval between adjacent doses must be constant for all treatments, and the
Korean standard for antithrombin concentrate potency of the candidate material was calculated relative to the second international standard for AT concentrate, coded 96/520 [3]. Except for the chromogenic substrate S-2238 (Chromogenix AB, Sweden), all reagents for the tests were obtained from Sigma.
Stability studies The stability studies of the candidate material were performed through real-time stability tests and accelerated thermal degradation tests. For accelerated degradation tests which is used for predicting long-term stability, 15 vials were stored at 20, 4, 20, 37, and 45 8C, respectively. After 6 months the predicted potency loss at 20 8C were calculated by the DEGTEST program at NIBSC [9—11].
593 Table 2 Comparison of potencies (IU/mL), grouped by each laboratory Laboratory code
n
GM
95% confidence intervals
1 2 3 4 5 6 Grand mean
19 18 18 18 8 8 6
56.43 54.74 49.89 54.24 47.03 49.63 51.92
55.66—57.21 53.66—55.85 49.51—50.30 53.37—55.12 45.42—48.71 46.39—53.10 48.24—55.88
Grand mean: calculated as unweighted mean of the laboratory means.
significance of differences in potencies at each storage temperature in the stability tests.
Statistical analysis
Results
Validity of each assay and relative potency of samples were estimated with the bStatistical Analysis of Results of Biological Assays and TestsQ of the European Pharmacopoeia [12]. If the results of the analysis showed linearity in dose— response relationships and intersected at the zero dose in the slope—ratio model, the assay was considered as dstatistically validT [7]. After the validity test of each assay by the slope—ratio model, the relative potencies were calculated using the ratio of their slopes [13]. The potencies of all assays from each laboratory were combined, and then they were calculated as geometric means of results from all laboratories. The results are shown as histograms with 95% confidence intervals. Variation tests using within-assay variability, between-assay variability, within-day variability, between-day variability, intra-laboratory variability, and inter-laboratory variability were estimated with the geometric coefficients of variations (%GCV) of potency estimates [13,14]. Student’s t-test was used to assess the differences between valid and invalid results. Also, ANOVA, Kruskal—Wallis test, and Bonferroni’s comparison test were utilized to assess the
Assessment of assay data
Table 1 Comparison of potencies (IU/mL) by validity of assay results Validity
n
GM
95% confidence intervals (GM)
GCV (%)
Invalid Valid
16 148
57.11a 52.96a
55.38—58.65 52.38—53.55
5.5 7.0
a
Invalid N valid ( p b 0.01) by Student’s t-test.
Six participants conducted 90 independent assays, with data from a total of 164 assays. According to validity tests using the slope—ratio model, data from 148 assays (= 90.2%) were accepted as valid, while 9.8% of results were invalid (non-interaction and/or non-linearity, Table 1). Since the invalid results showed significant differences from the valid results ( p b 0.01), they were not included in calculating the overall mean potency of this standard.
Analysis of assay data The valid results from each laboratory are presented in Table 2. The unweighted geometric mean
30
20
10
0
41.0 43.0 45.0 47.0 49.0 51.0 53.0 55.0 57.0 59.0 61.0 63.0 65.0
Relative Potency
Figure 1
Distribution of titer in valid results.
No. of Labs Participated
594
H.N. Kang et al. Table 4
Accelerated thermal degradation test % Activity remaininga
4 3 4
2 1 5
6
2
3
1
+4 8C +20 8C +37 8C +45 8C
40 45 50 55 60 Estimated potency related to International Standard
Observed
Predicted
100.55 96.73 94.52 91.28
99.38 97.99 94.35 91.29
Predicted loss at 20 8C (% per year) 0.227
Figure 2 Distribution of the potency estimates individual laboratories relative to the second International Standard Antithrombin concentrate. Each box represents a potency estimate calculated as a geometric mean with the numbers in the square denoting the laboratory code.
The potency of vials of the candidate for a National Standard stored at elevated temperatures for 6 months, was expressed as a percentage of the potency of vials stored at 20 8C. The predicted loss of activity per year was calculated using the Arrhenius equation. a Relative samples to 20 8C.
was used to calculate the overall potency of the standard. The distribution of titers in valid results is shown in Fig. 1, and one of the individual laboratories potency relative to the International Standard is presented as a histogram (see Fig. 2).
of potency per year calculated by the Arrhenius equation was 0.227%.
Variability
Basically, a collaborative study to establish the first national standard for AT concentrate was conducted according to the WHO Guideline for the Preparation of International and Other Standards and Reference Materials for Biological Substances [5]. KFDA as bproject leaderQ of the study examined candidate material and documented it, studied design, reviewed procedure, and conducted statistical analysis to confirm that the candidate material conformed to national standards [15]. More specifically, KFDA reviewed related documents such as flow diagrams of manufacturing process, certificates of analysis, process validation, viral clearance, and safety as manufacturer’s products. The heparin-cofactor chromogenic assay based on the Minimum Requirements for Biological Products in Korea [6] and on the European Pharmacopoeia [7] was accepted by all participants. The method was designed to analyze the results statistically. Also, KFDA analyzed the results from the raw data generated by each laboratory, and assessed the suitability of the candidate material as a national standard. This controlled collaborative study has the advantage of resulting in lower inter-laboratory variability than field-type collaborative studies, in
The variability of each laboratory and the overall mean that was estimated within assay, within day, between days, within laboratory, and between laboratories was expressed as geometric coefficients of variation (%GCV) of the potency estimates (Table 3). At each laboratory, the variability within an assay and between assays (within laboratory) ranged from 1.5 to 3.4 and from 1.6 to 8.4, respectively. The variability ranged from 1.5 to 2.7 for within a day and from 0.9 to 3.5 for between days. Overall, inter-laboratory variability was estimated as 7.3.
Stability study The candidate materials which had been respectively stored at 20, 4, 20, 37, and 45 8C for 6 months were assayed, and the mean potencies of samples from 4, 20, 37, or 45 8C were calculated against samples at 20 8C as the reference (Table 4). The mean potencies at each temperature demonstrated no significant differences compared to ones from other temperature. The predicted loss Table 3
Discussion
Variability (% GCV)
Variability
Within-assay variability Within-day variability Between-day variability Intra-laboratory variability (between-assay variability) Inter-laboratory variability
Laboratory code 1
2
3
4
5
6
Overall mean
1.5 2.0 2.1 2.9 7.3
3 2.7 3.5 4.1
1.5 1.5 0.9 1.6
3.4 2.7 2.3 3.3
— — — 4.3
— — — 8.4
2.4 2.2 2.2 41
Korean standard for antithrombin concentrate
595
which participants use their own method in the tests and analysis [16]. Altogether, 148 of 164 assays (= 90.2%) were accepted as statistically valid results according to the slope—ratio model (Table 1) and were included in subsequent analyses. 16 assays (= 9.8%) showed statistical invalidity (e.g. non-interaction: 11 assays, non-linearity: 3 assays, and both analyses of variance: 2 assays) were excluded from the study due to significant differences from valid outcomes. For reference, to establish the first and second international standards for AT [2,3], results from 88% and 80.5% of total assays, respectively, were used subsequent analyses in a collaborative study. The unweighted geometric mean, which can measure the results more appropriately in a log-normal distribution (Fig. 1), was used to calculate overall potency of the standard [13]. Overall, the mean potency was 51.92 IU/vial (95% confidence interval = 48.24~55.98 IU/vial) as the geometric mean. Also, the results of the assays were plotted in histogram form to prevent overlooking significant information in numerical study as established by WHO guidelines [5]. While relative potencies were being determined, differences arose from several sources, such as biological and experimental sources. The differences at the lowest level may have come from within-assay variations. Differences at the next level may have been due to between-assay variation, while differences at the highest level may have occurred from between-laboratory variation [13]. In this study we expressed the differences as %GCV, which showed good agreement at each level (Table 3). Predicted loss per year of the standard was 0.227%, when stored at 20 8C (Table 4). However, the predicted loss in the accelerated thermal degradation study seems to have been overestimated because the real-time stability study demonstrated that the candidate material was far more stable than the prediction. It should be enough to monitor the stability of the standard in breal-timeQ studies throughout its lifetime, usually considered as 5 to 6 years in case of the National Standard [9]. We conclude that the candidate reference standard is judged as suitable to serve as a Korean Reference Standard for AT Concentrate, with an assigned potency of 51.9 IU/vial.
sor Kyu-Yoon Hwang of Soonchunhyang University for statistical analysis. Especially, we are indebted to Dr. Elain Gray and Dr. Gill Creeber at National Institute of Biological Standardization and Control (NIBSC) for their valuable encouragement and advice about the collaborative study design and the accelerated thermal degradation study.
Acknowledgements
References
The participants’ contributions are gratefully acknowledged. We are grateful to Korea Red Cross for the supply of plasma, to Korea Green Cross PBM for the manufacture of AT concentrate, and to Profes-
Appendix A. List of participants Hye Na Kang Biologics Evaluation Department Korea Food and Drug Administration Nokbun-Dong, Eunpyung-Ku Seoul 122-704, Korea Elain Gray Division of Haematology NIBSC Blanche Lane, South Mimms, Potters Bar Herts., EN6 3QG, UK Yoshiaki Okada Laboratory of Blood Products Department of Safety Research on Blood and Biological Products National Institute of Infectious Diseases Japan Jeong Sup Shin Department of Quality Control and Assurance Korea Green Cross PD Corporation 227, Kuga-Ri, Kiheung-Eup, Yongin-Si Kyunggi-Do, Korea Seok Hwa Ryu Quality Control Section Korea Red Cross Blood Products Research Institute 41-1 Tanpyong-Ri, Kamgok-Myun, Umsong-Gun Chungbuk 369-850, Korea Jae-Myung Cha R&D Part Dong Shin Pharmaceutical Co. 12 Kohyun-Dong, Osan-Si Kyunggi-Do 447-320, Korea
[1] Lechner K, Kryle PA. Antithrombin III concentrates—are they clinically useful? Thromb Haemost 1955;73(3):340 – 8. [2] Kirwppd TBL, Barrowcliffe TW, Thomas DP. An international collaborative study establishing a reference preparation for antithrombin III. Thromb Haemost 1980;43:10 – 5.
596 [3] Gray E, Walker AD, Heath AB. A collaborative study to establish the 2nd international standard for antithrombin concentrate. Thromb Haemost 1999;82:46 – 50. [4] Campbell PJ. International biological standards and reference preparations: I. Preparation and presentation of materials to serve as standards and reference preparations. J Biol Stand 1974;2:249 – 58. [5] WHO Technical Report Series No. 800, Guidelines for the preparation, characterization and establishment of international and other standards and reference reagents for biological substances; 1990. [6] Freeze-dried concentrated human antithrombin III. The Minimum Requirements for Biological Products in Korea. Korea Food and Drug Administration; 2000. [7] Assay of Human Antithrombin III. European pharmacopoeia. 4th ed. 2002. p. 176. [8] Kang HN, Kin SN, Lee SH, Hong SH. A collaborative study to establish a Korean standard for factor VIII:C concentrate. Thromb Res 2004;113:261 – 7. [9] Tydeman MS, Kirwood TBL. Design and analysis of accelerated degradation tests for the stability of biological standards: I. Properties of maximum likelihood estimators. J Biol Stand 1984;12:195 – 206.
H.N. Kang et al. [10] Kirkwood TBL, Tydeman MS. Design and analysis of accelerated degradation tests for the stability of biological standards: II. A flexible computer program for data analysis. J Biol Stand 1984;12:207 – 14. [11] Kirkwood TBL. Design and analysis of accelerated degradation tests for the stability of biological standards: III. Principles of design. J Biol Stand 1984;12:215 – 24. [12] Statistical analysis of results of biological assays and tests. European pharmacopoeia supplement. Council of Europe; 2000. p. 263—93. [13] Kirkwood TBL, Seagroatt VA, Smith SJ. Statistical aspects of the planning and analysis of collaborative studies on biological standards. J Biol Stand 1986;14:273 – 87. [14] Kirkwood TBL. Geometric means and measures of dispersion. Biometrics 1979;35:908 – 9. [15] The European Pharmacopoeia Forum. Special issue BIO 96-1. EDQM; 1996. [16] Barrowcliffe TW, Raut S, Hubbard AR. Discrepancies in potency assessment of recombinant FVIII concentrates. Haemophilia 1998;4:634 – 40.
intl.elsevierhealth.com/journals/thre
REGULAR ARTICLE
A collaborative study to establish a Korea national biological standard for antithrombin concentrate Hye Na Kang *, Sung Han Lee, Soon Nam Kim, Choong Man Hong, Seok Ho Lee, Seung Hwa Hong Biologics Evaluation Department, Korea Food and Drug Administration, 5 Nokbun-Dong, Eunpyung-Gu, Seoul 122-704, Republic of Korea Received 1 February 2005; received in revised form 8 May 2005; accepted 8 May 2005 Available online 1 July 2005
KEYWORDS National biological standard; Antithrombin concentrate
Abstract Background and objectives: Six laboratories consisting of three manufacturers and three national control laboratories participated in a collaborative study to evaluate the suitability of a candidate material to serve as the first Korean National Standard for Antithrombin (AT) concentrate. Materials and methods: The potency of this candidate preparation was determined using the heparin cofactor chromogenic method. The method is described in the Minimum Requirements for Biological Products in Korea and in the European Pharmacopoeia. The candidate was calibrated against the second International Standard for AT concentrate, coded as 96/520. Results: The participants contributed data from a total of 90 independent assays and the results were accepted as statistically valid when the outcome of the analysis exhibited linear dose—response relationships and intersected at a common point at zero dose in the slope—ratio model. The combined potency estimates were obtained by taking the geometric means of results from all assays at each laboratory, and overall potency estimates were calculated as unweighted geometric means of results from all laboratories. The results were expressed in the form of histograms with 95% confidence intervals. Conclusions: According to the results of the collaborative study, the candidate preparation showed excellent intra- and inter-laboratory correlations and is judged to be suitable to serve as the Korean National Standard for AT concentrate with the following potency: 51.9 IU/vial (95% confidence intervals = 48.24~55.98 IU/vial). D 2005 Elsevier Ltd. All rights reserved.
* Corresponding author. Tel.: +82 2 380 1715; fax: +82 2 380 1349. E-mail address: [email protected] (H.N. Kang). 0049-3848/$ - see front matter D 2005 Elsevier Ltd. All rights reserved. doi:10.1016/j.thromres.2005.05.016
592
Introduction Antithrombin (AT) is a natural inhibitor of the blood coagulation cascade. AT concentrate has been used for hereditary or acquired AT deficiencies [1]. Like other biologics, the establishment of the standard for AT concentrate is necessary for authorities at the national control laboratories and for quality control of manufactures. In 1980, the first International Reference Preparation for AT was established as a plasma preparation, and then the first and the second International Standards for AT concentrate were established in 1990 and in 1999, respectively, to decrease disagreement of inter-laboratory variability when using a plasma standard for concentrates [2,3]. The Korea Food and Drug Administration (KFDA) has been a leader in international collaborative studies for establishing several national standards. The purpose of these studies has been to overcome shortages in supply of international standards and to encourage comparison of results among different laboratories through a collaborative effort [4]. We organized an international collaborative study to calibrate the biological activity of the candidate material AT concentrate against the second International Standard for AT concentrate and to evaluate the suitability of this candidate to serve as the first National Standard for AT Concentrate.
Materials and methods Participants Six participants involving 3 manufacturers and 3 National Control Laboratories contributed to this study (see Appendix). Each laboratory was assigned a randomly generated code number for this study.
Materials The second international standard A freeze-dried concentrate of AT concentrate, coded 96/520 was used throughout the study, with the following potencies: functional potency 4.7 IU/vial; antigenic potency 5.1 IU/vial [3]. The proposed Korean standard According to the World Health Organization (WHO) Guideline for the Preparation of International and Other Standards and Reference Materi-
H.N. Kang et al. als for Biological Substances [5] and the Minimum Requirements for Biological Products in Korea, issued by the KFDA [6], the candidate material was manufactured by Korea Green Cross PBM. During the manufacturing, the candidate material was purified from human plasma supplied by the Plasma Fractionation Center, Republic of Korea Red Cross. The candidate material was formulated with inactive ingredients (10 mg of Glycine, 3 mg of Tris sodium citrate and 8 mg of sodium chloride per vial) and was adjusted to approximately 50 IU/mL before distribution. The potency of the final bulk solution was calculated based on values in the standard of NIBSC (96/520). The material was distributed into 5000 rubber-stoppered vials to a final volume of 1.0 mL. The mean filling weight was 1.014 g (CV 0.71%), and the material in each vial was lyophilized. The vials of AT concentrate for the national standard were stored at 20 8C. Meanwhile, the human pooled plasma was tested and found negative for HBsAg, anti-HIV 1/ 2 and anti-HCV before use. AT was purified through heparin Sepharose gel from the filtrate of Fractions II + III. In order to address the potential inactivation of live viruses, the heat inactivation process was introduced for 10 h at 60 F 10 8C. Further purification was achieved using heparin affinity chromatography. The entire manufacturing process was validated by the heparin-binding fraction test as described in the European Pharmacopoeia [7].
Study design Except for two laboratories in NIBSC and NIID which conducted assays on 8 vials per day, all domestic participants performed assays using 3 vials per day. Per vial, each assay was repeated to estimate intra-assay variation which is meaningful for statistical validity and precision. As in our previous study, independent assays on 6 different days were required to calculate the variability within a single day and between days [8]. The participants returned the raw data, and KFDA conducted statistical analysis of the data.
Assay methods To determine the potency of AT, all participants used the heparin co-factor chromogenic method. The assay was designed for slope—ratio model analysis in which the interval between adjacent doses must be constant for all treatments, and the
Korean standard for antithrombin concentrate potency of the candidate material was calculated relative to the second international standard for AT concentrate, coded 96/520 [3]. Except for the chromogenic substrate S-2238 (Chromogenix AB, Sweden), all reagents for the tests were obtained from Sigma.
Stability studies The stability studies of the candidate material were performed through real-time stability tests and accelerated thermal degradation tests. For accelerated degradation tests which is used for predicting long-term stability, 15 vials were stored at 20, 4, 20, 37, and 45 8C, respectively. After 6 months the predicted potency loss at 20 8C were calculated by the DEGTEST program at NIBSC [9—11].
593 Table 2 Comparison of potencies (IU/mL), grouped by each laboratory Laboratory code
n
GM
95% confidence intervals
1 2 3 4 5 6 Grand mean
19 18 18 18 8 8 6
56.43 54.74 49.89 54.24 47.03 49.63 51.92
55.66—57.21 53.66—55.85 49.51—50.30 53.37—55.12 45.42—48.71 46.39—53.10 48.24—55.88
Grand mean: calculated as unweighted mean of the laboratory means.
significance of differences in potencies at each storage temperature in the stability tests.
Statistical analysis
Results
Validity of each assay and relative potency of samples were estimated with the bStatistical Analysis of Results of Biological Assays and TestsQ of the European Pharmacopoeia [12]. If the results of the analysis showed linearity in dose— response relationships and intersected at the zero dose in the slope—ratio model, the assay was considered as dstatistically validT [7]. After the validity test of each assay by the slope—ratio model, the relative potencies were calculated using the ratio of their slopes [13]. The potencies of all assays from each laboratory were combined, and then they were calculated as geometric means of results from all laboratories. The results are shown as histograms with 95% confidence intervals. Variation tests using within-assay variability, between-assay variability, within-day variability, between-day variability, intra-laboratory variability, and inter-laboratory variability were estimated with the geometric coefficients of variations (%GCV) of potency estimates [13,14]. Student’s t-test was used to assess the differences between valid and invalid results. Also, ANOVA, Kruskal—Wallis test, and Bonferroni’s comparison test were utilized to assess the
Assessment of assay data
Table 1 Comparison of potencies (IU/mL) by validity of assay results Validity
n
GM
95% confidence intervals (GM)
GCV (%)
Invalid Valid
16 148
57.11a 52.96a
55.38—58.65 52.38—53.55
5.5 7.0
a
Invalid N valid ( p b 0.01) by Student’s t-test.
Six participants conducted 90 independent assays, with data from a total of 164 assays. According to validity tests using the slope—ratio model, data from 148 assays (= 90.2%) were accepted as valid, while 9.8% of results were invalid (non-interaction and/or non-linearity, Table 1). Since the invalid results showed significant differences from the valid results ( p b 0.01), they were not included in calculating the overall mean potency of this standard.
Analysis of assay data The valid results from each laboratory are presented in Table 2. The unweighted geometric mean
30
20
10
0
41.0 43.0 45.0 47.0 49.0 51.0 53.0 55.0 57.0 59.0 61.0 63.0 65.0
Relative Potency
Figure 1
Distribution of titer in valid results.
No. of Labs Participated
594
H.N. Kang et al. Table 4
Accelerated thermal degradation test % Activity remaininga
4 3 4
2 1 5
6
2
3
1
+4 8C +20 8C +37 8C +45 8C
40 45 50 55 60 Estimated potency related to International Standard
Observed
Predicted
100.55 96.73 94.52 91.28
99.38 97.99 94.35 91.29
Predicted loss at 20 8C (% per year) 0.227
Figure 2 Distribution of the potency estimates individual laboratories relative to the second International Standard Antithrombin concentrate. Each box represents a potency estimate calculated as a geometric mean with the numbers in the square denoting the laboratory code.
The potency of vials of the candidate for a National Standard stored at elevated temperatures for 6 months, was expressed as a percentage of the potency of vials stored at 20 8C. The predicted loss of activity per year was calculated using the Arrhenius equation. a Relative samples to 20 8C.
was used to calculate the overall potency of the standard. The distribution of titers in valid results is shown in Fig. 1, and one of the individual laboratories potency relative to the International Standard is presented as a histogram (see Fig. 2).
of potency per year calculated by the Arrhenius equation was 0.227%.
Variability
Basically, a collaborative study to establish the first national standard for AT concentrate was conducted according to the WHO Guideline for the Preparation of International and Other Standards and Reference Materials for Biological Substances [5]. KFDA as bproject leaderQ of the study examined candidate material and documented it, studied design, reviewed procedure, and conducted statistical analysis to confirm that the candidate material conformed to national standards [15]. More specifically, KFDA reviewed related documents such as flow diagrams of manufacturing process, certificates of analysis, process validation, viral clearance, and safety as manufacturer’s products. The heparin-cofactor chromogenic assay based on the Minimum Requirements for Biological Products in Korea [6] and on the European Pharmacopoeia [7] was accepted by all participants. The method was designed to analyze the results statistically. Also, KFDA analyzed the results from the raw data generated by each laboratory, and assessed the suitability of the candidate material as a national standard. This controlled collaborative study has the advantage of resulting in lower inter-laboratory variability than field-type collaborative studies, in
The variability of each laboratory and the overall mean that was estimated within assay, within day, between days, within laboratory, and between laboratories was expressed as geometric coefficients of variation (%GCV) of the potency estimates (Table 3). At each laboratory, the variability within an assay and between assays (within laboratory) ranged from 1.5 to 3.4 and from 1.6 to 8.4, respectively. The variability ranged from 1.5 to 2.7 for within a day and from 0.9 to 3.5 for between days. Overall, inter-laboratory variability was estimated as 7.3.
Stability study The candidate materials which had been respectively stored at 20, 4, 20, 37, and 45 8C for 6 months were assayed, and the mean potencies of samples from 4, 20, 37, or 45 8C were calculated against samples at 20 8C as the reference (Table 4). The mean potencies at each temperature demonstrated no significant differences compared to ones from other temperature. The predicted loss Table 3
Discussion
Variability (% GCV)
Variability
Within-assay variability Within-day variability Between-day variability Intra-laboratory variability (between-assay variability) Inter-laboratory variability
Laboratory code 1
2
3
4
5
6
Overall mean
1.5 2.0 2.1 2.9 7.3
3 2.7 3.5 4.1
1.5 1.5 0.9 1.6
3.4 2.7 2.3 3.3
— — — 4.3
— — — 8.4
2.4 2.2 2.2 41
Korean standard for antithrombin concentrate
595
which participants use their own method in the tests and analysis [16]. Altogether, 148 of 164 assays (= 90.2%) were accepted as statistically valid results according to the slope—ratio model (Table 1) and were included in subsequent analyses. 16 assays (= 9.8%) showed statistical invalidity (e.g. non-interaction: 11 assays, non-linearity: 3 assays, and both analyses of variance: 2 assays) were excluded from the study due to significant differences from valid outcomes. For reference, to establish the first and second international standards for AT [2,3], results from 88% and 80.5% of total assays, respectively, were used subsequent analyses in a collaborative study. The unweighted geometric mean, which can measure the results more appropriately in a log-normal distribution (Fig. 1), was used to calculate overall potency of the standard [13]. Overall, the mean potency was 51.92 IU/vial (95% confidence interval = 48.24~55.98 IU/vial) as the geometric mean. Also, the results of the assays were plotted in histogram form to prevent overlooking significant information in numerical study as established by WHO guidelines [5]. While relative potencies were being determined, differences arose from several sources, such as biological and experimental sources. The differences at the lowest level may have come from within-assay variations. Differences at the next level may have been due to between-assay variation, while differences at the highest level may have occurred from between-laboratory variation [13]. In this study we expressed the differences as %GCV, which showed good agreement at each level (Table 3). Predicted loss per year of the standard was 0.227%, when stored at 20 8C (Table 4). However, the predicted loss in the accelerated thermal degradation study seems to have been overestimated because the real-time stability study demonstrated that the candidate material was far more stable than the prediction. It should be enough to monitor the stability of the standard in breal-timeQ studies throughout its lifetime, usually considered as 5 to 6 years in case of the National Standard [9]. We conclude that the candidate reference standard is judged as suitable to serve as a Korean Reference Standard for AT Concentrate, with an assigned potency of 51.9 IU/vial.
sor Kyu-Yoon Hwang of Soonchunhyang University for statistical analysis. Especially, we are indebted to Dr. Elain Gray and Dr. Gill Creeber at National Institute of Biological Standardization and Control (NIBSC) for their valuable encouragement and advice about the collaborative study design and the accelerated thermal degradation study.
Acknowledgements
References
The participants’ contributions are gratefully acknowledged. We are grateful to Korea Red Cross for the supply of plasma, to Korea Green Cross PBM for the manufacture of AT concentrate, and to Profes-
Appendix A. List of participants Hye Na Kang Biologics Evaluation Department Korea Food and Drug Administration Nokbun-Dong, Eunpyung-Ku Seoul 122-704, Korea Elain Gray Division of Haematology NIBSC Blanche Lane, South Mimms, Potters Bar Herts., EN6 3QG, UK Yoshiaki Okada Laboratory of Blood Products Department of Safety Research on Blood and Biological Products National Institute of Infectious Diseases Japan Jeong Sup Shin Department of Quality Control and Assurance Korea Green Cross PD Corporation 227, Kuga-Ri, Kiheung-Eup, Yongin-Si Kyunggi-Do, Korea Seok Hwa Ryu Quality Control Section Korea Red Cross Blood Products Research Institute 41-1 Tanpyong-Ri, Kamgok-Myun, Umsong-Gun Chungbuk 369-850, Korea Jae-Myung Cha R&D Part Dong Shin Pharmaceutical Co. 12 Kohyun-Dong, Osan-Si Kyunggi-Do 447-320, Korea
[1] Lechner K, Kryle PA. Antithrombin III concentrates—are they clinically useful? Thromb Haemost 1955;73(3):340 – 8. [2] Kirwppd TBL, Barrowcliffe TW, Thomas DP. An international collaborative study establishing a reference preparation for antithrombin III. Thromb Haemost 1980;43:10 – 5.
596 [3] Gray E, Walker AD, Heath AB. A collaborative study to establish the 2nd international standard for antithrombin concentrate. Thromb Haemost 1999;82:46 – 50. [4] Campbell PJ. International biological standards and reference preparations: I. Preparation and presentation of materials to serve as standards and reference preparations. J Biol Stand 1974;2:249 – 58. [5] WHO Technical Report Series No. 800, Guidelines for the preparation, characterization and establishment of international and other standards and reference reagents for biological substances; 1990. [6] Freeze-dried concentrated human antithrombin III. The Minimum Requirements for Biological Products in Korea. Korea Food and Drug Administration; 2000. [7] Assay of Human Antithrombin III. European pharmacopoeia. 4th ed. 2002. p. 176. [8] Kang HN, Kin SN, Lee SH, Hong SH. A collaborative study to establish a Korean standard for factor VIII:C concentrate. Thromb Res 2004;113:261 – 7. [9] Tydeman MS, Kirwood TBL. Design and analysis of accelerated degradation tests for the stability of biological standards: I. Properties of maximum likelihood estimators. J Biol Stand 1984;12:195 – 206.
H.N. Kang et al. [10] Kirkwood TBL, Tydeman MS. Design and analysis of accelerated degradation tests for the stability of biological standards: II. A flexible computer program for data analysis. J Biol Stand 1984;12:207 – 14. [11] Kirkwood TBL. Design and analysis of accelerated degradation tests for the stability of biological standards: III. Principles of design. J Biol Stand 1984;12:215 – 24. [12] Statistical analysis of results of biological assays and tests. European pharmacopoeia supplement. Council of Europe; 2000. p. 263—93. [13] Kirkwood TBL, Seagroatt VA, Smith SJ. Statistical aspects of the planning and analysis of collaborative studies on biological standards. J Biol Stand 1986;14:273 – 87. [14] Kirkwood TBL. Geometric means and measures of dispersion. Biometrics 1979;35:908 – 9. [15] The European Pharmacopoeia Forum. Special issue BIO 96-1. EDQM; 1996. [16] Barrowcliffe TW, Raut S, Hubbard AR. Discrepancies in potency assessment of recombinant FVIII concentrates. Haemophilia 1998;4:634 – 40.