The second international standard for polymyxin B*

Bull. Org. mond. Sante 1973, 48, 85-90 Bull. Wld Hlth Org. The second international standard for polymyxin B J. W. LIGHTBOWN,1 A. H. THOMAS,1 B. GRAB,2 & A. S. OUTSCHOORN3 Since supplies of the first International Standard for Polymyxin B were exhauvted, it was replaced by a second international standard the potency of which was estimated from the results of a collaborative assay carried out by 5 laboratories in 4 countries. The wide variations in the results probably resultedfrom difficulties experienced in handling the first international standard. The potency finally agreed upon by the collaborating laboratories, on the basis of the overall mean values obtained after rejection of the most discrepant assays, was 8403 IU/mg. That value was accepted by the WHO Expert Com- mittee on Biological Standardization (1970), which consequently defined the International Unit of polymyxin B as the activity contained in 0.000119 mg of the second international standard. In its twenty-first report, the Expert Committee on Biological Standardization (1969) noted that stocks of the first International Standard for Polymyxin B, established in 1955, were depleted; that material suitable for its replacement had been obtained by the National Institute for Medical Research, London; and that a collaborative assay was in progress. The properties of the material and the results of the collaborative assay are described below. MATERIALS FOR THE STUDY The proposed second international standard for poly- myxin B A sample of about 400 g of polymyxin B sulfate, Batch No. 33274, was generously made available by Burroughs Wellcome & Co., Dartford, Kent, England, through the good offices of Dr G. A. Stewart. The sample was received at the National Institute for Medical Research, in May 1967, in a single container, and was stored at -20°C in the dark, protected from moisture. The following data were supplied by the manu- facturer. * From the WHO International Laboratory for Biological Standards, National Institute for Medical Research, Mill Hill, London, England. 1 Division of Biological Standards. 'Statistician, Health Statistical Methodology, World Health Organization, Geneva, Switzerland. 'Chief Medical Officer, Biological Standardization, World Health Organization, Geneva, Switzerland. description: white, almost odourless powder stability: very soluble in water, yielding a bright colour- less solution identity tests: complies with tests A, B, and C given under " Identification " in the British Pharmacopoeia, 1963 (page 629) acidity: pH of 2 Y. w/v solution=6.6 loss on drying at 600 in vacuo: 3.64% sulfated ash: 0.05% potency: estimated potency=7 743 IU/mg; limits of error 98.1-101.9% of estimated potency (P=0.95) pyrogen test (USP): passed (20 000 IU per kg of body weight in rabbits; summed response for 3 rab- bits=0.8°C; no individual response greater than 0.590) toxicity test (USP): passed (600 IU in 0.5 ml of 0.9% saline injected i.v. into each of 5 mice; no mice died) toxicity test (BP): passed (600 IU in 0.2 ml of water injected i.v. into each of 5 mice; no mice died) composition: hydrolysis and GLC of the resulting fatty acids revealed the following composition: polymyxin B1 sulfate, 77.5%; polymyxin B2 sulfate, 21.4%; uniden- tified product yielding an unknown fatty acid, 1.10% Preliminary attempts to distribute the material into ampoules as a dry powder were unsatisfactory. After exhaustive drying, it was found impossible to seal the ampoules by fusion of the glass without the charring of particles of powder in the vicinity of the seal. The material was therefore dissolved in water, distributed into ampoules as a solution, and freeze- dried. The details of this procedure are as follows: In December 1967, the sample was dissolved at a concentration of approximately 75 mg per ml, in double-distilled water prepared with a glass still. The solution was filtered through Millipore AP20 and SC 2983 -85 J. W. LIGHTBOWN ET AL. membrane filters and distributed in 1-ml quantities into approximately 4 000 nonactinic glass ampoules. After freeze-drying, a vented polyethylene plug was inserted into each of the ampoules, which were fur- ther dried in vacuum over phosphorus(V) oxide for 10 days. The ampoules were then filled with pure dry nitrogen, sealed by fusion of the glass, tested for leaks, and stored at -20°C in the dark. No significant loss in weight was found on drying at 60°C in vacuo at a pressure of < 0.04 mmHg over phosphorus(V) oxide to constant weight. The (first) International Standardfor Polymyxin B This was described by Humphrey et al. (1959). Its assigned potency was 7 874 IU/mg. The quantity of powder in each ampoule was small (approxi- mately 20 mg) and it was therefore recommended that, for the collaborative assay, samples should be weighed with a micro or semimicro balance. THE COLLABORATIVE ASSAY Five laboratories in four countries agreed to par- ticipate in the collaborative assay. The names of the participants are listed in Annex 1, but throughout the report they are referred to by a number that has no connexion with the order of listing. Every labora- tory was supplied with 5 ampoules, each containing 75 mg of the proposed second international standard for polymyxin B, together with 5 ampoules of the International Standard for Polymyxin B, each con- taining approximately 20 mg. The participating laboratories were asked to assay the proposed second international standard against the first. They were informed that any recognized method of assay would be acceptable and that each assay should contain sufficient information to pro- vide an estimate of the potency of the proposed second standard in terms of the International Stan- dard, as well as evidence of validity and fiducial limits for that assay from its own internal evidence. It was hoped that the results of at least 4 assays, using at least 2 independent sets of weighings and dilutions, would be provided by each laboratory. The participants were reminded that the prepara- tions should not be dried before use. The material in the ampoules of the proposed second international standard was found to take up moisture at the rate of approximately 0.5% w/w per 5 min when exposed in an open weighing bottle at a relative humidity of 36%. It had been hoped that some of the participating laboratories might use chemical or physical methods for the evaluation of polymyxin B, but none of them did. Table 1. Details of assay methods used in different laboratories Labor- Dose levels of each No. of Temper- atory Test organism Type of preparation Structure of assay assays ature ofatorysa ( /l per- incuba-No. assay lU mi) ~~~~~~~~~~~~~~~~~~~~~~~~~formed tion 1 Bordetella diffusion, 8.0; 10.0; 12.5 3 + 3; 1 of each concentration of each 6 37°C bronchiseptica plates preparation on a plate; 9 plates per assay (ATCC 4617) 2 B. bronchiseptica diffusion, 150; 300; 600 3 + 3; 4 replicates of each concentration 78 37°C (Wellcome large of each preparation; 1 plate per assay CN385) plates 3 B. bronchiseptica diffusion, 500; 1 000; 2 000 3 + 3; 1 of each concentration of each 8 37°C (NCTC 8344) Petri preparation on a dish; approx. 20 dishes dishes per assay 4 B. bronchiseptica diffusion, 25; 50; 100 3 + 3; 3 replicates of each concentration of 2 (ATCC 4617) Petri each preparation on a dish; 30 dishes per dishes assay 5 B. bronchiseptica diffusion, 40.0; 89.4; 200.0 3 + 3; 2 replicates of each concentration of 3 370C (NCTC 8344) large one preparation and 1 of each concentra- plates tion of the other preparations arranged in a 9 x 9 Latin square design; 1 plate per assay 40; 200 2 + 2; 2 replicates of each concentration 24 of each preparation arranged in an 8 x 8 Latin square design; 1 plate per assay 86 POLYMY)IN B RESULTS The 5 laboratories performed a total of 121 assays, all by the diffusion method. Table 1 gives details of the assays performed and indicates the methods used. STATISTICAL ANALYSIS The conventional statistical method for parallel line assays was used for the analysis of the experi- mental results. The diameter of the zone of inhibi- tion of growth was taken as the response metameter and was analysed in relation to the logarithm of the dose. The potency estimate of the proposed second standard and its precision were determined for each individual assay, and the significance of departures from parallelism and linearity of dose-response regression lines was statistically tested. The homogeneity of the potency estimates was studied through the X2-test, according to the method of Humphrey et al. (1953). When the results of a particular laboratory did not show significant heterogeneity, a weighted aver- age logarithmic potency was computed by weighting each potency value with the reciprocal variance based on the internal evidence of the assay. The variance of this average potency was then simply the reciprocal of the total of the individual weights. For laboratories that obtained statistically hetero- geneous results, the between-assay variance was com- puted and the weight of each logarithmic potency value was redetermined by taking the reciprocal of the total variance-i.e., including both within-assay and between-assay variances according to the method described by Bliss (1952). The new weights were applied to the individual potencies in computing the average logarithmic potency, and its variance was taken as the reciprocal of the sum of the new individual weights. The same method was used for computing the combined weighted log potency estimate from the laboratory average log potencies. The correspond- ing unweighted average estimate was also established for the purpose of comparison. The results of the statistical tests of deviation from parallelism and linearity are summarized in Table 2. This shows that the number of assays showing a sig- nificant departure from parallelism at the 5% proba- bility level is very small. As for departure from linearity, 12 of the 97 three-point assays carried out were nonlinear. It can be seen from the last two columns of Table 2 that the majority were concave lines. Log potency estimates obtained by repeated assay within a single laboratory were heterogeneous in the case of 3 out of the 5 laboratories. Laboratory 1 The 6 potency values obtained in this laboratory were homogeneous. However, the choice of doses and dose intervals was such that for several assays there was no marked difference between the responses at the low and medium doses and between those at the medium and high doses. As a consequence, 3 of the assays were statistically invalid, showing a signifi- cant departure from linearity at the 1% probability level. Laboratory 2 Significant heterogeneity at the 1% probability level was observed among the individual potencies obtained by this laboratory, which varied from Table 2. Significant departure a from parallelism and linearity-summary results Laboratory No o No. of non- No. No. Sign for sum of quadratics No. Y No. of assays parallel of 3-point of nonlinearassays assays assays b -(convex) + (concave) 1 6 1 6 3 0 3 2 78 3 78 4 1 3 3 8 3 8 2 0 2 4 2 0 2 2 0 2 5 27 4 3 1 0 1 total 121 11 97 12 1 11 a At the 5 % probability level of significance. b With significant sum of squares for quadratics 87 J. W. LIGHTBOWN ET AL. Table 3. Summary results of assays of proposed second international standard for polymyxin B Weighted Wihn-laboratory Total Labor- No. of Homogeneity Weic Weighted Total within- variance of Total laboratory adjusted atory assays within laboratory average average aboraty average log variance a laboratorypotency)log potency (iv) potency weight 1 6 homogeneous 7 902.9 3.89779 19 661.9 0.000 050 860 0.001 157 607 863.9 2 78 heterogeneous 9429.7 3.97450 b 41 844.3C 0.000 023 898 0.001 130 645 884.5 3 8 heterogeneous 9 136.8 3.96079 b 26 951.2 C 0.000 037 104 0.001 143 851 874.2 4 2 homogeneous 8 202.9 3.91397 8 106.5 0.000123 358 0.001 230 105 812.9 5 27 heterogeneous 8 115.7 3.90933 b 29 307.9 C 0.000 034120 0.001 140 867 876.5 total 121 125 871.8 4 312.0 a Including between-laboratory variance (0.001 106 747). b Weighted with weights adjusted for within-laboratory heterogeneity. c Total of adjusted weights. 9 336.2 IU/mg to 11 911.6 IU/mg. Possible sources of heterogeneity were examined, and it was seen that the greatest contribution to the total X2 for heterogeneity arose from variations between " am- poules/weighings " and between " days ". Laboratory 3 There was significant heterogeneity at the 1 % prob- ability level between the potencies of the 8 assays for this laboratory. This heterogeneity was traced to the potencies obtained from one set of " ampoules/ weighings " which yielded low potency estimates. Laboratory 4 This laboratory did 8 assays. However, only 2 were parallel line assays that could be tested for validity. The latter were found to be statistically invalid because of departure from linearity. Never- theless, the potencies were homogeneous. Laboratory S There was significant heterogeneity at the 1 % prob- ability level of the 27 potency values obtained in this laboratory. It was traced to the potencies obtained on 1 out of the 5 assay days from one set of " ampoules/weighings ". The final laboratory results are summarized in Table 3. The average potency values range from 7 902.9 in Laboratory 1 to 9 429.7 in Laboratory 2. Coverage All assays Valid assays Un weighted geometric No. of assays mean potency (IU/mg) 121 9 015.0 100 9 092.0 The X2 for between-laboratory homogeneity was much greater than the critical value at the 1% proba- bility level for 4 degrees of freedom. It was necessary, therefore, to take into account the additional variance component-the between- laboratory variance-in order to estimate the weighted average potency for all laboratories and its precision. The last two columns of Table 3 give the total variance and the corresponding adjusted weight for each laboratory. The combined weight log potency estimate was computed by applying to each laboratory log potency given in column (v) of Table 3 the corresponding weight shown in column (ix). The variance of this general average log potency was estimated as the reciprocal value of the sum of the total adjusted laboratory weights (1/4 312=0.00023191). The resulting combined potency of the proposed second international standard for polymyxin B was 8 544.6 IU/mg with a 95% confidence interval of 7 977.0-9 152.5 IU/mg. For the puipose of comparison, the general un- weighted geometric mean potency was calculated separately for all the 121 assays and for the valid assays only (i.e., excluding the assays with a statisti- cally significant departure from parallelism and/or linearity). The results were as follows. 95% confidence interval(lU/mg) 8 838.4-9 195.1 8 898.4-9 289.7 88 POLYMYXIN B Table 4. Summary results of assays of proposed second international standard for polymyxin B (based on 70 assays selected to produce homogeneity within laboratories) WegeometrcTtldihn Within-laboratory TotalLabor- No. of Homogeneity Weic Weighted variance of Total laboratory adjusted atory assays within laboratory average average a gratory average log variancea laboratorypotency log potency wegt potency weight(lU/mg)(i) (ii) (iii) (iv) (v) (vi) (vii) (viii) (ix) 1 6 homogeneous 7 902.9 3.89779 19 661.9 0.000 050 860 0.000 745 434 1 341.5 2 36 homogeneous 8650.4 3.93704 214 224.1 0.000 004 668 0.000 699 252 1 430.1 3 6 homogeneous 9319.7 3.96940 84 579.3 0.000 011 823 0.000 706 414 1 415.6 4 2 homogeneous 8 202.9 3.91397 8106.5 0.000123 358 0.000 817 929 1 222.6 5 20 homogeneous 8 368.0 3.92262 225 188.3 0.000 004 441 0.000 699 007 1 430.6 total 70 551 760.1 6840.4 a Including between-laboratory variance (0.000 694 578). Table 4 shows the effect of excluding from the final analysis the assays from Laboratories 2, 3, and 5 that appeared to be the most discrepant and the exclusion of which left a homogeneous group of assays within each laboratory. In this way a com- bined weighted mean potency of 8 489.8 IU/mg with ,a 95% confidence interval of 8 039.0-8 966.0 IU/mg was obtained. The results from Laboratory 3 form two groups: one group consisting of 2 assays, which was rejected in the above-mentioned calculation, and another group of 6 assays. When the latter group was rejected instead of the former, a combined weighted mean potency of 8 362.2 IU/mg was obtained with confidence limits of 8 099.2-8 633.7 IU/mg. A value lying between the two weighted -mean potency values was considered to be the best estimate of the potency of the second international standard for polymyxin that the study could provide. DISCUSSION The very great heterogeneity of the potency values obtained in this study, both within and between laboratories, may stem in part from differences in composition between the first international standard, which was manufactured prior to 1954, and the second standard, which is representative of current production. This difference may have resulted from variations in the content of polymyxins B1 and B2, but unfortunately insufficient material was available to allow this analysis to be carried out on the first international standard. Apart from this possible source of variation in assay, a number of laboratories found great difficulty in opening ampoules of the first international standard without contaminating the contents with glass fragments. This, together with the small amount of powder in each ampoule, the difficulty of removing it from the ampoules owing to static charge, and its rapid uptake of moisture during manipulation, made it very difficult to obtain samples weighed with great precision. Since supplies of the first international standard were exhausted, the National Institute for Medical Research, in accordance with the authorization of the WHO Expert Committee on Biological Stan- dardization (1969), established the material that had been examined as the second International Standard for Polymyxin B and suggested that the International Unit of polymyxin B be defined as the activity con- tained in 0.000119 mg of the preparation-a value corresponding to a potency lying within the range suggested by the collaborative assay, i.e., 8 362.2 to 8 489.8 IU/mg. ESTABLISHMENT OF THE INTERNATIONAL STANDARD AND DEFINITION OF THE INTERNATIONAL UNIT The WHO Expert Committee on Biological Stan- dardization (1970) considered the data that had been obtained in the collaborative assay and decided that, since it was unlikely that a more accurate estimate of potency could be made and since there was an urgent need for the unit to be defined, this should be done on the basis of the available results. The value suggested was acceptable to the collaborating laboratories, and the Expert Committee therefore defined the International Unit of polymyxin B as the activity contained in 0.000119 mg of the second International Standard for Polymyxin B. This cor- responds to a potency of 8 403 IU/mg. 89 90 J. W. LIGHTBOWN ET AL. ACKNOWLEDGEMENTS The authors acknowledge the help received from Dr P. J. Campbell, WHO International Laboratory for Biological Standards, who was responsible for the distribution of the material into ampoules. RI2SUM1t DEUXI[ME tTALON INTERNATIONAL DE POLYMYXINE B Les stocks du premier 6talon international de poly- myxine B, constitue en 1955, etant epuises, le National Institute for Medical Research, de Londres, s'est procure un 6chantillon de 400 g de sulfate de polymyxine B pouvant servir a constituer un deuxieme etalon. Dissous dans 1'eau, reparti en ampoules et lyophilise, ce materiel a ete distribue 'a cinq laboratoires de quatre pays pour etre soumis a un titrage comparatif avec le premier etalon international. On a pratique au total 121 6preuves par diffusion en utilisant comme organisme d'6preuve Bordetella bronchi- septica. Elles ont donne des r6sultats tr&s h6terogenes, les estimations de l'activit6 variant a la fois dans un meme laboratoire et entre les laboratoires. La principale raison de ces 6carts semble etre les difficult6s rencontrees lors de la manipulation des ampoules contenant le premier 6talon international. Apres exclusion des resultats les plus discordants, on a obtenu en combinant les donn6es fournies par les laboratoires deux valeurs moyennes d'activit6: 8489,8 UT/mg, avec des limites de confiance 95% de 8039,0 a 8966,0 UI/mg, et 8362,2 UT/mg, avec des limites de confiance 95% de 8099,2 A\8633,7 UT/mg. Estimant qu'il 6tait peu probable qu'on puisse 6valuer l'activite du deuxieme etalon avec davantage de pr6cision en pro- c6dant a de nouvelles 6preuves, les participants au titrage comparatif ont constitue le mat6riel en deuxieme etalon international de polymyxine B et propos6 que l'unit6 internationale de polymyxine B soit definie comme l'activite de 0,000119 mg du deuxieme etalon internatio- nal. Cette valeur, correspondant a une activit6 de 8403 Ur/mg, a et6 acceptee par le Comite OMS d'experts de la Standardisation biologique (1970) lorsqu'il a defini l'unite internationale de polymyxine B par rapport au deuxieme etalon international. REFERENCES Bliss, C. 1. (1952) The statistics of bioassay with special reference to vitamins, New York, Academic Press, pp. 580-582 Humphrey, J. H. et al. (1953) Bull. Wld Hlth Org., 9, 15-28 Humphrey, J. H. et al. (1959) Bull. Wld Hlth Org., 20, 1229-1232 WHO Expert Committee on Biological Standardization (1969) Wid Hlth Org. techn. Rep. Ser., No. 413, p. 13 WHO Expert Committee on Biological Standardization (1970) Wid Hlth Org. techn. Rep. Ser., No. 444, p. 8 Annex PARTICIPATING LABORATORIES National Biological Standards Laboratory Department of Health Canberra, Australia (Dr L. F. Dodson) Burroughs Wellcome & Co., Ltd Dartford, Kent England (Dr G. A. Stewart) Division of Biological Standards National Institute for Medical Research London, England (Mr J. W. Lightbown & Mr P. Isaacson) Department of Microbiology Research Institute for Pharmaceutical Chemistry Budapest, Hungary (Dr I. Horvath) Food and Drug Administration Department of Health, Education, and Welfare Washington, D.C., USA (Dr W. W. Wright)