Neomycin in Serum. through granulation tissue. This paper reports only. the results of the neomycin assay.

Transcription

1 APPLED MICROBIOLOGY, Sept. 1969, p Vol. 18, No. 3 Copyright 1969 American Society for Microbiology Printed in U.S.A. Large-Plate Method for the Assay of Neomycin in Serum A. A. van SOESTBERGEN Department of Pathology, College of Medicine, Ohio State University, Columbus, Ohio 4321 Received for publication 14 April 1969 A large-plate method employing radial diffusion from small paper discs for assaying serum levels of neomycin is described. More than 12 discs placed on a loading plate and loaded with 2 Mliters of sample could all be brought into contact with the agar plate at one time. The requirement for elaborate statistical design to compensate for time-dependent bias was thus eliminated. The dose-response curve was linear for a range of at least.5 to 2.5 jag/ml. The experimental limits for the actual zone width (distance from the edge of the paper disc to the outer edge of the inhibition zone) for 12 repetitions of the assay of neomycin in one and the same serum, carried out simultaneously on one large plate, were about i 6% (95% confidence interval). The 95% confidence interval for the distribution of difference between duplicate zones obtained for the assay of neomycin in 34 different sera, also carried out on one plate, was about i 1%. The dose-response lines for a standard and for three unknown sera, when carried out together on one plate, were parallel within the variability (± SD one) of the zone width. The large-plate method is considered to be more efficient than the use of the smaller petri dishes. The method is suitable for the assay of penicillin in serum and can most probably be used for the assay of a wide variety of substances for which radial diffusion from paper discs into agar is feasible. Diffusion methods to assay the level of an antibiotic in serum are based upon radial diffusion of the antibiotic from paper discs or other containers into the surrounding agar. Diffusion methods are preferred above other methods whenever possible because of their degree of precision, their relatively wide range, and the speed and simplicity with which they can be carried out The use of small (1 mm in diameter) petri dishes is indicated for the assay of only one or two sera. The large-plate method is the method of choice for a greater number of assays. The large-plate method is accurate, very efficient, and simple to perform, and it also removes the variable of depth of agar which occurs when the assays have to be carried out in several small petri dishes (8). Complicated statistical designs to compensate for bias introduced by the time difference between application of the first and last samples on the plate are not required if the discs are applied all at one time (6). This can be done through the use of a loading plate. MATERIALS AND METHODS Serum samples. Neomycin content was measured in 34 samples obtained from four cats used in an investigation of the rate of absorption of neomycin through granulation tissue. This paper reports only the results of the neomycin assay. Standard. Neomycin sulfate (Mycifradin sulfate, The Upjohn Company, Kalamazoo, Mich.) was dissolved in sterile cat serum (Colorado Serum Company, Denver, Colo.) to a concentration of 1 mg/ml. The solution was kept stable for several days by means of refrigeration. Small samples dispensed in 1-ml ampoules which were sealed and stored at -3 C retained their potency for at least several months. Dilutions containing 2.5, 2., 1.5, 1., and.5 jug/ml were made in serum. Indicator organism. The test organism was Staphylococcus epidermidis (ATCC 12228). The stock culture was stored at -3 C in 3% rabbit serum in Trypticase Soy broth. A suspension of the organisms was obtained by washing the overnight growth from a slant of stock culture agar (Difco) with 5 ml of sterile.85% saline. The turbidity of the suspension was adjusted by diluting with saline to an optical density between.2 and.25 at a wavelength of 61 nm; the measurement was made in a 1-cm cell by means of a Spectronic 6 spectrophotometer (Bausch & Lomb, Inc., Rochester, N.Y.). The suspension then contained about.5 X 18 colony-forming cells per ml. The seed agar, after being cooled to 48 C, was inoculated with a.1 dilution of this suspension. Assay plates and agar. A two-layer agar system was used. The "base" agar layer, consisting of 3 ml of 35

2 VOL. 18, 1969 ASSAY OF I' Penassay Seed Agar (Difco) adjusted to ph 7.9 to 8. with NaOH, was poured into a carefully leveled, large 2.84-liter Pyrex baking dish (outside dimensions, 39 by 23 by 5 cm). After this agar layer had set, a "seed" agar consisting of 14 ml of the same agar inoculated with a 1.4-ml suspension of indicator organisms was poured on top of the base layer. When the seed agar had set, the plates were placed in the refrigerator for at least 1 hr. Loading prcedure, incubation, and reading of the assay. Filter-paper discs (6.35 mm, 74 E, Schleicher & Schuell Co., Keene, N.H.) were measured and found to vary less than.5 mm from an average diameter of about 6.4 mm. More than 14 discs could be placed on the agar all at one time by the following procedure. A glass plate, later a perspex plate,.64 cm thick and somewhat smaller than the inside dimensions of the Pyrex dish, was covered with a layer of parafilm (parafilm "M," American Can Co., Neenah, Wisc.) to obtain a water-repellant surface. The discs were placed on the parafilm surface according to a pattern indicated on a piece of graph paper underneath the plate. Dilutions of the standard or the sample were applied to the discs in volumes of 2 Milters by means of a capillary pipet (Becton, Dickinson, & Co., Rutherford, N.J.). When all the discs were loaded (loading time approximately 3 min for 12 discs), the agar plate was taken from the refrigerator, inverted, and moved into contact with the discs so that the discs could be placed on the agar surface. The Pyrex dish was then covered with Saran Wrap (The Dow Chemical Co., Midland, Mich.) and was placed in an incubator at 37 C. After approximately 18 hr of incubation, the distance from a point on the outer edge of the inhibition zone to a point diametrically opposite was measured to :i:.1 mm with vernier scale calipers. The size of this measurement is indicated as the "diameter" of the zone. RESLMTS Precision of the assay. The diameter of the zone is actually twice the distance from the edge of the paper disc to the outer edge of the inhibition zone (the zone "width") plus the diameter of the paper disc. The variation of the diameter of the paper disc is small compared to the overall variation of the zone diameter. However, to include the diameter of the paper disc in calculating the standard deviation of the zone diameter could result in a somewhat misleading estimate of the precision of the assay. Therefore, the present calculations on the precision of the assay are based upon the value of the zone width which was derived from the zone diameter by subtracting the diameter of the disc and by dividing by two. To determine experimentally the precision with which measurements of the zone width could be obtained, 12 discs were loaded with identical samples containing approximately 1.5 1Ag of neomycin per ml of serum and were placed on IIEOMYCIN 351 the agar. The diameters of the inhibition zones were measured the next day, and the zone widths were calculated. Arranged in 11 classes, the zone widths were normally distributed (linear distribution on a probit scale; Fig. 1) with chi square = 1.68 and P =.38 for 1 degrees of freedom. The mean of the distribution was 3.5 mm and the standard deviation was.11 mm. Thus, a zone width was determined with a 95 % confidence interval of -6.2%. This level of precision was obtained repeatedly in consecutive assays of this type. A measure of the precision of the measurements when carried out on one plate was also obtained ZONE (mmn) FIG. 1. Probit distribution of zone widths obtained in 12 assays of neomycin content of one serum carried out simultaneously on one large plate o DIFFERENCE IN ZONE WIDTH (mm) FIG. 2. Probit distribution of differences between zone widths obtained in duplicate assays of neomycin content of 34 sera containing various amounts of neomycin. The assays were carried out simultaneously on one large plate.

3 352 VAN SOESTBERGEN APPL. MICROBIOL. as the distribution of differences between duplicate measurements of each of 34 samples of sera containing different amounts of neomycin. These differences were normally distributed (linear distribution on a probit scale; Fig. 2) with chi square = 16.8 and P =.3 at 14 degrees of freedom. The mean of the distribution was zero and the standard deviation was.2 mm. A zone width of, for instance, 3.5 mm could thus be determined with a 95% confidence of 1.4%. A third measure of the precision of the assay E z 42.5j, ZONE (mm) FIG. 3. Relationship between the mean zone width and the standard deviation obtained for five repetitions of assays of various amounts of neomycin present in 2 sera. The assays were carried out simultaneously on one large plate. 3 : o as 4, T -L LOG DOSE FIG. 4. Relationship between inhibition zone and neomycin content (,ug/ml) of a standard (S) and the inhibition zone and dilution of a test serum (T). The plotted points represent the values obtained witk-five assays carried out simultaneously on one large plate. 9 was made by making five dilutions of the standard and of three sera containing unknown quantities of neomycin and by measuring the zone diameters around discs loaded with 2 Mliters of each dilution. Five discs were taken for each dilution and the mean and standard deviations were calculated for the zone widths. The 2 mean values thus obtained were not correlated to a significant degree with the size of the standard deviation (Fig. 3). The average value of the standard deviation was.17 mm. Validity of assay. Although in bioassay, using agar diffusion techniques, linearity is usually satisfied by plotting log concentration or dilution against the square of the zone width (3), exceptions frequently occur. For the assay of erythromycin in serum, linearity was found for log concentration on the diameter of the zone (1). Other investigators, using agar-well diffusion, obtained a curvilinear relationship between zone diameters and antibiotic concentrations (2). For the present assay, the relationship of the width of the inhibition zone to the log of the neomycin concentration in the standard and in a test serum appeared to be linear (Fig. 4). The data are tabulated in Table 1. The analysis of variance was completed as shown in Table 2. A detailed description of the statistical technique used to analyze the present data is given by Finney (4) Ṫhe assumption of linear regression obtained by studying the dose-response diagram plotted in Fig. 1 is clearly justified by a mean square for deviations from linearity (indicated as Linearity in Table 2) not significantly greater than the mean square for error within doses. The value of 1.53 for the variance ratio is well below the 5% significance level. In regard to differences between preparations (the standard and the test serum) the assay appears well planned. A variance ratio test for the mean square of the difference between preparations to the mean square within doses is 2.32, which is also well below the 5% significance level of 4.2 expected for 1 and 4 degrees of freedom. The responses to the dose in the test serum lie within the range of responses to the standard. Table 2 shows no evidence of deviation from parallelism. A large mean square for this component would indicate invalidity of the assay, and the assumption that the test preparation and the standard contained identical material would have been false. In that case, the data would not have been of any use for the assay. The mean square for regression is, of course, large, confirming the impression that the doseresponse relationship is suitable for the estima-

4 VOL. 18, 1969 ASSAY OF NEOMYCIN 353 Value determined Log 2yij 511 TABLE Width of inhibition zones around discs in an assay of serum for neomycin Standard, neomycin concn (jug/ml) Dilutions of test serum TABLE 2. Analysis of variance for the data of Table I Nature of Degofees Sum of Mean Variance variation freedom squares square ratio Preparations a Regression Parallelism a Linearity a Between doses Within doses Total anot significant. b P <.1. tion of neomycin concentration in a serum. Further evidence of validity was obtained when a Bartlett's test for homogeneity of variances indicated no departure from homogeneity. The primary object of the assay is to obtain a valid estimate of the concentration of neomycin in the test serum. In the present investigation, the concentration is assessed relative to a standard, and a numerical assessment of the precision of the estimate of the concentration must be given. With a statistical treatment of the data as carried out for the present assay, these requirements are met. Let p denote the estimate of the potency (concentration of neomycin in the test serum relative to the standard). A value of loglop = Thus, p is equal to 1.41, meaning that one unit of the test serum is found to give a response equal to 1.41 units of the standard. Through the use of a variance formula (4) the fiducial limits of this estimate can be calculated; a 95% interval was found to be pi = 9.87 and Pu = The potency of the test serum had thus been assayed with a precision equal to ±5.4%. DISCUSSION The commercially available Pyrex baking dish has been recommended by other investigators (5) as a container for the agar. The bottoms of the dishes were shown to be sufficiently flat by placing a flat piece of plate glass on the bottom and running a little water between the dish and the glass plate. The dishes were sterilized by autoclaving. With the large-plate method, duplicate assays of each of 34 sera could be carried out all at one time on one agar plate. A sufficient number of dilutions (five) of the standard, with three discs per dilution, were included in every assay. Because only about 25 % of the zones are used for standards, the method is considered appreciably more efficient than assays carried out on small petri dishes in which the standard zones may comprise 5 to 9% of the total. Difference may occur in protein binding of antibiotics between sera of different species and, although this did not appear to be the case for neomycin in human or cat serum, cat serum was used throughout the present investigation as a diluent for the standard. (Dilution of the standard in human or cat serum gave regression lines with the same slopes and values for the mean.) Because of the small size of the sample required to load the discs (2 Mliters), repetition of the assay usually is feasible. It is also possible to assay neomycin in small samples of blood obtained, for example, from experimental animals or from humans by skin puncture. The method is sensitive enough to measure concentrations of neomycin in serum from.2 to about 2.5,g/ml without prior

5 354 VAN SOESTBERGEN APPL. MICROBIOL. dilution of the serum. Therapeutically effective levels of neomycin in serum can thus be detected. If the serum contains more than 2.5,ug/ml, a single dilution of one in two, and sometimes of one in four or one in eight, must be tested. However, this is of no particular inconvenience and does not decrease the effectiveness of the method. Although several attempts were made to obtain a water-repellant surface on the glass loading plate by covering its surface with silicon, as has been recommended (6), the use of parafilm proved to be superior in our hands. For smaller plastic petri dishes, however, the use of a silicon spray (Dricote D-155, Fisher Scientific Co., Pittsburgh, Pa.) appears to be the method of choice. The precision attained with the use of disposable micropipets appeared to be less dependent upon the technical expertise of the operator than did methods involving larger pipets, capillary action, or dipping the discs into serum. The precision compares favorably with the precision of a micromethod carried out on small petri dishes, as recently described for the assay of colicins (7) and for the assay of serum levels of erythromycin (1) Ḟor the present investigation, the standard deviation of the distribution of the measurements was calculated on the actual width of the inhibition zone (distance from the edge of the paper disc to the outer edge of the inhibition zone). If the calculations are based upon the diameter of the zone (distance from a point on the outer edge of the inhibition zone to a point diametrically opposite), the 95% confidence interval of 1.2% obtained for the measurements referred to in Fig. 4 is reduced to 4.2% and that for the measurements in Fig. 1 from 6.1 to 3.5%. We have shown our version of the large-plate method to be applicable to the assay of penicillin in serum, and it probably can be used in most assays for which diffusion from filter papers is feasible. ACKNOWLEDGMENTS I thank Darrol W. Heggen, Richard R. Lanese, and Martin D. Keller, Department of Preventive Medicine, and D. Ransom Whitney, Department of Mathematics, for helpful discussions regarding statistical treatment of the data and for assistance with the calculations. David R. Kelly, Department of Otolaryngology, supplied the cat sera. Ching ho Lee rendered valuable technical assistance. LITERATURE CITED 1. Bell, S. C., J. W. Hamman, and W. E. Grundy Micromethod for assaying serum levels of erythromycin. Appl. Microbiol., 17: Bennett, J. V., J. L. Brodie, E. J. Benner, and W. M. M. Kirby Simplified, accurate method for antibiotic assay of clinical specimens. Appl. Microbiol. 14: Cooper, K. E The theory of antibiotic inhibition zones, p. 74. In F. Kavanagh (ed.), Analytical microbiology. 4. Finney, D. J., Statistical method in biological assay, 2nd ed. Hafner Publishing Co., New York. 5. Gerke, J. R., J. D. Levin, and J. F. Pagano Antibiotic substances, p In F. Kavanagh (ed.), Analytical microbiology. 6. Kavanagh, F., Cephalosporin C, p In F. Kavanagh (ed.), Analytical microbiology. Academic Press Inc., New York. 7. Richardson, H., A. H. Emslie-Smith, and B. W. Senior Agar diffusion method for the assay of colicins. Appl. Microbiol. 16: Simpson, T. S Microbiological assay using large-plate methods, p. 93. In F. Kavanagh (ed.), Analytical microbiology.