will be shown, these serum levels were usually 2 to 5 times higher than those

Transcription

1 THE EFFECTIVE CONCENTRATIONS OF PENICILLIN IN VITRO AND IN VIVO FOR STREPTOCOCCI, PNEUMOCOCCI, AND TREPONEMA PALLIDUM HARRY EAGLE, RALPH FLEISCHMAN, AND ARLYNE D. MUSSELMAN Section on Experimental Therapeutics, National Institutes of Health, Bethesda 14, Maryland Received for publication November 2, 1949 In the therapeutic use of penicillin, the question as to whether the minimum effective concentration in vivo is the same, greater, or less than that necessary to kill the same organ'sm in vitro is of obvious importance. Data are here reported with respect to the minimal effective serum concentrations of penicillin in a number of experimental infections in mice and rabbits. As will be shown, these serum levels were usually 2 to 5 times higher than those necessary to kill the same organism in vitro. However, in view of the known concentration differential between the serum and tissue fluids, this suggests that the effective concentration of penicillin at the actual site of infection is in most instances the same as that effective in vitro. METHODS AND MATERIALS The use of procaine penicillin to provide reasonably stable serum concentrations of penicillin. In order to obtain a reliable measure of the effective serum concentrations of penicillin, it was necessary to keep these levels reasonably constant for the time that the drug was acting on the organisms. This was accomplished by the use of procaine penicillin G suspended in peanut oil, gelled by the addition of 2 per cent aluminum monostearate (Buckwalter and Dickison, 1948). A series of such suspensions were prepared, varying in penicillin content in approximately 2-fold steps from 320,000 down to 80 units per ml. The co-operation of the staff of the Bristol Laboratories, by whom these penicillin suspensions were prepared, is gratefully acknowledged. The various suspensions were injected intramuscularly in a fixed volume of 0.1 ml in mice weighing 17.5 to 22.5 grams. In rabbits (2.5 to 3.5 kg) the dosage volume in ml was one-fifth of the body weight in kilograms. The serum concentrations of penicillin at varying intervals thereafter are summarized in table 1. The relatively slow rate at which the penicillin concentration fell, reflecting its slow and continuing absorption from the intramuscular depot, is evident from the table. For purposes of interpolation, lines were fitted to the experimental data by the least squares method. Penicillin assays. Serum penicillin assays were carried out by a serial dilution technique previously described (Eagle and Newman, 1947). Ten-ml specimens were obtained from rabbits by cardiac puncture, and mice were exsanguinated from the exposed heart. Horse blood was used as the hemolytic indicator in assaying the mouse sera, and rabbit blood for the rabbit sera. The results shown in the tables have been corrected for the inhibitory effect of serum in the assay (Eagle and Tucker, 1948). These corrective factors in the case of mouse serum were found to be 1.9 for whole serum, 1.3 for 1:2 serum, and 1.15 for 1:4 serum. 625

2 626 H. EAGLE, R. FLEISCHMAN, AND A. D. MUSSELMAN [VOL c. *e H + Cq CZ m m C14 0) 00Q H, WC! 1) 0 *ew ".Z --b.1e WC!S 0Dt -H-H-H- *4- * $ o 04 Pi 0w A 41 0 ps co c -H -H-H -H -H-HI 001km 10 t- e cli cq co 00.)-400 oo5 - oo 000 la -H -H -H -H -H-H -H 'o o; Co _ o ; oo; la +C00._ QOO oo5-h 00 -HH U: o Cooo C-00 0Q0 00 CS4 _4 _ o q _6 _ CO CD C_ C- CO 0~ ee_o o ooo oooo o0 o U C O 0 O O 1000 C;C; Ce n000 0 co etr- - -H4i-H-N-H!'-R~ 10 9C9 o co -o C0 C COC 9 9 blo 0 A U-i co t- = Q..d N co m U-. t: C C IC1,5 I cq -,.- ll. cl. Cs cl ;, Ci' -. e

3 1950] CONCENTRATIONS OF PENICILLIN 627 In order to make the dosages and concentrations comparable to those reported in a previous paper dealing with sodium penicillin, they have been expressed as mg per ml or mg per kg of sodium penicillin G, assuming an activity of 1,667 units per mg. One hour after the injection of the penicillin, the animals were inoculated with an actively growing culture of the several organisms, calibrated by microscopic count (Magnuson, Eagle, and Fleischman, 1948). The colony count on a poured plate averaged 1.1 times the microscopic count, with a range of 0.13 to 2.8. The smaller values usually reflected the presence of bacterial clumps or chains in the culture, the average number of organisms per chain varying from 2 to 7 in individual experiments. Determination ofminimum dose of penicillin with bactericidal action in vivo and of minimum effective serum concentrations. Two types of experiment were conducted. In one, a relatively small inoculum was used (100 organisms in the mice and 20 organisms in the rabbits), and death or survival served as the criterion of penicillin efficacy. The size of the inoculum was deliberately set at this low level so that the effective concentration of penicillin would have to be provided for only a short period of time in order to abort infection (Eagle, Fleischman, and Musselman, 1949b). Under these circumstances, the effective concentration of penicillin would be that present during the first few hours after its administration (see below). The abortive doses of penicillin (ED5o) indicated in the tables were calculated by Karber's procedure (1931). The assistance of Mr. Nathan Mantel, of the Office of the Statistical Coordinator, Division of Public Health Methods, Public Health Service, in calculating the EDi dosages and fitting lines to the experimental data is gratefully acknowledged. In the second type of experiment the bactericidal effect of various serum concentrations of penicillin was determined directly. One hour after the injection of varying doses of procaine penicillin 20,000 organisms were inoculated intramuscularly into mice. At intervals thereafter the muscle was removed and emulsified in a Waring blender with 50 ml of blood broth; the number of viable organisms was determined by plating out in blood agar (3 ml of the tissue emulsion in a total of 12 ml, and serial 40-fold dilutions from the first tube). All plate counts were made in duplicate. This technique was not feasible in syphilitic infection, since Treponema pallidum has not yet been cultivated in artificial media. Instead, rabbits were treated with penicillin 10 to 12 days after the intratesticular inoculation of approximately 4( million organisms and after the development of an acute orchitis. The immobilization and disappearance of organisms from the primary lesion after varying doses of procaine penicillin served as the criterion of the minimum treponemicidal concentration. EXPERIMENTAL RESULTS The Effective Serum Concentrations in a Number of Experimental Infections Judged by the Dosages of Procaine Penicillin Necessary to Abort an Early Infection Experiments to determine the dosages of procaine penicillin necessary to abort the several infections here studied are summarized in tables 2 and 3. As there

4 628 H. EAGLE, R. FLEISCHMAN, AND A. D. MUSSELMAN [VOL. 59 shown, the effective dosage of penicillin and, thus, the effective serum concentration were usually sharply defined. In the mice particularly a 2-fold difference in dosage often made the difference between cure rates of less than 20 per cent and more than 80 per cent. TABLE 2 The dosage of procaine penicillin G (in peanut oil gelled with 2 per cent aluminum monostearate) necessary to abort a number of experimental mouse infections Suspensions containing varying concentrations of penicillin were injected intramuscularly in a fixed volume of 0.1 ml into mice weighing 17.5 to 22.5 g. The dosages in mg per kg have been calculated on the basis of the sodium rather than the procaine salt (1,667 units per mg). The animals were inoculated 1 hour later. units/kg DOSAGE 25,000 12,500 8,700 6,250 4,400 3,200 2,200 1, mg/kg GROup A p-eemolytic STREPTOCOCCUS (100 ORGAMSMS INTRMUS- CUJLALLY) 0/20 0/20 9/20 18/20 19/20 GROUP B p-memolytic STREPTOCOCCUS (100 ORGANSMS INTRAMUS- CULARLY) 1/20 2/18 17/20 14/17 13/19 15/15 DIPLOCOCCUS PNEUMONIAE TYPE I (100 ORGAMSMS INTRAPERI- TONEALLY) 0/20 0/30 6/30 1/29 9/30 40/60 25/30 18/20 DIPLOCOCCUS PNEUMONIAE TYPEm m (100 ORGAMISS) Intonperi 2/30 3/30 4/30 3/30 16/30 24/30 Intramuscularly 2/24 0/25 5/25 18/25 22/25 25/25 Mortality in con /43 17/20 54/55 34/35 23/25 trol animals re /33 3/15 27/30 9/10 16/25 ceiving indicated 1 5/35 0/15 15/30 6/10 no. of organisms Dosage that abor- mg/ * 3.0 =1: 0.46t j 0.16$ ted infection in kg half the animals (ED5o) -~~~~~ * Natural survivorship of 10 per cent assumed in calculating ED5o dosage. t Natural survivorship of 15 per cent assumed in calculating ED50 dosage. t Natural survivorship of 5 per cent assumed in calculating ED60 dosage. The serum levels of penicillin after these EDso dosages may be taken as those that provide effective concentration at the site of inoculation. From the data of table 1 one could estimate that in the mice inoculated intraperitoneally with (a) group A streptococci, (b) type I pneumococci, and (c) type III pneumococci and intramuscularly with (d) group B streptococci and (e) type III pneumococci, the serum concentrations provided by the EDw dosages of procaine penicillin at the time of inoculation' were (a) 0.015, (b) 0.06, (c) 0.055, (d) 0.11, and (e) One hour after the injection of penicillin.

5 19]0 CONCENTRATIONS OF PENICILLIN 629 micrograms per ml, respectively (table 4, coluimn 2). Similarly, in rabbits inoculated with the group B streptococcus, the ED50 dosage had provided initial serum concentrations of 0.07 to 0.14 micrograms per ml, varying according to the site of inoculation. In all these cases, dosages of penicillin that initially provided approximately half these concentrations were almost wholly ineffective (table 4, columns 1 and 2). The minimal effective serum levels were thus only slightly less than those provided by the ED50 dose at the time of inoculation. TABLE 3 The dosage of procaine penicillin G (in peanut oil gelled with 2 per cent aluminum monostearate) necessary to abort infection in rabbits inoculated with group B 5-hemolytic streptococci (20 organisms) Suspensions containing varying concentrations of penicillin were injected intramuscularly in rabbits at a dosage of 0.5 to 0.7 ml bod. They were inoculated 1 hour later as indicated in the table. Dosages in mg per kg have been calculated as sodium penicillin G (1,667 units per mg). PENICILLIN DOSAGE urnis/kg mg/kg MORLTALTY IN RABBITS INOCULATED WITH GROUP B p-hemaolytic STREPTOCOCCI Intramuscularly Subcutaneously Intratesticularly Intrapuly monarily* 8, /6 4, /8 0/7 2, /8 0/8 0/6 1/6 1, /8 5/8 2/6 2/ /8 7/8 4/6 5/ /4 4/4 9/10 6/10 Mortality in untreated 12/12 9/10 7/7 5/7 control animals Dosage which aborted units/kg 1,300 1, infection in half of mg/kg 0.78 :1: :i i * animals * Fifteen per cent natural survivors assumed in calculating EDs0 dosage of penicillin; if one assumes 30 per cent natural survivors, the EDso value becomes There was reason to believe that, under the conditions of the present experiments, the infections had been effectively aborted by the ED6o dose of penicillin within less than 1 hour, and that the group A infection had been aborted somewhat faster than the other three. (1) In mice inoculated intraperitoneally or intramuscularly with 20 to 100 organisms but treated with aqueous sodium penicillin instead of the suspension of procaine penicillin in oil, the abortive doses in the case of group B streptococci, Diploous pneumoniae type I, and Diplococcus pneumoniae type III had previously been found to range between 3 and 10 mg per kg (Eagle, Fleischman, and Muselman, 1949b). These doses of aqueous penicillin provided serum concentrations in excess of 0.05 ug per ml, for 1 to 1.6

6 630 H. EAGLE, R. FLEISCHMAN, AND A. D. MUSSELMAN [VOL. 59 hours (Eagle, Fleischman, and Musselman, 1949a, 1950). The abortive dose of 0.35 mg per kg in a similar group A streptococcal infection provided that concentration for approximately 0.5 hours and a concentration of 0.012,gg per ml for 0.82 hours. (2) When these organisms were exposed to effective concentrations of penicillin for a period of 15 to 45 minutes, the then surviving bacteria did not resume multiplication for a number of hours. During this recovery period varying TABLE 4 The minimal effective serum concentration of penicillin in a number of experimental infections (based on the dosage of procaine penicillin necessary to abort early infections) SERUM CONCS. OF (O1A)G PENICILLIN (JG/ML) (4) NO. OF DOSAGE PROVIDED BY EDio ESTDIATED ORGAN- THAT DOSAGE OF PENICILLIN MNINMUM ORGAMSM ORGANISM HOST ~ ROUTE OP sm ABORTED INOCULATION FET INOCU INFECTION SERETIV LNATE IN HALF (2) SERUM LATD E At time of (3)ou zconc.no ANIMALS* inocula- 1lhourtPNCILN tiont itr mg/kg pg/mi,-hemolytic Mice Intramuscular : streptococcus, group A p-hemolytic Mice Intramuscular streptococcus, Rabbits Intramuscular group B Subcutaneous Intratesticular Intrapulmonary Diplococcus Mice Intraperitoneal pneumoniae, type I Diplococcus Mice Intraperitoneal pneumoniae, Intramuscular type III * From data of tables 2 and 3. All dosages and concentrations are expressed in terms of the sodium rather than the procaine salt. From data of table 1. t Based on an extrapolation of the serum penicillin curves beyond the measurable range. Although the effective concentration for the type III organisms seemed to be somewhat less than for type I, the difference was not significant, and the effective concentration for both organisms has been rounded off to 0.05 Ag per ml. proportions of the damaged organisms were disposed of by host mechanisms, in the absence of drug (Eagle, 1949; Eagle and Musselman, 1949; Eagle, Fleischman, and Musselman, 1950b). It thus appeared that, under the conditions of the present experiments, the EDi dose of procaine penicillin had probably completed its abortive action within approximately 1 hour. In that case, the minimum effective serum level would be slightly less than the concentration provided by the EDo dose at the moment of inoculation and slightly greater than that present an hour after inoculation. The

7 1950] CONCENTRATIONS OF PENICILLIN 631 serum concentrations at those two times are given in columns 2 and 3 of table 4 for each of the infections here studied; and the last column of that table lists an intermediate value, which is the estimated minimum effective concentration of serum for each of those infections. Because of the slow rate of fall of the serum penicillin levels, the error in this approximation is relatively small. In mice those effective levels were estimated at ug per ml for the group A streptococcus inoculated intramuscularly, 0.05 ug per ml for the type I or III pneumococcus inoculated intraperitoneally, 0.08,g per ml for the group B streptococcus inoculated intramuscularly, and 0.08,g per ml for the type III pneumococcus inoculated intramuscularly. In rabbits inoculated with the group B streptococcus the minimum effective serum concentration of penicillin was approximately 0.12 Mg per ml for an intramuscular inoculum, 0.1 Mug per ml for a subcutaneous or intrapulmonary inoculum, and 0.07 ug per ml for an intratesticular inoculum. These minimal effective serum concentrations are approximately 2 to 4 times those necessary to kill the same organisms in vitro and are estimated to have provided just the latter bactericidal concentration at the actual site of inoculation (see below). The Minimal Effective Serum Concentration of Penicillin as Determined by the Direct Measurement of Its Bactericidal Action in Vivo The data and conclusions of the preceding section were confirmed by experiments in which the bactericidal action of penicillin was studied directly. A relatively large number of organisms (20,000) were inoculated intramuscularly in mice, and the minimum dosage of sodium penicillin that sufficed to cause an initial reduction in the number of viable organisms was then determined. These experiments will be reported in detail elsewhere (Eagle, Fleischman, and Musselman, 1950b), but a single illustrative figure (figure 1) is reproduced here. With all four organisms tested (Diplococcus pneumoniae, type I and type III, and group A and group B,B-hemolytic streptococci) a dosage of 3 mg per kg, supplying a 30-minute serum level of 0.5 Mg per ml, was actively bactericidal. A dosage of 0.6 mg per kg, supplying a 30-minute level of 0.11 Mug per ml and an estimated 1-hour level of 0.01 Mg per ml, was actively bactericidal against the group A streptococcus (for which the effective serum concentration is estimated as Mg per ml); it was, however, only slightly active against the group B streptococcus and the pneumococci (requiring approximately 0.08,ug per ml). A dosage of 0.15 mg per kg, providing a 30-minute serum concentration of Mg per ml, was relatively ineffective against all four species. (A somewhat higher activity was perhaps to have been expected against the group A streptococcus.) With the latter organism the minimum dosage that had a bactericidal effect in vivo was also determined by means of a suspension of the procaine salt in oil. As shown in figure 3, a dosage of 0.24 mg per kg was wholly ineffective, and a dosage of 0.48 mg per kg was definitely bactericidal. Although the serum concentration provided by the latter dose is less than the measurable level, it may be estimated by extrapolation of the data in table 1 to be on the order of 0.01 to 0.02,ug per ml one-half hour after the injection. This agrees with the previous estimate of

8 632 H. EAGLE, R. FLEISCHMAN, AND A. D. MUSSELMAN [VOL ;g per ml as the minimum serum concentration effective against this organism. A special case was afforded by syphilitic infection. When syphilitic rabbits were given a single injection of procaine penicillin, the smallest dosage causing the temporary disappearance of the organisms from the primary lesion was 0.3 mg per kg (500 units per kg). At this threshold dosage the organisms began to decrease significantly in numbers only after 12 to 24 hours and disappeared altogether from the lesion after 48 to 60 hours. The mean serum concentration provided by this dosage was ug per ml after 1 hour, falling slowly to ug z10.000~~~~~~~~~~~~~~~~~~~~~~~~.1 0~~~~~~~~~~~~~~~~~~~~~0 > 1,000 O IG \,$~~~~~~'01 X z _000 to~~~~~~~~~~~~~ot '.eefet FNAPNIILI L.) D wl a:. ON TYP IN PNEUSEMOSCLECCUS HOURS AfTER INJECTION OF PENICILLIN Figure 1. The bactericidal action of penicillin on type III pneumococci in vivo (after Eagle, Fleischman, and Musselman, 1950b). The dotted portion of each curve indicates the time for which the indicated dose provided serum concentrations in excess of the bactericidal level. per ml in 24 hours. There was, however, reason to suspect that even smaller doses of penicillin, and correspondingly lowver serum concentrations might be effective if those levels were sustained for a sufficiently long period. This proved to be the case. As shown in table 5, when the injections were repeated every 12 hours, the minimal effective dosage of procaine penicillin per injection fell to mg per kg (64 units per kg), at which dosage the organisms began to disappear from the primary lesion only after 48 to 60 hours and after 3 to 4 injections. At this dosage of procaine penicillin in oil the serum never contains penicillin in concentrations demonstrable by the bioassay technique here used. An unreliable extrapolation of the data in table 2 indicates that the effective dosage of mg per kg might provide a serum concentration in excess of 0.01 pg per

9 1950] CONCENTRATIONS OF PEMCILLIN 633 ml for approximately 1 to 2 hours, and in excess of 0.001,ug per ml for perhaps 3 to 4 hours. That this rough approximation may be of the correct order of magnitude is indicated by the fact that, when syphilitic rabbits were treated with aqueous penicillin at 4-hour intervals to a total of 50 injections, the CDw dosage was mg per kg per injection (Eagle and Magnuson, 1946). This dosage also fails to provide a measurable level in the serum, but it may be estimated to have provided a 30-minute serum concentration of approximately 0.005,g per TABLE 5 The doeage of procaine penicillin G (in peanut oil with 2 per cent aluminum mono8tearate) nece8sary to affect T. pallidum in a rabbit testicular chancre Rabbits weighing 2.5 to 3.5 kg were inoculated intratesticularly with T. pallidum. Ten to 14 days later, at the height of the inflammatory reaction, the animals were injected intramuscularly with suspensions containing varying concentrations (80 to 320,000 units per ml) of procaine penicillin G. The dosage in ml was one-fifth the body weight in kg. The testes were aspirated for dark-field examination 3, 6, 9, 12, 24, 36, 48, 60, and 72 hours after treatment, or until they became dark-field-negative. Two to three rabbits were tested at each time period. PENICILLIN DOSAGE PER AVERAGE SERUM CONC. OF PENICILIN, HOURtS RlEQUIRED INJECTION og/l FOR SIGNICANT EDUCTION IN NO. units/lcg mg/kg 1 hr after injec- 4 hr after injec-. OR OTILTY OF units/kg ~~tion tion T. PALLMDUM Single injection 64, , , , , Injections repeated at 12-hour inter < vals cot * Concentrations and dosages are expressed in terms of sodium penicillin G rather than the procaine salt (cf. table 1). t No apparent effect on number or motility of organisms after 8 injections, 96 hours after beginning of treatment. ml; and the treponemicidal concentration provided by this dosage at the focus of infection may be estimated to be on the order of Ag per ml (see below). Treponema pallidum is thus at the same time one of the most sensitive as well as one of the most resistant of all bacteria to the action of penicillin. It is one of the most sensitive in terms of the smallest concentration which is bactericidal; it is one of the most resistant in terms of the time for which it must be exposed to that concentration in order to be killed. Syphilitic infection may therefore be cured by a relatively minute concentration acting over a relatively long period of time. Treponema pallidum differs from other bacteria in another im-

10 6H. EAGLE, R. FLEISCHMAN, AND A. D. MUSSELMAN 634 [VOL. 59 portant respect. With all other species studied so far the maximally effective concentration of penicillin is usually no more than 2 to 4 times as great as that which barely suffices to kill a significant proportion of the organisms (Eagle and Musselman, 1948), and only occasionally, as with the C-203 strain of group A streptococci, is the ratio an 8-fold one. In the case of Treponema pallidum, however, there was a more than 50-fold ratio between the concentration that was most rapidly effective and that which was demonstrably, if slowly, treponemicidal. The maximal effect was produced by a dose of 2.4 mg per kg, which provided an initial serum concentration of 0.36 Ag per ml (0.6 units per ml). At this dosage the organisms disappeared from the primary lesion in 6 to 9 hours. At the other extreme, at a dosage of mg per kg, which provided an initial serum concentration below the measurable level, but estimated to be on the order of to 0.01 jig per ml (0.008 to units per ml), the organisms disappeared in 48 to 96 hours. This means that the rate of its treponemicidal action in vivo increases progressively over a 50-fold range of serum concentration. Although low doses providing concentrations below the measurable level suffice to effect cure, larger doses would be more rapidly effective. The Correlation between the Effective Conceration of Penicllin in Vitro and in Vivo When bacteria are exposed to penicillin in vitro, one may define four "effective" concentrations of the drug. (1) At the lowest level the drug serves only to reduce the rate of multiplication but does not kill a measurable proportion of the organisms. (2) A slightly higher concentration is bactericidal for the susceptible organisms but does not kill the more resistant variants in the culture and therefore fails to sterilize the suspension. (3) At a somewhat higher concentration the drug kills all the organisms in a given suspension. (The "sensitivity" of a bacterial strain as ordinarily defined, i.e., the concentration at which a given inoculum fails to grow out visibly in culture, lies between concentration levels 2 and 3 as here defined. It is not a fixed value, but will vary in magnitude according to the size of the inoculum and the conditions of the test). (4) Beyond the level that suffices to kill all the organisms still higher concentrations serve to increase the rate of that bactericidal action. Eventually, however, one attains a concentration at which organisms are killed at a maximal rate and are not affected by further increase. The latter three "effective" levels in vitro are given in table 6 for each of the four bacterial species under consideration. With all four species the serum concentration that sufficed to abort an early infection (table 6, column 1) was regularly 1.5 to 3 times the concentration that sufficed to sterilize suspensions containing 106 organisms per ml (column 3) and 2 to 5 times as great as the concentration that killed the majority of the organisms in a culture suspension (column 2).2 The fact that approximately half the serum penicillin is bound to serum protein and thus not available for diffusion into the tissues (Tompsett, Shultz, and 2 Since only small inocula were used in the present experiments, the latter is the more appropriate comparison. The minimal effective concentrations in vivo as here determined are therefore not those that kill the organisms at the maximal rate but merely those that have a definite bactericidal action.

11 1950] CONCENTRATIONS OF PENICILLIN 635 McDermott, 1947a,b) would account for most of the differences observed between the minimum effective serum concentrations and those barely effective in vitro. In addition, there is probably a concentration gradient between the diffusible penicillin in the serum and the concentration at the actual focus of infection in the tissues.8 One must conclude that the effective concentration of penicillin in vivo, i.e, the concentration necessary to kill an organism in the tissue fluids, is essentially the same as that effective against that same organism in vitro. TABLE 6 Correlations between the effective concentrations of penicillin in vivo and in vitro EFFECTVE CONCENTRATION IN VITRO, /AG/ML ESTIMATED (2) MINIMAL Lethal for (3 ORGANISM HOST ROUTE OF INOCULATION EFFECTIVE most Suffices to (4) SERUM sterilize Kils organ- sms suspensin sterilizet 106/m1* irgn CONC. PENICI[L.LrN* but fails Cotainin= aia to up to ratet B-Hemolytic strep- Mice Intramuscular tococcus, group A B6-Hemolytic strep- Mice Intramuscular 0.08 tococcus, group B Intramuscular 0.12 Rabbits Subcutaneous Intratesticular 0.07 Intraperitoneal 0.1 Diplococcus pneu- Mice Intraperitoneal moniae, type I Diplococcus pneu- Mice Intraperitoneal moniae, type III Intramuscular 0.08 * From data of table 4: serum concentrations afforded by dose of penicillin that aborted infection in half of animals. All dosages and concentrations are expressed in terms of the sodium (1,667 units per mg) rather than the procaine salt. t From Eagle and Musselman, 1948; Eagle, Fleischman, and Musselman, These results suggest that in the therapeutic use of penicillin, a dosage regimen which provides at the focus of infection that concentration which kills the organism at a maximal rate in vitro would be similarly effective in vivo. The 3The widely varying concentrations of penicillin that have been demonstrated in different organs (e.g., Cooke and Goldring, 1945; Cutting et al., 1945; Richardson et al., 1946; Struble and Bellows, 1944) offer no direct evidence on the actual concentration of the drug in the extracellular tissue fluids. The magnitude of that gradient between serum and tissue fluids therefore remains to be determined. In the present experiment the ED50 dosage, and thus the effective serum concentrations, were the same in rabbits inoculated subcutaneously and into the lung, somewhat higher in those inoculated intramuscularly, and somewhat lower in those inoculated in the testis. In mice also an intramuscular inoculum required a higher serum concentration than an intraperitoneal inoculum with the same organism. This suggests corresponding minor differences in the concentration of penicillin in the tissue fluids of these organs.

12 63600 H. EAGLE, R. FLEISCHMAN, AND A. D. MUSSELMAN [vol. 59 serum concentration necessary to provide that effective level at the focus will vary with the tissue involved, the rate of penicillin diffusion into the involved area, possible local destruction of penicillin, and similar factors. Early vs. established infections. The experiments of the preceding sections relate only to freshly inoculated animals, and a different situation might conceivably obtain after theinfection has become established, with widely disseminated foci of inflammation and much larger numbers of bacteria. Such differences have, however, not been encountered. It is true that when mice or rabbits were treated a number of hours after inoculation with either pneumococci or streptococci, when rabbits were treated 6 weeks after inoculation with T. pallidum, the single curative dose (CDwo) of aqueous penicillin was as much as 1,000 times greater than when they were treated immediately after inoculation;4 (Eagle, Fleischman, and Musselman, 1949b). However, When treatment was then continued over a sufficiently long period of time, the curative dose per injection was of the same order of magnitude as that which sufficed to abort infection with a small number of organisms (Eagle, Fleischman, and Musselman, 1950). Clearly, the established infection did not require higher concentrations of penicillin but merely the provision of the same effective level for a longer period of time.' The 2- to 5-fold ratio here noted between the serum level of penicillin that sufficed to provide an effective concentration at a tissue focus and the minimal bactericidal concentration in vitro clearly indicates that, under the conditions of the present experiments, the penetration of penicillin into the bacterial depot offered no problem. A larger excess in the serum would, however, be necessary in order to provide the effective concentration in an area in which the penicillin was being inactivated. Further, if the bacteria were in a walled-off densely fibrotic or necrotic area, only poorly supplied with blood, the diffusion of penicillin into the bacterial focus from the surrounding capillary bed and tissue fluids might be greatly retarded. Under the latter conditions, the effective concentration would have to be maintained in the serum for a relatively long period of time in order to assure eventual penetration into the focus of infection, and a high concentration in the serum and tissue fluids would accelerate the diffusion of the drug into that focus. These considerations, however, do not affect the validity of the thesis that the concentration of penicillin at the bacterial focus which kills the organisms in vivo is of the same order of magnitude as that which is effective against the same organism in vitro. Supposed peristence of peniciuin in the tissues. This conclusion is apparently at variance with the repeated observation that patients can be effectively treated with penicillin even if the serum concentration falls below effective or even measurable levels for a number of hours between injections (cf. Tillett, McCor- 4 With the group A i3-hemolytic streptococcal infections in mice and the group B infections in rabbits, it was not possible to cure a 24-hour infection with any single dose of aqueous penicillin short of the toxic level (Eagle, Fleischman, and Musselman, 1949b); and divided doses (4 times at 3- or 24-hour intervals) were similarly ineffective. This failure was not related to the magnitude of the penicillin levels, and its cause is under present study. 5 See footnote 4. or

13 1950] CONCENTRATIONS OF PENICILLIN 637 mack, and Cambier, 1945; Tompsett, Timpanelli, Goldstein, and McDermott, 1949). Further, in experimental infections bacteria may continue to die for hours after the serum penicillin has fallen below the concentration effective in vitro (Jawetz, 1946; Grunberg, Schnitzer, and Unger, 1948; Eagle, Fleischman, and Musselman, 1950b). Of the four bacterial species under present study the group B streptococcus in particular was characterized by the continuing death I I I.I I I 100- SERUM CONCENTRATION OF PENICILLIN G IN MICE AFTER INJECTION OF 200 MG /KG so MORTALITY IN MICE INOCULATED INTRAMUSCULARLY WITH 100 TYPE z 60 - m PNEUMOCOCCI AT INDICATED v TIME AFTER PENICILLIN I&J o TIME IN HOURS Figure 2. The time for which a single injection of aqueous penicillin serves to protect mice against a following small inoculum. Mice weighing 17.5 to 22.5 g were injected intramuscularly with 0.1 ml of 4 per cent aqueous penicillin G (200 mg per kg). At varying intervals thereafter they were inoculated into the opposite leg muscle with 0.1 ml of a bacterial suspension diluted to contain 100 Diplococcus pneumoniae type III (by microscopic count). The upper curve in the figure is the mean serum concentration of penicillin after this dosage, as previously reported (Eagle, Fleischman, and Musselman, 1949a). The lower curve shows the proportion of animals that died during the period of observation. Each experimental group comprised 25 to 37 animals, and the mortality in untreated controls was 96 per cent. of the organisms for as long as 24 to 48 hours after penicillin was no longer demonstrable in the serum. The simplest explanation for this paradox would be that penicillin persists at effective levels in the tissues for hours after it has disappeared from the blood (cf. Ercoli, Lewis, Schwartz, and Whitehead, 1948). However, this has not been found to be the case under the conditions of the present experiments. When mice were injected intramuscularly with massive doses of aqueous penicillin, which provided initial levels of several hundred ug per ml, the animals became suscep-

14 638 H. EAGLE, R. FLEISCHMAN, AND A. D. MUSSELMAN [VOL. 59 tible to a minimal inoculum of a pathogenic organism even before the serum concentration had fallen below the effective level (figure 2). Similar results have been obtained in rabbits inoculated with a group B i3-hemolytic streptococcus (table 7). As the table shows, 4 hours after the intramuscular injection of penicillin G at 60 mg per kg, when the serum concentration still averaged 0.6 pg per ml, an intramuscular inoculum of 20 organisms killed 62 per cent of the animals, and after 6 hours, when the serum level averaged 0.06,ug per ml, the inoculum was regularly fatal. Since as little as 0.25 mg per kg sufficed to abort TABLE 7 The time for which sodium penicillin G persists at effective level8 in rabbit muscle Rabbits were injected intramuscularly with sodium penicillin G at either 60 or 3 mg per kg. At varying intervals thereafter they were inoculated into the opposite leg muscle with 0.5 ml of a suspension of group B fi-hemolytic streptococci calibrated by microscopic count to contain 20 organisms. The minimal protective dosage of penicillin when it was injected simultaneously with the organisms was 0.25 mg per kg; and it is a reasonable surmise that rabbits that died contained less than this amount of available penicillin in the body tissues at the time of inoculation. DOSAGE OF PENICILIN mg/kg TINE OF INOCULATION (HOURS AFTER INJECTION OF PENICILLIN) Average serum cone. of peni t t 0. 06t 60 Mortality in rabbits inocu- - 0/16 0/7 11/16 7/8 13/14 lated at indicated time Average serum conc. of peni ft Mortality in rabbits inocu- 0/7 3/7 7/8 6/6 lated at indicated time Conclusion: Penicillin did not persist at effective levels in rabbit muscle for a significant period after it had fallen below those levels in the serum. A small number of organisms caused a fatal infection when inoculated even before the serum penicillin had fallen to ineffective levels. * After Eagle, Fleischman, and Musselman, 1949a. t Estimated by interpolation. this inoculum if given at the time of the injection, it is apparent that only negligible amounts of penicillin were present in the tissues 4 to 6 hours after the injection, and that it did not persist at effective concentrations in the muscle for a significant period after it had fallen below these effective levels in the serum. The sensitivity to penicillin of bacteria damaged by previous exposure to the drug. A second possible explanation for the continuing death of bacteria in vivo might be that bacteria, once exposed to bactericidal concentrations, continue to die in the presence of much smaller concentrations, which are ordinarily wholly ineffective (cf. Grunberg, Unger, and Eldridge, 1948). Since the hourly decrease

15 1950] CONCENTRATIONS OF PENICILLIN 639 in the serum penicillin concentration after the intramuscular injection of aqueous solutions in mice averages more than 90 per cent, one would have to assume at least a 10-fold increase in susceptibility in order to explain the persistence of the bactericidal action for just 1 hour after the serum penicillin had fallen below the normally bactericidal level. This was not found to be the case. Although some bacterial species previously exposed to maximally effective concentrations of penicillin were thereby rendered somewhat more sensitive to low concentrations of the drug (cf. table 8), the differences were slight. Thus, when group A streptococci and type III pneumococci were exposed in vitro to 0.12,ug per ml TABLE 8 The susceptibility to sodium penicillin G of bacteria exposed to bactericidal concentrations of the drug in vitro and damaged but not yet killed by that exposure Bacterial cultures in the logarithmic phase of growth were diluted to contain 106 to 107 organisms per ml by microscopic count and exposed to 0.12,ug per ml of penicillin for the times indicated in the table. The suspensions were then plated out in blood agar containing varying concentrations of penicillin in order to determine the levels bactericidal for the surviving and damaged organisms. PRELrMINAILY SUSCEPTIBILITY OF SURVIVnG ORGANSMS TO PENICILLIN EXPOSURE TO penicillin Smallest conc. with definite Conc. necessary to kill BACTERIAL SPECIES bactericidal action, pg/ml 99t999%t % Penicillin- Penicillinbacteianl bacteria Hours hklled exposed Normal bacteria exposed Normal bacteria Streptococcus pyogenes (C- 1j ) Diplococcus pneumoniae, type III Staphylococcus aureus (Smith) Group B,-hemolytic strep tococcus (F-20) * Determined by plating out the reacting mixture in blood agar containing varying concentrations of penicillin (in the case of the group A streptococcus, e.g., 0.002, 0.003, 0.004, 0.006, 0.008, and micrograms per ml). t I.e., per cent of original suspension. long enough to kill approximately 99 per cent of the organisms, the concentrations then necessary to kill the surviving organisms were approximately half those necessary to kill normal organisms of the strain. (There is no information as to whether this difference represents an actual increase in sensitivity to penicillin or whether the organisms had irreversibly bound some penicillin during their initial exposure to the higher concentration.) With the strains of group B streptococcus and Staphylococcus aureus, however, the surviving organisms were not abnormally susceptible to penicillin, requiring just as high a concentration as normal untreated bacteria. In vivo also, when group A streptococci were exposed to relatively high con-

16 640 H. EAGLE, R. FLEISCHMAN, AND A. D. MUSSELMAN [VOL ,000 z A 10,000 V. A B 6Xo, w 0.24 mg/kg Z 1,000-1,000 0 <z o.j ;i- ( Z mo ano ADDITIONAL z TREATMENTg, _ 0.1 o!i mg/kg o g/kg 0 I,-g 0~~~~~~~~926Mg /kg12mgk 0 ~ ~ BCE /k A PENICILLIN-DAMAGED NORMAL BACTERIA BACTERIA 0.01 ic I I 0.01 "lw v I PRELIMINARY WITREATMENTS AQUEOUS HOURS AFTER ADMINISTRATION OF PROCAINE PENICILLIN WITH PENICILLIN Figure S. The susceptibility to penicillin of group A streptococci that have previously been exposed in vivo to bactericidal concentrations of the drug. Mice were inoculated intramuscularly with 200,000 group A streptococci (by microscopic count) and immediately treated with 1 mg per kg of aqueous penicillin G. One and one-half hours later, when approximately 83 per cent of the organisms had been killed and the serum penicillin had fallen to wholly ineffective levels, the mice were retreated with varying doses of procaine penicillin G (suspended in oil with 2 per cent aluminum monostearate), as indicated in section A of the figure. At intervals thereafter the inoculated muscle was removed from groups of 3 mice and emulsified in a Waring blender, and the number of surviving organisms was determined by plating out on blood agar containing 0.02 per cent "clarase" (Takamine) in order to inactivate the tissue penicillin. The points were plotted in the figure as the geometric means of the organisms surviving at each time period, referred to the number recovered from control animals immediately after inoculation as 100. The number so recovered initially was 1.8 times the number of colonies obtained by plating out the actual inoculum, the paradoxical increase reflecting the fact that small bacterial chains and clumps were partially broken up in the blender. Section B of the figure shows the results in a parallel experiment, in which mice inoculated with 20,000 organisms were immediately treated with varying doses of procaine penicillin, without preliminary exposure to penicillin. As shown by a comparison of sections A and B, there was no difference in the effective dose of procaine penicillin in the two experiments and, thus, no difference in the penicillin-sensitivities of normal bacteria and bacteria previously exposed to maximally effective concentrations of the drug. In both instances a dosage of 0.24 mg per kg had no significant effect, 0.48 mg per kg was definitely bactericidal, and dosages of 0.96 and 1.92 mg per kg were correspondingly more effective. Despite the small numbers of animals used and the large individual variation, the standard error of the logarithms of the means plotted in the figure varied only between 0.02 and 0.3 (X ) in the 1-hour groups, and (1 value excepted) and (X ) in the 2-hour groups. By the fourth hour, the differences between individual mice were much greater, and the standard error was correspondingly larger, to 0.82 (X ). m/kg

17 1950] CONCENTRATIONS OF PENICILLIN centrations of the drug, there was no indication that the surviving and damaged organisms had thereby been rendered more vulnerable to low concentrations. In the experiment of figure 3 more than 80 per cent of the organisms had been killed by a preliminary dose of sodium penicillin, and the survivors had been so damaged that even in the absence of further treatment they did not resume multiplication for a period of more than 4 hours. Nevertheless, as is shown in the figure, the same dosage of penicillin was necessary in order to continue the bactericidal action on the damaged organisms as was required to initiate it in normal, previously untreated bacteria. The continuing death of bacteria in vivo after the serum penicillin has fallen below the supposedly minimum effective level is therefore not due to the increased susceptibility of the surviving organisms to low concentrations of penicillin. The phenomenon is, however, adequately explained by the fact that, with some bacterial species, organisms that have been damaged but not killed by penicillin remain susceptible to the defense mechanisms of the host for a number of hours thereafter (Eagle, 1949; Eagle and Musselman, 1949). In this indirect manner the therapeutic activity of penicillin continues for a time after the drug itself has disappeared, until the bacteria have recovered sufficiently from the toxic effects of the drug to resume multiplication. SUMMARY AND CONCLUgIONS The abortive dose of penicillin was determined for a number of experimental infections in mice and rabbits, using the procaine salt suspended in oil with aluminum monostearate. Because of the relatively stable and reproducible serum concentrations observed with this preparation, one could thus estimate from the ED5o dose the minimum serum concentration that provided an effective level at the focus of infection. (a) In mice the minimum effective serum concentrations were approximately 0.012, 0.08, and 0.08 &g per ml for a group A streptococcus, a group B streptococcus, and a type III Diplococcus pneumonia, respectively, all inoculated intramuscularly. For both type I and type III Diplococcus pneumoniae inoculated intraperitoneally, the minimum effective serum concentration was 0.05,ug per ml. (b) In rabbits inoculated with the group B streptococcus intramuscularly, subcutaneously, intratesticularly, and directly into the lung, the minimal effective serum concentrations were approximately 0.12, 0.1, 0.07, and 0.1 yg per ml, respectively. (c) In the case of Treponema pallidum in rabbits the minim serum concentrations that supplied a treponemicidal level at a testicular lesion were estimated to be on the order of to 0.01 Mg per ml. These values were not significantly affected by the size of the inoculum or the age of the infection. With each of four organisms tested the smallest dose of aqueous sodium penicillin that had a bactericidal effect at an intramuscular focus provided serum concentrations of the same order as those provided by the minimum effective dose of procaine penicillin in oil. These minimum effective serum concentrations were regularly 2 to 5 times 641

18 642 H. EAGLE, R. FLEISCHMAN, AND A. D. MUSSELMAN [VOL. 59 the concentrations necessary to kill the same organisms in vitro. However, since approximately half of the serum penicillin is bound to the serum proteins and not immediately available for diffusion into the tissues, and since in addition there is probably a concentration gradient between the diffusible serum penicillin and the tissue fluids, one may conclude that the effective concentration of penicillin at the actual focus of infection is of the same order of magnitude as that which is operative in vitro. There was no evidence in these experiments that penicillin persists at effective levels in the tissues for a significant period after it has fallen below those levels in the serum. Further, when bacteria were exposed to lethal concentrations of penicillin in vivo or in vitro, the surviving and damaged organisms were not thereby sensitized to the drug. The concentrations necessary to continue the bactericidal action on the damaged bacteria were essentially the same as those originally necessary to initiate it. From the foregoing considerations one may conclude that for the foci of infection here studied the duration of the direct bactericidal action of penicillin in vivo is generally approximated by the time for which the serum concentrations remain in excess of 2 to 5 times the minimal bactericidal level in vitro. REFERENCES BUCKWALTER, F. H., AND DIcusoN, H. L A new absorption delaying vehicle for penicillin. J. Am. Pharm. Assoc., 37, COOKE, J. V., AND GOLDRING, D The concentration of penicillin in various body fluids during penicillin therapy. J. Am. Med. Assoc. 127, 80. CUTTING, W. C., LUDUENA, F. P., FIESE, M., ELLIOTT, H. W., AND FIELD, J., II Distribution and fate of penicillin in the body. J. Pharmacol., 85, EAGLE, H The slow recovery of bacteria from the toxic effects of penicillin. Read at the Second Antibiotics Symposium, National Institutes of Health, April 11, 1949, Washington, D. C. EAGLE, H The evolution of penicillin resistance in bacterial cultures. In preparation. EAGLE, H., FLEISCHMAN, R., AND MUSSELMAN, A. D. 1949a The serum concentration of penicillin G in mice, rabbits, and men after its intramuscular injection in aqueous solution. J. Bact., 57, EAGLE, H., FLEISCHMAN, R., AND MUSSELMAN, A. D. 1949b The effect of the size of the inoculum and the age of the infection on the curative dose of penicillin and experimental infections with streptococci, pneumococci and T. pallidum. J. Exptl. Med., 90, EAGLE, H., FLEISCHMAN, R., AND MUSSELMAN, A. D. 1950a The effect of the schedule of administration on the therapeutic efficacy of penicillin: the importance of the aggregate time for which penicillin remains at effectively bactericidal levels. Am. J. Med. In press. EAGLE, H., FLEISCHMAN, R., AND MU5SELMAN, A. D. 1950b The participation of the host in the therapeutic activity of penicillin. In preparation. EAGLE, H., AND MAGNUSON, H. J Effect of the method of administration on the therapeutic efficacy of sodium penicillin in experimental syphilis. Bull. Johns Hopkins Hosp., 79, EAGLE, H., AND MUSSELMAN, A. D The rate of bactericidal action of penicillin in vitro as a function of its concentration, and its paradoxically reduced activity at high concentrations against certain organisms. J. Exptl. Med., 88,

19 1950] CONCENTRATIONS OF PENICILLIN 643 EAGLE, H., AND MUSSELMAN, A. D The slow recovery of bacteria from the toxic effects of penicillin. J. Bact., 58, EAGLE, H., AND NEWMAN, E The renal clearance of penicillins F, G, K, and X in rabbits and man. J. Clin. Investigation, 26, EAGLE, H., AND TUCKER, H. A The effect of human serum on the dilution bioassay of penicillins G, X, and K. J. Bact., 56, ERCOLI, N., LEWIs, M. N., SCHWARTZ, B. S., AND WHITEHEAD, M Fate and distribution of penicillin in the body. II. Duration of blood concentration and chemotherapeutic effectiveness. Proc. Soc. Exptl. Biol. Med., 69, GRUNBERG, E., SCHNITZER, R. J., AND UNGER, C Mechanism of the topical effect of penicillin G in experimental local streptococcal infections. Yale J. Biol. Med., 20, GRUNBERG, E., UNGER, C., AND ELDRIDGE, D The sensitivity of streptococci to penicillin G after exposure to the antibiotic in vivo. Yale J. Biol. Med., 21, JAWETZ, E Dynamics of the action of penicillin in experimental animals. Arch. Internal Med., 77, KARBER, G Beitrag zur kollektiven Behandlung pharmakologischer Reihenversuche. Arch. exptl. Path. Pharmakol., 162, MAGNUSON, H. J., EAGLE, H., AND FLEISCHMAN, R The minimal infectious inoculum of Spirochaeta pallida (Nichols strain), and a consideration of its rate of multiplication in vivo. Am. J. Syphilis, Gonorrhea Venereal Diseases, 32, RICHARDSON, A. P., MILLER, I., SCHUMACHER, C., JAMBOR, W., PANSY, F., AND LAPEDES, D Physiological disposition of penicillin G and K in dogs. Proc. Soc. Exptl. Biol. Med., 63, STRUBLE, G. C., AND BELLOWS, J. G Studies on the distribution of penicillin in the eye and its clinical application. J. Am. Med. Assoc., 125, TILLETT, W. S., MCCORMACK, J. E., AND CAMBIER, M. J The treatment of lobar pneumonia with penicillin. J. Clin. Investigation, 24, TOMPSETT, R., SHULTZ, S., AND MCDERMOTT, W. 1947a Influence of protein-binding on the interpretation of penicillin activity in vivo. Proc. Soc. Exptl. Biol. Med., 65, TOMPSETT, R., SHULTZ, S., AND MCDERMOTT, W. 1947b The relation of protein-binding to the pharmacology and antibacterial activity of penicillins X, G, dihydro F, and K. J. Bact., 53, TOMPSETT, R., TIMPANELLI, A., GOLDSTEIN, O., AND MCDERMOTT, W Discontinuous therapy with penicillin. J. Am. Med. Assoc., 139,