Effect of Antibiotic Treatment Delay on Therapeutic Outcome of Experimental Acute Otitis Media Caused by Streptococcus pneumoniae Strains with Different Susceptibilities to Amoxicillin

Abstract

The effect of delayed administration of amoxicillin on the course of acute otitis media (AOM) caused by two Streptococcus pneumoniae strains with different susceptibilities to amoxicillin (MICs of 0.016 and 1 μg/ml for strains A and B, respectively) was evaluated in the gerbil model. The organisms were inoculated by transbullar challenge into the middle ear, and antibiotic treatment was administered at various times thereafter. The bacteriological and clinical efficacies of treatment diminished significantly with the delay of antibiotic administration. The bacterial eradication rates when antibiotic treatment was started at 2, 5, 8, 18, and 21 h post-bacterial inoculation were different for both strains (95, 95, 90, 55, and 55% for strain A and 95, 95, 65, 10, and 0% for strain B). Results of further experiments using strain B with higher antibiotic doses and numbers of administrations and different follow-up times indicate that the failures observed with the delayed administration were not related to the bacterial burden, selection of antibiotic-resistant mutants, or inadequate pharmacodynamic parameters. Such failures may be related to the metabolic bacterial status. The delayed amoxicillin treatment of AOM caused by S. pneumoniae may lead to therapeutic failures, mainly when organisms with diminished antibiotic susceptibility are involved.

Otitis media is one of the most frequent illnesses of childhood and is responsible for nearly one-third of visits to health care centers (26). Acute otitis media (AOM) is usually suppurative or purulent, the predominant organism involved being Streptococcus pneumoniae. Infection with pneumococci seems to be more serious and less likely to clear spontaneously than infection with other AOM-causing pathogens (3, 27). Although other organisms, even β-lactamase producers, may be responsible for AOM, amoxicillin continues to be one of the most recommended drugs for treatment (5, 14). Up to 80% of children with AOM will resolve the infection within 1 week without antibiotic treatment, and only 13% will be benefited by such treatment (24). The generalized used of antibiotics for this clinical condition increases health care costs and creates numerous adverse effects, including selection of antibiotic-resistant mutants. For all these reasons, new strategies to minimize unnecessary prescribing of antibiotics have been suggested. One approach is to delay treatment for 48 to 72 h after the diagnosis to determine whether there is spontaneous clinical improvement (10, 17). This policy has been associated with a reduction of antibiotic use for AOM to 31% and with decreased resistance among causative organisms (10). However, the effect of delaying antibiotic treatment has not been extensively studied. At the beginning of the antibiotic era, when penicillin-resistant pneumococci were not a problem, several studies carried out by Eagle et al. (6-8) brought such an approach to animal models. In experimental infections in mice inoculated with either fully penicillin-sensitive S. pneumoniae or S. pyogenes, the impaired effect of the delayed treatment with benzylpenicillin was established (6, 8). At the present time, the effect of delayed treatment on antibiotic efficacy must also be evaluated within the context of the worldwide problem of antibiotic resistance in S. pneumoniae. For these reasons, we have studied the effect of delayed antibiotic administration on the course of experimental AOM caused by both penicillin-sensitive and -resistant S. pneumoniae strains. This model has been very useful for evaluating antibiotic treatment either for AOM or for otitis media with effusion (OME) (1, 2, 4, 21, 23, 25). Most of these experimental models have shown that many organisms inoculated into the middle ear (ME) are cleared by antibiotics active in vitro, but in most published studies the antibiotics were administered very early.

The aim of this study was to evaluate the effect of delayed administration of amoxicillin on the course of experimental AOM caused by either a penicillin-sensitive or a penicillin-resistant S. pneumoniae strain.

MATERIALS AND METHODS

Bacteria

Two clinical isolates of S. pneumoniae serotype 23F (strains A and B, for which both MICs and minimal bactericidal concentrations [MBCs] of benzylpenicillin were 0.016 and 2 μg/ml, respectively) were used. Bacterial virulence was maintained by passage in gerbils.

Antibiotic

Amoxicillin trihydrate (SmithKline Beecham Pharmaceuticals, Worthing, England) was used in the in vitro studies.

For in vivo (therapeutic) use, amoxicillin in commercial vials (Clamoxyl; SmithKline Beecham S.A., Madrid, Spain) was reconstituted in apyrogen sterile distilled water to the desired concentrations.

In vitro studies

MICs and MBCs of amoxicillin were determined by microdilution as previously described (18, 19). Modal values from three separate determinations were considered. In vitro susceptibilities were also determined for 52 strain B isolates collected from animals receiving treatment (5 and 20 mg of amoxicillin/kg of body weight administered 21, 24, and 27 h after bacterial inoculation) and for 4 isolates collected from untreated controls 48 h after bacterial inoculation. Retest was performed using the E-test method (AB Biodisk, Solna, Sweden) according to the manufacturer's recommendations.

Animals

Eight- to 9-week-old adult female Mongolian gerbils (Meriones unguiculatus) weighing 49 ± 5 g each were purchased from the Centre d′Élevage R. Janvier (Le Genest, St.-Isle, France) and managed as previously described (25). The study was performed in accordance with prevailing regulations regarding the care and use of laboratory animals in the European Community and approved by our ethical committee for animal experimentation.

Experimental otitis

Overnight cultures of the organisms were kept in aliquots at −70°C. On the day of the experiment, a freshly thawed aliquot of S. pneumoniae was incubated for 4 h at 37°C in a 5% CO₂ atmosphere in brain heart infusion broth (Oxoid, Unipath, Basingstoke, Hampshire, England) enriched with 5% horse serum (Biomerieux, Marcy-l'Étoile, France). The number of viable bacteria was determined by colony counting. Animals were inoculated bilaterally with approximately 10⁶ CFU of S. pneumoniae per 20 μl, which was introduced directly into the ME bullae. The tympanic membranes were left intact and swelled without rupture during the inoculation. A normal tympanic aspect and correct inoculation were verified with an operating microscope. AOM was defined as otorrhea through a perforation in the tympanic membrane and/or inflammatory signs accompanied by changes in the membrane's normal yellowish-pink appearance, that is, development of a gray, dark brownish-yellow, or whitish opaque area with a very rough surface texture. OME was defined as no inflammatory signs in the tympanic membrane with the presence of air fluid levels or ME fluid (MEF) with or without signs of negative ME pressure. Animals showing otoscopic signs of both AOM and OME were considered to have intermediate otitis media (IOM). Three animals (six ears) were inoculated with 20 μl of sterile culture medium/ear to examine the possible role of the broth as an inductor of otitis media.

Treatment regimens and efficacy studies

Amoxicillin was administered subcutaneously (s.c.) in 500 μl as a single dose of 5 mg/kg and as repeated doses (three shots) of 5, 10, and 20 mg/kg. Different times of treatment initiation (postinoculation [p.i.]) were evaluated. For strain A, amoxicillin at 5 mg/kg was administered at 2, 5, 8, 18, or 21 h p.i. For strain B, three different administration schedules were used: (i) 2, 5, 8, 10, 18, or 21 h p.i. for the 5-mg/kg dose; (ii) 2, 5, and 8 h p.i. for the 5-mg/kg dose administered as three shots; and (iii) 21, 24, and 27 h p.i. for the 5-, 10-, and 20-mg/kg doses administered as three shots. Animals in the control groups received apyrogen sterile distilled water as a placebo. Groups of 6 to 10 animals per treatment and control groups were included. Efficacy was evaluated at 48 h p.i. for single and repeated doses. For strain B, additional experiments were carried out in parallel with two antibiotic doses (5 and 20 mg/kg; three shots). Treatment was initiated at 2 or 21 h p.i., and bacteriological evaluation took place 27 and 46 h after treatment initiation.

Treated and control animals were studied longitudinally for the presence of otorrhea, changes in weight, otoscopic aspects, and composition of ME samples. At different periods, otoscopic aspects were examined in both ears and ME samples were obtained from both ears by washing the ME fossae with 20 μl of saline solution injected and withdrawn with a 0.33-mm needle via the epitympanic membranes. Bacterial counts in ME washing fluid (MEWF) were then determined. Aliquots of serial 10-fold dilutions in saline were plated onto sheep's blood agar and incubated for 24 h at 35°C in a 5% CO₂ atmosphere. Bacterial counts are expressed as log₁₀ numbers of CFU per 20 μl; the lowest detectable bacterial count was 4 CFU/20 μl (0.60 log₁₀ CFU/20 μl). To evaluate the presence of polymorphonuclear leukocytes, 3 μl of MEWF was extended over a 6-cm² slide surface for Gram staining and observed under a high-power (magnification, ×1,000) microscope.

Pharmacokinetic studies

Amoxicillin concentrations in MEF without washing were determined in groups of 10 animals bilaterally inoculated with strain B under the same conditions as previously indicated in the experimental otitis model. One group received a single 5-mg/kg dose of amoxicillin administered s.c. 5 h after bacterial inoculation, and other groups received a single dose of the antibiotic (5 or 20 mg/kg) administered s.c. 21 h after bacterial inoculation. MEF samples were obtained with a 0.33-mm needle via the epitympanic membranes 60, 90, 120, 180, and 240 min after antibiotic administration. Aliquots of MEF samples having ≥2 μl of exudate were pooled and frozen at −70°C until determination of antibiotic levels.

Antibiotic concentrations were determined, after filtration, by a microbiological assay using Micrococcus luteus ATCC 9341 (15). Standard curves for determination of antibiotic concentrations were derived from standard solutions prepared in 0.1 M phosphate buffer, pH 6.0. Calibration standard responses were linear over the range of 0.063 to 4 μg/ml. The lower limit of quantitation was 0.063 μg/ml. Assay variability for individual samples was <10%.

Antibiotic concentration-time curves for each antibiotic dose were analyzed by a noncompartmental approach using the Win-Nonlin program (Pharsight, Mountainview, Calif.), and areas under the concentration-time curves (AUCs) were determined by the trapezoidal rule. The duration for which the drug concentration was above the MIC was calculated graphically from the semilogarithmic representation of the concentration-time curve and the regression line representing the apparent elimination rate constant.

Statistical analysis

The number of ear samples with a positive count divided by the total number of ear samples was calculated to give the percentage of positive ear samples in each group of animals. To detect differences in eradication rates in each group, the Fisher exact test was used. Bacterial counts for untreated and treated animals are expressed as arithmetic mean log₁₀ numbers of CFU per 20 μl of MEWF, culture-negative samples being included in the calculation of means by assuming a value at the detection limit. Analysis of covariance was used to compare the reduction in log₁₀ numbers of CFU and loss of body weight at 48 h in each group, and our calculations controlled for basal log₁₀ numbers of CFU, basal weights, and differences by groups. When the analysis of covariance P value was significant (P < 0.05), contrast between groups was made using the Tukey-Kramer test to adjust the type I experimentwise error.

RESULTS

In vitro studies

Both the MICs and MBCs of amoxicillin for strains A and B were 0.016 and 1 μg/ml, respectively. Retest of susceptibility of isolates of strain B collected from control and treated animals showed identical values (1 μg/ml for 92.9% and 0.75 μg/ml for 7.1% of isolates).

Experimental otitis and therapeutic efficacy

Animals inoculated with sterile culture medium showed neither clinical nor otoscopic signs of disease, and at 48 h the MEWF was culture negative.

Table 1 summarizes the bacteriological and clinical efficacies for strain A, evaluated at 48 h p.i., of amoxicillin administered in a single dose (5 mg/kg) at different intervals. After inoculation with 6.61 ± 0.43 log₁₀ CFU (mean ± standard deviation [SD] of amounts for four experiments), bilateral AOM was obtained at 48 h p.i. in 95% of all ears from animals that did not receive the antibiotic. Most MEWF specimens contained between 6.7 and 8.6 polymorphonuclear cells per field, with intra- and extracellular organisms. Animals showed lethargy and significant weight loss, and 100% of ears showed otorrhea. The antibiotic reduced the number of culture-positive ear specimens as well as the number of organisms recovered compared to numbers for untreated controls (P < 0.001). However, the number of culture-negative ear specimens was much higher and the number of colonies was lower when the antibiotic was administered within the 8 h after bacterial challenge than when the drug was administered at 18 h and thereafter (P < 0.05). The amount of body weight lost was significantly smaller in animals treated between 2 and 8 h than in those treated later and in untreated controls (P < 0.001). Most (90%) ears from animals treated early (between 2 to 8 h) showed OME in the otoscopic examination, while the majority (85%) of those from animals receiving delayed treatment (18 and 21 h) showed AOM.

TABLE 1.

Bacteriological and clinical efficacies^a of amoxicillin administered in a single dose (5 mg/kg) at different intervals after inoculation of animals with S. pneumoniae strain A

Group	Time (h)		% of culture-positive ME samples	Mean bacterial count (log10 no. of CFU per 20 μl) ± SD	Mean % of body wt loss ± SD
	Challenge to treatment	Antibiotic administration to ME sampling
Untreated control	NAb	NA	95	2.79 ± 0.84	12.54 ± 2.93
1	2	46	5c,d	0.64 ± 0.16c,d	1.74 ± 1.42c,d
2	5	43	5c,d	0.61 ± 0.07c,d	2.72 ± 1.14c,d
3	8	40	10c,d	0.61 ± 0.07c,d	5.34 ± 4.43c,d
4	18	30	45c	1.07 ± 0.86c	10.52 ± 1.95
5	21	27	45c	1.02 ± 0.62c	10.88 ± 4.62

^aEvaluated 48 h post-bacterial inoculation; 10 animals (20 ears) per group.

^bNA, not applicable.

^cSignificant difference (P < 0.001) from value for untreated control.

^dSignificant difference (P < 0.05) from values for animals treated at 18 and 21 h.

Table 2 summarizes the bacteriological and clinical efficacies for strain B, evaluated at 48 h p.i., of amoxicillin administered in a single dose (5 mg/kg) at different intervals. After inoculation with 6.35 ± 0.37 log₁₀ CFU (mean ± SD of amounts for five experiments), bilateral AOM was obtained at 48 h p.i. in 95% of all ears from animals that did not receive the antibiotic. Most MEWF specimens contained between 7.5 and 10.8 polymorphonuclear cells per field, with intra- and extracellular organisms. Animals showed lethargy and significant weight loss, and 85% of ears showed otorrhea. When the antibiotic was administered 2 and 5 h p.i., significant (P < 0.001) decreases in numbers of culture-positive ear samples and colony counts compared with those for untreated animals were found. Furthermore, the amount of body weight lost was significantly smaller than that in untreated animals (P < 0.001). On the other hand, when the antibiotic was administered either 18 or 21 h p.i., the number of culture-positive ear samples and the amount of weight lost showed no significant differences from those in controls (untreated animals). However, treated animals showed a slight reduction in numbers of organisms isolated compared with untreated animals, although the difference was not statistically significant. The results when the antibiotic was administered 8 or 10 h p.i. showed intermediate values, with 35 to 55% of samples culture positive and moderate decreases in the number of organisms and the amount of weight lost compared to those in untreated controls (P < 0.05). Persistence of AOM was linked to bacteriological failure and was observed, by otoscopic examination, in untreated animals and in animals in which antibiotic treatment was delayed until ≥18 h p.i. On the other hand, animals showing bacteriological eradication (those treated 2 and 5 h p.i.) had OME. There was not a clear pattern in animals showing moderate bacteriological success (those treated 8 and 10 h p.i.), which had different otoscopic appearances (AOM, OME, and IOM).

TABLE 2.

Bacteriological and clinical efficacies^a of amoxicillin administered in a single dose (5 mg/kg) at different intervals after inoculation of animals with S. pneumoniae strain B

Group	Time (h)		% of culture-positive ME samples	Mean bacterial count (log10 no. of CFU per 20 μl) ± SD	Mean % of body wt loss ± SD
	Challenge to treatment	Antibiotic administration to ME sampling
Untreated control	NAb	NA	95	3.25 ± 1.29	9.37 ± 2.25
1	2	46	5c	0.60 ± 0.00c	2.67 ± 1.27c
2	5	43	5c	0.64 ± 0.17c	3.34 ± 2.14c
3	8	40	35c	1.08 ± 0.86c	5.89 ± 2.86c
4	10	38	55c	1.36 ± 0.99c	6.49 ± 3.20c
5	18	30	90	2.46 ± 1.09	9.93 ± 2.50
6	21	27	100	2.76 ± 1.02	10.06 ± 2.20

^aEvaluated 48 h post-bacterial inoculation; 10 animals (20 ears) per group.

^bNA, not applicable.

^cSignificant difference (P < 0.05) from values for untreated control and groups treated at 18 and 21 h.

Table 3 summarizes the natural evolution of infection in untreated animals inoculated with strain B. Shown are the numbers of culture-positive ear samples, bacterial counts (expressed as arithmetic mean log₁₀ numbers of CFU per 20 μl [± SD]) in MEWF, and results of otoscopic examination at different intervals after inoculation with 6.09 ± 0.25 log₁₀ CFU in 20 μl (mean ± SD of amounts for four experiments). All MEWF samples taken between 2 and 48 h were culture positive. The maximal bacterial burden was achieved 5 h after bacterial inoculation and was maintained until 18 h, after which the burden decreased. Polymorphonuclear leukocytes in MEWF were present 2 h p.i. The number increased at 5 h p.i., and the elevated level was maintained throughout the rest of the observation period. AOM was detected 5 h p.i., with frank otorrhea appearing at 18 h post-bacterial challenge.

TABLE 3.

Evolution of results of otoscopic examination and analysis of culture-positive MEWF at different intervals after inoculation of animals with 6.09 ± 0.25 (mean ± SD) log₁₀ CFU of S. pneumoniae strain B per ear^a

Time (h) from bacterial inoculation	% of culture-positive ME samples	Mean bacterial count (log10 no. of CFU per 20 μl) ± SD	Finding(s) of otoscopic examination
2	100	5.17 ± 1.28	Slight tympanic redness
5	100	6.90 ± 1.59	AOM with dry ears
8	100	6.84 ± 0.43	AOM with wet ears
18	100	6.55 ± 0.50	AOM with otorrhea
21	100	5.39 ± 1.16	AOM with otorrhea
29	100	4.04 ± 0.85	AOM with otorrhea
48	100	3.22 ± 1.12	AOM with otorrhea
67	75	2.22 ± 1.02	AOM, IOM, and OME

^aEight animals (16 ears) per group.

Table 4 presents the bacteriological and clinical results for strain B after administration of different doses of amoxicillin (5, 10, and 20 mg/kg) in three shots, with treatment starting at 21 h pi. There were no significant differences in the numbers of culture-positive samples, otoscopic appearances (all animals showed AOM), and amounts of weight loss among treated and untreated groups. However, a significant (P < 0.001) decrease in the numbers of organisms was found in treated animals compared with the numbers in controls, with no significant differences between animals receiving different amoxicillin doses.

TABLE 4.

Bacteriological and clinical efficacies^a of amoxicillin administered in three different doses (three shots) at 21, 24, and 27 h after inoculation of animals with S. pneumoniae strain B

Group (dose in mg/kg)	No. of positive ear samples/total no. of samples (%)	Mean bacterial count (log10 CFU per 20 μl) ± SD	Mean % of body wt loss ± SD
Untreated control	14/14 (100)	3.73 ± 0.84	9.7 ± 2.3
1 (5)	16/18 (88.9)	2.03 ± 1.05b	9.8 ± 3.2
2 (10)	17/18 (94.4)	1.94 ± 0.76b	8.0 ± 3.8
3 (20)	13/14 (92.8)	1.96 ± 0.75b	8.5 ± 1.6

^aEvaluated 48 h post-bacterial inoculation.

^bSignificant difference (P < 0.001) from value for untreated control.

When the results obtained after the administration of 5 mg/kg of amoxicillin as a single dose were compared to those obtained with the same dose administered as three shots (with treatment starting at 21 h p.i.) (Tables 2 and 4), no significant differences were observed in the numbers of culture-positive ear samples and amounts of weight loss, but significant differences (P < 0.05) were found in bacterial counts, which were lower after three shots.

Table 5 presents the results obtained in studies carried out in parallel with strain B to rule out the possible influence of the evaluation time point. In these studies, two antibiotic doses (5 and 20 mg/kg; three shots) were used, treatment was started at 2 or 21 h p.i., and bacteriological evaluations were made 27 and 46 h after treatment initiation. When antibiotic treatment (5 mg/kg) was started 2 h p.i., significant differences in bacteriological results (P < 0.001) between treated and untreated animals were found, regardless of whether evaluation took place 27 or 46 h after treatment initiation. When antibiotic treatment (5 or 20 mg/kg) was started 21 h p.i., no significant differences in numbers of culture-positive samples between treated and untreated animals were observed. A significant (P < 0.05) reduction in the colony counts versus those in untreated controls was observed in animals treated with 5 mg of amoxicillin/kg and evaluated at either 27 or 46 h after starting the treatment. However, after administration of 20 mg/kg, a significant (P < 0.05) reduction in the colony counts was observed only when evaluation was performed 27 h after treatment initiation. No significant differences in either numbers of culture-positive samples or colony counts were found when the results obtained with the 5- and 20-mg/kg doses were compared.

TABLE 5.

Bacteriological efficacies, evaluated 27 and 46 h after initiation of treatment, of two amoxicillin doses administered 2 and 21 h after inoculation of animals with S. pneumoniae strain B

Groupa	Time(s) (h)			% of culture-positive samples after dose (mg/kg) of:			Mean bacterial countd ± SD after dose (mg/kg) of:
	Challenge to treatment	Challenge to ME sampling	Treatmentb to ME sampling	5	20	0 (untreated control)	5	20	0 (untreated control)
1	2, 5, 8	29	27	0.0c		100	<0.60c		3.92 ± 0.88
2	2, 5, 8	48	46	8.3c		100	0.64 ± 0.14c		3.04 ± 1.01
3	21, 24, 27	48	27	91.7		100	1.93 ± 0.93c		3.04 ± 1.01
4	21, 24, 27	67	46	41.7		75	1.12 ± 0.86c		2.16 ± 1.26
5	21, 24, 27	48	27		83.3	100		2.22 ± 0.93c	3.04 ± 1.01
6	21, 24, 27	67	46		66.7	75		1.26 ± 0.86	2.16 ± 1.26

^aSix animals (12 ears) per group.

^bTime from start of treatment.

^cSignificant difference (P < 0.05) by dose and time compared to value for untreated control.

^dValues are mean log₁₀ numbers of CFU ± SD in 20 μl of MEWF. Data for culture-negative samples were included in the calculation of means by assuming a value at the detection limit (0.60 log₁₀ CFU).

Pharmacokinetic and pharmacodynamic data

Table 6 presents the pharmacokinetic and pharmacodynamic data in relation to bacterial eradication for the MEF obtained following administration of two different doses of amoxicillin 5 and 21 h after inoculation with strain B. The delay in administering the same dose (5 mg/kg) of amoxicillin resulted in a MEF drug concentration of less than 50% of that obtained when the antibiotic was administered earlier, and this reduction correlated with a lower eradication rate (10 versus 95%). The delayed administration of a higher dose (20 mg/kg) did not improve the bacteriological results obtained with the delayed administration of 5 mg/kg, in spite of achieving MEF antibiotic concentrations double those obtained with the lower dose. For antibiotic doses achieving similar MEF drug concentrations (5 and 20 mg/kg administered 5 and 21 h p.i., respectively), the amoxicillin half-lives in the MEF were different. The half-life determined for the 5-mg/kg dose (administered 5 h after inoculation) was longer than the half-life determined for the 20-mg/kg dose (administered 21 h after inoculation), and this resulted in a higher AUC/MIC ratio, a longer amount of time for which the drug concentration in the ME sample exceeded the MIC for the pathogen (t > MIC), and a higher eradication rate (95 versus 11.5%).

TABLE 6.

MEF pharmacokinetic and pharmacodynamic data in relation to eradication of S. pneumoniae strain B after treatment of animals with a single dose of amoxicillin

Groupa (dose in mg/kg)	Time (h) from challenge to treatment	C90 minb (μg/ml)	Half-life (min)	AUC/MIC	Amt of time (h) for which drug concn exceeded MIC	% Eradication
1 (5)	5	5.55	77	16.4	4.5	95
2 (5)	21	2.33	NDc	ND	ND	10
3 (20)	21	5.05	30	9.6	2.5	11.5

^aTen animals (20 ears) per group.

^bC_{90 min}, concentration of amoxicillin at 90 min after treatment.

^cND, not determined.

DISCUSSION

The 5-mg/kg amoxicillin dose was chosen because it achieves serum and MEF drug concentrations in gerbils similar to those obtained in children receiving a standard dose for the treatment of AOM (4, 12, 21). This dose administered 2 h after bacterial inoculation prevented the development of AOM and, when administered 5 h after bacterial inoculation, was efficacious in the treatment of AOM. Such efficacy was obtained after the administration of a single dose. However, the clinical (judged by weight loss), otoscopic, and bacteriological efficacies clearly decreased as the antibiotic treatment was delayed. These results were more evident with strain B (a penicillin-resistant strain for which the MIC of amoxicillin is 1 μg/ml). For this reason, most of our further studies were focused on this strain.

The decrease in efficacy observed with the delayed administration of the antibiotic cannot be explained by selection of subpopulations for which MICs were higher since this possibility was discarded by testing the in vitro susceptibilities of the organisms recovered. Furthermore, such results cannot be explained by the bacterial burden present at the time of treatment initiation because the antibiotic efficacy was greater at 5 h than at 21 h, when the bacterial burden was significantly lower in untreated controls. As a single 5-mg/kg dose of amoxicillin failed when administered 21 h after bacterial inoculation, additional experiments (also with delayed administration) were carried out in which three shots were used and the antibiotic dose was increased to 10 and 20 mg/kg. The dosing intervals (every 3 h) were chosen based on the amoxicillin half-life in gerbil serum that is approximately one-third of that obtained in humans (4, 21). Although the administration of three shots of amoxicillin at 5 mg/kg did not significantly reduce the number of culture-positive samples or the effect of the infection on weight loss, colony counts in ME samples were significantly reduced compared to those in untreated animals and animals receiving the single dose. When the antibiotic was administered in three shots at higher doses (10 and 20 mg/kg), the clinical, otoscopic, and bacteriological results were not better than those obtained with the lower dose. These results indicate that antibiotic efficacy is lower when treatment is delayed, and the administration of higher doses does not improve efficacy. However, the administration of repeated doses of amoxicillin decreased the bacterial burden. These results are in agreement with those previously described for the peritonitis and muscle infection models (6, 8, 11, 16), but this model adds a new variable due to the clearing effect of otorrhea, a factor commonly associated with AOM. Furthermore, our work shows the relation among the in vitro antibiotic susceptibility of the pathogen, the time at which the drug is administered, and the in vivo efficacy.

The results of experiments done in order to evaluate the influence that different periods between treatment initiation and efficacy evaluation might have on outcome confirmed that the early administration of the antibiotic leads to better results than the delayed administration, regardless of the time of evaluation.

From the pharmacokinetic-pharmacodynamic perspective, early antibiotic treatment resulted in amoxicillin concentrations in ME samples that were 2.4 times higher than those obtained with delayed administration, with 95 and 10% bacteriological eradication, respectively. With the dose increased to 20 mg/kg (with delayed treatment), the antibiotic concentration achieved in ME samples at 90 min was similar to that obtained with the early administration of the low dose but the efficacy of the treatment was not improved. On the other hand, the antibiotic half-life in the ME samples was longer with early administration than with delayed administration, which may be due to the presence of otorrhea at the time of delayed administration. Consequently, the t > MIC was greater with the low dose administered early than with the high dose administered late (4.5 versus 2.5 h, equivalent to 100 versus 83.3% of the dosing interval, respectively). However, the differences in half-lives, AUC/MIC ratios, and t > MIC between the two regimens (5 mg/kg administered early and 20 mg/kg administered late) do not explain the differences in efficacy (95 versus 11.5%) found between the regimens. Moreover, the high dose in the delayed-administration regimen achieved very favorable pharmacodynamic parameters, including the t > MIC (83.3% of the dosing interval), but in this case, they were not predictive of antibiotic efficacy.

The diminished efficacy of penicillin in experimental infections caused by penicillin-sensitive pneumococci after late administration of the antibiotic was reported by Eagle in 1949 (6), suggesting that the different metabolic statuses of the bacteria could explain such an effect. The activity of beta-lactams against different organisms in a stationary or logarithmic phase of growth is controversial (8, 11, 16), probably due to the very different methodologies used for its evaluation. Although the number of organisms in our experimental model remained relatively stable between 5 and 18 h, an active growth phase occurred between 2 and 5 h, with colony counts starting to decrease at 21 h. As suggested by Knudsen et al. (16), the stationary phase may reflect two different conditions: (i) a balanced rate of growth and death of bacteria, where the generation time is not affected (during the first hours), and (ii) slower growth due to a longer generation time (after the first hours). It is possible that organisms were in the logarithmic phase of growth during the first period (2 to 5 h) of our experimental model, in the early stationary phase of growth between 5 and 10 h, and in the stationary phase of growth afterwards. As the organisms approached the stationary phase, the efficacy of amoxicillin would be diminished. Another factor that may explain the results obtained in this study is the presence of bacterial biofilms, where generation time is increased. Recently, Ehrlich et al. (9) have shown, in a chinchilla experimental model of AOM, the early development of mucosal biofilms. In such biofilms, bacteria may be protected from both antibiotics and antibodies. There are also several reports showing that bacteria living in biofilms are in general less susceptible to antibiotics than free-living bacteria (20, 29).

What could be the clinical relevance of our results for human medicine? With few exceptions, most authorities recommend antibiotic treatment for AOM although bacteriological and clinical failures occur occasionally even when the organism involved is susceptible to the prescribed drug (22). However, most clinical trials evaluating the efficacy of antibiotic treatment of AOM do not stratify results according to the times at which clinical symptoms appeared and antibiotic treatment was initiated or according to the presence or absence of otorrhea in patients at the time that the first dose was administered. Such an approach could provide more information on the actual efficacy of antibiotics in the treatment of AOM as well as explain some failures. In order to attain the maximal drug efficacy and avoid resistance, it is recommended that antibiotics and doses be chosen based on results of local antimicrobial susceptibility studies, that the organism most probably involved be taken into account, and that the drug be administered for short time periods (5, 13, 14). As far as the delay of antibiotic treatment is concerned, it is not easy to extrapolate the results obtained in the animal model to human medicine, but our results suggest that if antibiotics are indicated in cases of AOM they should be administered very early, as recently proposed by Handley (14), especially in countries where the organism may have diminished susceptibility to the administered drug. This attitude should lead to the achievement of maximal efficacy and at the same time may prevent serious complications (28).