Rabbit Streptococcus pneumoniae Keratitis Model and Topical Therapy

Transcription

1 Rabbit Streptococcus pneumoniae Keratitis Model and Topical Therapy James P. Guzek, 12 Dennis J. Cline 2 Paul K. Row, 5 Izak F. Wessels 2 ' 4 Scott Beeve, 5 Scott Ispirescu, 5 Raydolfo M. Aprecio, 5 James D. Kettering, 5 David L Gano, 2 and Gregory M. Nelson 5 PURPOSE. TO develop a model for experimental Streptococcus pneumoniae keratitis and to evaluate the chemotherapeutic efficacy of 12 common topical antibiotics in vivo. METHODS. Five-hundred CFUs of log-phase S. pneumoniae were injected into the central corneal stroma of 36 eyes of 18 rabbits. After 0, 4, 8, 16, 24, and 48 hours, the in vivo growth was assayed as the CFU per cornea. Epithelial removal (to promote antibiotic entry and mimic human keratitis) was evaluated. Disc or tube dilution verification of the sensitivity or resistance of three S. pneumoniae strains was performed: a penicillin sensitive ("S"), an intermediate sensitive ("I"), and a resistant ("R") strain. Keratitis was established with S. pneumoniae "S" in 65 eyes, 5. pneumoniae "I" in 107 eyes, and S. pneumoniae "R" in 78 eyes. Sixteen hours later, control corneas were harvested and the epithelium removed from treatment corneas. Every half hour saline, penicillin, gentamicin, bacitracin, ciprofioxacin, ofloxacin, erythromycin, vancomycin, ceftriaxone, cefotaxime, or chloramphenicol was applied for 5 hours. One hour later CFUs/cornea were assayed. RESULTS. After 24 hours, S. pneumoniae "S" and "I" had proliferated to 9.18 ± 6.65 X 10 6 CFUs and 9.26 ± 6.90 X 10 6 CFUs. Epithelial removal at 16 hours was not significant. The in vitro antibiotic sensitivity was as expected. However, in vivo, penicillin, gentamicin, or cefazolin sterilized S. pneumoniae "S." S. pneumoniae "R" responded best to fortified gentamicin with or without vancomycin; all others antibiotics were significantly less effective (P < 0.001). CONCLUSIONS. A small intracorneal S. pneumoniae inoculum in rabbit corneas grew and was maintained for 24 hours (with epithelial removal) to provide a model for testing antibiotic sensitivity in vivo. Topical penicillin is best for treating keratitis from penicillin-sensitive S. pneumoniae, whereas topical gentamicin or a combination of gentamicin and vancomycin was most effective against penicillin-resistant S. pneumoniae. (Invest Ophthalmol Vis Sci. 1998;39:12-17) Although Streptococcus pneumoniae (S. pneumoniae) has generally been sensitive to penicillin, the current worldwide trend is toward increasing resistance to various antimicrobials. 1 "'' There have been reports of treatment failures of meningitis, sepsis, pneumonia, and otitis media from penicillin-resistant S. pneumoniae. 5 ' 6 Currently in the United States, approximately 7% of S. pneumoniae isolates demonstrate intermediate penicillin resistance, and 1% is completely resistant. 4 This prevalence is even higher in certain regions: In Tennessee and Kentucky, 29% to 33% of nasopharyngeal S. pneumoniae isolates from children who visit county health departments for treatment and 61% from children in day-care settings are relatively penicillin resistant. 7 From the 'Margaret Marquart Catholic Hospital, Kpando, Ghana, West Africa; the 2 Department of Ophthalmology and the 3 School of Medicine, Loma Linda University, California; the 4 Department of Ophthalmology, College of Medicine, University of Tennessee, Chattanooga; and the 5 Department of Microbiology, Loma Linda University, California. Supported by the George Warren Trust Fund through the Department of Ophthalmology, Loma Linda University, California. Submitted for publication September 23, 1997; revised January 9 and May 28, 1998; accepted June 9, Proprietary interest category: N. Reprint requests: Izak F. Wessels, 1949 Gunbarrel Road, Suite 280, Chattanooga, TN It is therefore not surprising that penicillin-resistant 5. pneumoniae keratitis has been reported. 8 ' 9 The recommended treatment of choice for penicillin-sensitive S. pneumoniae keratitis is penicillin or cefazolin; whereas vancomycin or a cephalosporin has been suggested as therapy for penicillin-resistant S. pneumoniae keratitis. 89 S. pneumoniae is commonly associated with keratitis in the United States (3% to 15% of cases) and even more so in other countries (as much as 31%). *" 19 The recommended treatment includes 100,000 to 300,000 U/ml topical penicillin G or 50 mg/ml cefazolin every 15 to 30 minutes, which may not suffice for resistant organisms. 10 " 14 Two patients with penicillin-resistant S. pneumoniae keratitis failed to respond to topical cefazolin. Although topical vancomycin sufficed in one, 8 the other patient needed a penetrating keratoplasty despite treatment with topical and intravenous vancomycin. 9 The fluoroquinolones (ciprofioxacin and ofloxacin) have been suggested as alternative antibiotics but are less effective than penicillin against nonresistant organisms. 14 ' " 22 In one report, topical ciprofioxacin cured only 7 of 13 (52.8%) patients with S. pneumoniae keratitis. 16 In this study, the efficacy of various antibiotics against S. pneumoniae keratitis was evaluated in vivo. An animal model of S. pneumoniae keratitis was developed, and common topical antibiotics were tested in vivo against S. pneumoniae with various penicillin resistance. Investigative Ophthalmology & Visual Science, October 1998, Vol. 39, No. 11 Copyright Association for Research in Vision and Ophthalmology

2 IOVS, October 1998, Vol. 39, No pneumoniae Keratitis in Rabbit 13 MATERIALS AND METHODS The Institutional Review Board of Loma Linda University approved the project design. All animals were maintained in animal care facilities fully accredited by the American Association of Laboratory Animal Science, and the procedures performed were in accordance with the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research, with one exception: both corneas were tested in each animal. This was decided only after careful consideration and consultation with veterinarians involved in animal research and after confirmation by other prominent investigators studying rabbit keratitis that the protocol was acceptable. Some of the reasons were because of the short duration of the experiment (less than 24 hours) and because no signs of distress or changes in behavior were observed during the pilot study that preceded it (authors' unpublished observations). The major benefit was that it was possible to avoid killing as many animals as would otherwise have been the case. We used three encapsulated S. pneumoniae strains: (1) a penicillin-sensitive strain (S. pneumoniae "S") WU-2, serotype 3, obtained from Mary-Kay Johnson, Tulane University, New Orleans, LA; (2) an intermediate penicillin-resistant strain (S. pneumoniae "I") ATCC 49619, serotype nonspecific; and (3) a penicillin-resistant strain (5. pneumoniae "R") A-N92 to 017, also known as JG strain, from George McCracken, MD, Dallas Memorial Hospital (Dallas, TX; serotype nonspecific). Antibiotic resistance is defined by the minimal inhibitory concentrations (MIC) according to the Centers for Disease Control guidelines. 1 Penicillin "sensitive" 5. pneumoniae strains (by definition) demonstrate less than 0.06 /xg/ml as MIC; "intermediate" resistant strains between 0.12 and 1.0 /xg/ml; and "resistant" strains 2 /xg/ml or more. The Centers for Disease Control definitions were adhered to, where available, but none are reported for fluoroquinolones, aminoglycosides, or bacitracin. MIC values were determined with the E-test (AB Biodik, Remel, Lenexa, KS). When no disks were commercially available (for bacitracin, cefazolin, ceftriaxone, gentamicin, or tobramycin), tube dilution methods (i.e., the lowest concentration inhibiting growth as evidenced by an absence of gross turbidity at 24 hours) were used. S. pneumoniae Keratitis Rabbit Model S. pneumoniae "S" and S. pneumoniae "I" were prepared according to standard microbiologic procedures. 12 Briefly, one 5. pneumoniae colony was incubated for 6 hours at 37 C in 10 ml Todd-Hewitt medium (Difco Laboratories, Detroit, MI) enriched (5 ml per 100 ml) with 10% bacto-yeast extract (Difco), and filtered through a Nalgene 0.45 micron filter (Corning Glass Works, Corning, NY). This produced a suspension with approximately 5.0 X 10 8 CFUs/ml and an optical density of 0.4 at 650 nm that was diluted to 500 CFUs in 10 jui warmed medium. The repeatability of this inoculum was confirmed by serial dilution, by culturing 0.1-ml aliquots on blood agar plates and by determining CFUs after 16 hours of incubation. The recommended medium for S. pneumoniae "R" was Mueller Hinton Broth (Difco) fortified with sterile defibrinated horse blood (Hema Resource and Supply, Aurora, OR). 13 Adult New Zealand White rabbits weighing 2 to 3 kg were anesthetized with 50 mg/kg IM ketamine (Parke-Davis, Morris Plains, NJ) and 10 mg/kg IM xylazine (Phoenix Pharmaceuticals, St. Joseph, MO). With a Hamilton microsyringe 10 /xl with 500 CFUs S. pneumoniae "S" or "I" was injected into the central corneal stroma. Each rabbit received the same organism in both eyes. After 0, 4, 8, 16, 24, and 48 hours (six eyes at each interval with each organism, total 72), the animals were killed with an overdose of intravenous pentobarbital (Euthanasia-6 solution; Veterinary Laboratories, Lenexa, KS). The eyes were examined using a portable (model SI-S; Kowa, Torrance, CA) slit lamp to evaluate the size and extent of stromal infiltrates, hypopyon formation, and epithelial defects. Under aseptic conditions, the corneas were harvested, rinsed, minced, and homogenized with a Tissue-tearer (Fisher Scientific, Houston, TX) three times for 10 seconds each at 4 C. Serial dilutions were plated in duplicate on blood agar, and the CFUs/cornea were determined after 16 hours of incubation. Rabbit bacterial keratitis is rarely if ever associated with epithelial defects, in contrast to the almost invariable finding in humans. 14 Therefore, to more closely simulate the clinical circumstances and to standardize antibiotic penetration, all corneas had the epithelium removed after inoculation. With the aforementioned anesthesia and standard aseptic conditions, a fresh No. 15 Bard-Parker blade (Becton Dickinson Acute Care, Franklin Lakes, NJ) was used to remove the epithelium by scraping from the center outward in each quadrant under a microscope (Zeiss OPMI-6; Zeiss, Germany). A pilot study of 5. pneumoniae "S" placed in eight eyes with epithelial removal just after inoculation demonstrated highly variable CFUs/cornea after 21 and 45 hours (authors' unpublished data). In eyes when the epithelial removal was delayed 16 hours, the number of CFUs/cornea was found to be more consistent at 18, 21, and 45 hours after inoculation. For this report, the epithelium was removed at 14 hours to allow 2 hours of recovery from the anesthetic before the antibiotics were applied. Because of the potential risks from exposure to S. pneumoniae "R," all investigators were vaccinated with Pneumovax (Merck, West Point, PA). Likewise, the in vivo growth assessment was abbreviated to one determination 16 hours after inoculation, and 9 instead of 12 antibiotics were tested. Effects of Antibiotic Therapy The following commercially available antibiotics were dissolved in balanced salt solution (BSS; Alcon, Fort Worth, TX) to the concentrations listed and filtered through a 0.2-/xg filter (Millex, Bedford, MA). Concentrations of cefazolin (Marsan Pharmaceuticals, Cherry Hill, NJ), cefotaxime (Hoechst-Roussel, Somerville, NJ), ceftriaxone (Roche Laboratories, Nutley, NJ), chloramphenicol (Parke-Davis, Morris Plains, NY), and vancomycin (Abbott, Abbott Park, IL) were 50 mg/ml. Tobramycin (Apothecon, Princeton, NJ) and gentamicin (Elkins Sinn, Cherry Hill, NJ) were concentrated to 13-6 mg/ml. Bacitracin (Sigma Chemical, St. Louis, MO) was concentrated to 10,000 U/ml and penicillin G (Marsan Pharmaceuticals) to 300,000 U/ml. The following antibiotics were used as commercially available: chloramphenicol (Chloroptic; Allergan, Irvine, CA); ciprofloxacin (Ciloxan; Alcon); and ofloxacin (Ocuflox, Allergan) all at a 30 mg/ml concentration and erythromycin ophthalmic ointment (E. Fougera, Melville, NY) at 5 0 mg/g. Both chloramphenicol concentrations (30 mg/ml and 50 mg/ml) were tested against S. pneumoniae "I," and the com-

3 14 Guzek et al. IOVS, October 1998, Vol. 39, No. 11 TABLE 1. Antibiotic Sensitivity of the Three Organisms as Determined by the E-Test or Tube Dilution Method Antibiotic WU-2 {S} ATCC {1} A-N R} Penicillin G Cefazolin* Chloramphenicol Erythromycin Vancomycin Ciprofloxacin Ofloxacin Cefotaxime Ceftriaxone* Gentamicin* Tobramycin* Bacitracin,* U/ml 0.03 {S} 0.12 {S} 2{S} <0.25 {S} 0.75 {S} {S} 1(1} {I} 0.5(1} 1.5 {S} 0.25 {S} 0.5 {S} {S} 1(1} {R} 6{R} 3{S} 16 {R} 0.5 {S} {R} 1(1} Our findings confirmed the sensitivity levels as reported: the level is noted as {S}, fully sensitive; {I}, intermediate sensitivity; or {R}, resistant. All concentrations are /xg/ml, unless otherwise specified. bination of gentamicin with vancomycin was tested against 5. pneumoniae "I" and S. pneumoniae "R." Both eyes of a rabbit were inoculated with the same S. pneumoniae, and the rabbit was then randomized into a control or treatment group. Each group contained three rabbits (six eyes). In two rabbits, one cornea was inadvertently perforated before any organisms were injected. These eyes were excluded, but the fellow eye that had been successfully inoculated was included in one of the control groups. Therefore, two of the control groups had 11 instead of 12 eyes. The first control corneas were harvested 16 hours after inoculation. In the treatment groups, the epithelium was removed (as described above) 14 hours after inoculation. The animals were then allowed to recover from the effects of the anesthetic for 2 hours. Two applications (the second after a 2-minute wait) of medication were made every half hour for 5 hours, to a total of drops per eye. Each rabbit received the same antibiotic solution, ointment, or BSS in both eyes. For 5. pneumoniae "S," 7 antibiotics were tested and for S. pneumoniae "I" 12 were tested. One hour after the last application (22 hours after inoculation), the animals were killed and the corneas assayed to determine CFUs as described above. The School of Allied Health Professions of Loma Linda University assisted with the statistical analysis, including ANOVA and Student's -test, with statistical significance set at P = in S.pneumoniae "I." No epithelial defects developed with any of the organisms. Hypopya developed exclusively in eyes with 5. pneumoniae "I." The effects of epithelial removal are detailed in Table 2. Sixteen hours after inoculation, an average of 1.73 ± 127 X 10 6 CFUs/cornea were recovered. At 18 hours (2 hours after epithelial removal) the mean was 2.07 ± 0.55 X 10 6 CFUs/ cornea, and at 21 hours (5 hours after epithelial removal) 1.92 ± 0.49 X 10 6 CFUs/cornea. Statistical analysis revealed that the difference in CFUs/cornea after epithelial removal was not significant with the Student's f-test (P = 0.37). Antibiotic Therapy The antibiotic sensitivity testing connrmed that S. pneumoniae "S" was sensitive; S. pneumoniae "I" was of intermediate resistance, and S. pneumoniae "R" was resistant to penicillin and cefazolin (Table 3). Table 4 shows the results of topical antibiotic therapy. The inoculum sizes for 5. pneumoniae "S," "I," and "R" were 552 ± 70 CFUs, 413 ± 96 CFUs, and 502 ± 82 CFUs, respectively. Corneas with S. pneumoniae "S" yielded 2.8 ± 1.6 X 10 6 CFUs/cornea after BSS drops. Topical penicillin G sterilized all six corneas. Gentamicin and cefazolin were almost as effective; on average 6.7 ±12.1 CFUs/cornea and 44 ± 68 CFUs/cornea, respectively, were recovered. When compared with the effectiveness of penicillin, the difference did not attain statistical RESULTS Rabbit Keratitis Model The in vivo growth of 5. pneumoniae "S" and "I" in the rabbit cornea is shown in Table 1. Each inoculum for "S" and "I" contained 552 ± 70 CFUs and 407 ±104 CFUs (mean ± SEM), respectively. After 24 hours, S. pneumoniae "S" had proliferated to 918 ± 6.65 X 10 6 CFUs/cornea and S. pneumoniae "I" to 926 ± 6.90 X 10 6 CFUs/cornea. At 16 hours, examination with a slit lamp revealed central nebulas in corneas with S. pneumoniae "S" and "I," whereas S. pneumoniae "R" had denser infiltrates 3- to 4-mm in diameter. Small (2- to 3-mm diameter) infiltrates were seen at 24 hours in S. pneumoniae "S" corneas, and they were larger (3 to 4 mm) TABLE 2. Mean ± SEM of Inoculum Size and CFUs Recovered per Cornea at Various Times Inoculum size Oh 4h 8h 16 h 24 h 48 h WU-2 (S} 552 ± ± ± 1.99 X ± 0.84 X ± 4.8 X ± 6.65 X ± 1.71 X 10 5 ATCC ± ± ± 0.49 X 2.83 ± 1.11 X 8.96 ± 3.71 X 9.26 ± 6.90 x 5.85 ± 4.99 X HI {S}, fully sensitive; (I), intermediate sensitivity. At each time interval, the average was obtained from 6 eyes.

4 IOVS, October 1998, Vol. 39, No pneumoniae Keratitis in Rabbit 15 TABLE 3. Mean ± SEM of CFUs Recovered per Cornea at Various Times Time No. of Eyes CFU/Cornea 16 h 18 h 21 h 45 h ± 1.27 X ± 0.55 X ± 0.49 X ± 2.37 X At each time interval, the average was obtained from the number of eyes as noted. All eyes had the corneal epithelium removed 16 hours after inoculation. The probability values are derived from comparison with the CFU yield at 16 hours. significance (P = 0.08). Decreasing order of effectiveness was bacitracin, ciprofloxacin, erythromycin ointment, and 30 mg/ml chloramphenicol, a highly significant difference when compared with the effectiveness of penicillin using the Student's West (P < 0.001). After 5 hours of BSS drops, S. pneumoniae "I" yielded 1.78 ± 1.51 X 10 6 CFUs/cornea. Gentamicin sterilized all six corneas. Tobramycin and gentamicin with vancomycin were statistically similar, yielding 1.7 ± 4.1 CFUs/cornea and 3.33 ± 8.16 CFUs/cornea, respectively (P = 0.34 for each). The decreasing order of effectiveness for the other antibiotics was bacitracin, penicillin G, cefazolin, vancomycin, ofloxacin, ciprofloxacin, ceftriaxone, cefotaxime, 30 mg/ml chloramphenicol, erythromycin ointment, and 50 mg/ml chloramphenicol. Each of these differences was highly significant when the Student's Mest was used (P < 0.001). Ofloxacin was no different from ciprofloxacin (P = 0.22). The antibiotics that had no effect were 30 mg/ml chloramphenicol (P = 0.28), 50 mg/ml chloramphenicol (P = 0.67),. and erythromycin ointment (P = 0.29). After 5 hours of BSS drops, 5. pneumoniae "R" yielded 2.48 ± 1.06 X 10 7 CFUs/cornea. Gentamicin plus vancomycin sterilized all six of these corneas. Gentamicin alone resulted in 53 ± 103 CFUs/cornea, statistically equivalent to gentamicin plus vancomycin (P = 0.07). All other antibiotics were less effective, and included in decreasing order tobramycin, bacitracin, cefazolin, ofloxacin, penicillin G, ceftriaxone, and vancomycin (P < 0.01 for each). Among the aminoglycosides, tobramycin (1.7 ± 4.1 CFUs/cornea) was as effective as gentamicin against S. pneumoniae "I" (P = 0.34) but was less effective against S. pneumoniae "R" (7.67 ± 5.24 X 10 2 CFUs/ cornea), a highly statistically significant difference compared with gentamicin (P < 0.001) and with gentamicin plus vancomycin (P < 0.001). DISCUSSION The natural immunity of the rabbit cornea makes attempts at developing an animal S. pneumoniae keratitis model frustrating; only two reports of in vivo S. pneumoniae keratitis have been made, 23 ' 24 a model based on the injection of a very large inoculum (10 6 CFUs), one equivalent to the bacterial load of a well-developed keratitis. Although the CFU recovery remains stable for 24 hours, by 48 hours, natural immunity very profoundly decreases the yield to almost zero. 12 This model also demonstrates penicillin to be better than fluoroquinolones against a penicillin-sensitive S. pneumoniae. 22 With a smaller inoculum, we have found an initially rapid growth phase, followed by die same autosterilization by 48 hours, without stromal necrosis or epithelial defects developing. Despite the differences found when our results are compared with the typical clinical case, there does not appear to be a more appropriate model to use to compare antibiotic efficiency in vivo. Possibly, the smaller inoculum delayed the recruitment of the innate immunity and thereby permitted CFUs to be maintained for up to 48 hours more tiian the larger inoculum model. TABLE 4. The Number of Viable Organisms Recovered per Cornea WU-2 {S} ATCC {1} A-N R Controls Time zero (16 h) Balanced salt solution Treatment eyes Penicillin G Tobramycin Gentamicin Gentamicin plus vancomycin Cefazolin Bacitracin Ciprofloxacin Ofloxacin Erythromycin Vancomycin Ceftriaxone Cefotaxime Chloramphenicol 3 mg/ml 50 mg/ml 4.1 ± 1.3 X 10 6 * 2.8 ± 1.6 X 10 6 t ± ± ± 1.1 X ± 1.3 X ± 8.6 X ± 2.8 X ± 1.18 X 10 7 * 1.78 ± 1.51 X 10 7 f 1.97 ± 0.71 X ± ± 8.16 X ± 3-1 X ± 0.75 X ± 5.34 X ± 1.34 X ± 5.42 X ± 2.08 X ± 1.1 X ± 1.69 X ± 3.51 X ± 0.52 X ± 0.94 x 10 7 * 2.48 ± 1.06 X 10 7 * 4.46 ± 5.47 X ± 5.24 X ± ± 1.3 x ± 3.07 X ± 6.2 X ± 6.6 X ± 4.17 X 10 6 In all eyes the corneal epithelium was removed 16 hours after inoculation, and the treatment eyes had received 5 hours of topical drops. Each number set represents the mean ± SEM derived from six eyes, except in the controls, where *n = 12 or \n = 11 eyes. {S}, fully sensitive; {I}, intermediate sensitivity; {R}, resistant.

5 16 Guzek et al. IOVS, October 1998, Vol. 39, No. 11 To the best of our knowledge, this is the first report comparing the effectiveness of various antibiotics against penicillin-resistant 5. pneumoniae keratitis in vivo. Penicillin and (almost equally) gentamicin sterilized S. pneumoniae "S" keratitis. Penicillin dramatically reduced the 5. pneumoniae "I" CFUs/cornea by two orders of magnitude (from 1.63 ± 1.18 X 10 7 CFUs/cornea to 1.97 ± 0.71 X 10 5 CFUs/cornea). In contrast, gentamicin sterilized all six corneas. The resistant S. pneumoniae "R" did not respond to topical penicillin (from 2.31 ± 0.94 X 10 6 CFUs/cornea to 4.46 ± 3.47 X 10 6 CFUs/cornea), whereas with gentamicin the CFUs/cornea decreased to 53 ± 103. Gentamicin alone or in combination with vancomycin sterilized all six corneas. This combination is almost statistically equivalent to the use of gentamicin alone against S. pneumoniae "R" (P = 0.07). Standard pharmacology texts recommend against using aminoglycosides to treat S. pneumoniae, whereas one study demonstrated good effect. 11 This paradox can be explained by the low therapeutic index (the ratio between the maximum safe concentration and the required therapeutic concentration) of the aminoglycosides against S. pneumoniae. The sensitivity to gentamicin is markedly dose related: At 5 ng/ml only 9 of 50 (18%) S. pneumoniae strains respond, whereas at 10 ng/ml 48 of 50 (96%) do. 24 ' 25 Tobramycin has even less effective against S. pneumoniae in vitro, with only one organism sensitive at 3 0 mg/ml, three at 12.5 mg/ml, and one more at 25 mg/ml. 25 The optimal concentration is two to three times MIC ( to 30 mg/ml). Unfortunately, serum levels greater than 8 to 10 mg/ml are unacceptably toxic. However, topical gentamicin (13.6 mg/ml every half hour for 2 hours) attains a tissue concentration of 260 mg/g, more than 13 times the MIC. 26 Keratitis, even one caused by a relatively resistant organism, is best treated by topical antibiotics. In vivo antibiotic sensitivity testing should replace in vitro testing, because results from the latter do not automatically apply to the clinical situation. For example, on the basis of in vitro tests, vancomycin has been recommended as the treatment of choice for penicillin-resistant 5. pneumoniae. 8 ' 9 Our study demonstrated it to be statistically very significantly a poorer choice than gentamicin (P < 0.001). The report of 2 patients with poor outcomes further endorses the need for clinically relevant testing. 9 Bacitracin, cefazolin, penicillin G, ceftriaxone, cefotaxime, ofloxacin, and ciprofloxacin were all significantly less effective than gentamicin in vivo (P < 0.001). Furthermore, chloramphenicol appears to be very effective in vitro, but in vivo it was equivalent to saline against S. pneumoniae "I," even at fortified (50 mg/ml) concentrations. It has come to our attention that chloramphenicol is being added to a fluoroquinolone drop in the belief that it will complement the deficiency of fluoroquinolones against 5. pneumoniae. 8 However, the poor response found in vivo in this study suggests that this practice should be abandoned. Two third-generation cephalosporins (ceftriaxone, 50 mg/ml, and cefotaxime, 50 mg/ml) were tested. Although neither preparation has, to the best of our knowledge, been used clinically as eyedrops, they are effective against penicillin-resistant S. pneumoniae in vitro. 27 In vivo, neither drug was better than cefazolin against 5. pneumoniae "I" or "R." A recent report of 8 cases of S. pneumoniae keratitis resistant to fluoroquinolones has a quick response to an aminoglycoside combined with vancomycin or a cephalosporin. 28 If our findings are taken into consideration, it's possible that the aminoglycoside may have been the only effective agent. Although caution must be exercised in applying results from animal-based studies to clinical patient management, the results of in vivo testing appear more appropriate than those obtained in vitro. Appropriate therapy of keratitis from penicillin-sensitive S. pneumoniae includes topical penicillin; cefazolin, or gentamicin, whereas penicillin-resistant 5. pneumoniae requires topical gentamicin alone or in combination with vancomycin. References 1. Defining the public health impact of drug-resistant Streptococcus pneumoniae: report of a working group. Morb Mortal Wkly Rep. 1996;45(suppl):RR-l. 2. Baquero F, Martinez-Beltran J, Loza E. A review of antibiotic resistance patterns of Streptococcus pneumoniae in Europe. / Antimicrob Chemother. 1991;28(suppl C): Marton A, Gulyas M, Munoz R, Tomasz A. Extremely high incidence of antibiotic resistance in clinical isolates of Streptococcus pneumoniae in Hungary. / Infect Dis. 1991;l63: Friedland IR, McCracken GH. Management of infections caused by antibiotic resistant Streptococcus pneumoniae. N Engl J Med. 1994;6: Friedland IR, Klugman KP. Failure of chloramphenicol therapy in penicillin-resistant pneumococcal meningitis. Lancet. 1992;339: Pallares R, Gudiol F, Linares J, et al. Risk factors and response to antibiotic therapy in adults with bacteremic pneumonia caused by penicillin-resistant pneumococci. N Engl J Med. 1987;317: Drug-resistant Streptococcus pneumonia-kentucky and Tennessee, MMWR Morb Mortal Wkly Rep. 1994;43:23-26, Sutphin JE, Pflugfelder SP, Wilhemus KR, Jones DB. Penicillinresistant Streptococcus pneumoniae keratitis. Am J Ophthalmol. 1984;97: Wilkins J, Whitcher JP, Margolis TP. Penicillin-resistant Streptococcus pneumoniae keratitis. Cornea. 1996;13: Iiesegang TJ. Bacterial and fungal keratitis. In: Kaufman HE, Mc- Donald MB, Barren BA, Waltman SR. The Cornea. New York: Churchill Livingstone; 1988: Sande MA, Mandell GL. Antimicrobial agents: the aminoglycosides. In: Gilman AG, Rail TW, Nies AS, Taylor P. The Pharmacological Basis of Therapeutics. 8th edition. New York: Pergamon Press; 1990: Johnson MK, Hobden JA, Hagenah M, O'Callaghan RJO, Hill JM. The role of pneumolysin in ocular infections with Streptococcus pneumoniae. Curr Eye Res. 1990;9:1107-lll Friedland IR, Paris M, Shehon S, McCracken GH. Evaluation of antimicrobial combinations against penicillin-resistant and susceptible Streptococcus pneumoniae. J Antimicrob Chemother. 1994; 34: Osier HB. Diseases of the External Eye and Adnexa. Baltimore: Williams & Wilkins; 1994: Wilhemus KR, Hyndiuk RA, Caldwell DR, et al. 0.3% Ciprofloxacin ophthalmic ointment in the treatment of bacterial keratitis. Arch Ophthalmol. 1993;lll: Leibowitz ILM. Clinical evaluation of ciprofloxacin 0.3% ophthalmic solution for treatment of bacterial keratitis. Amf Ophthalmol. 1991;112(suppl):34S.

6 IOVS, October 1998, Vol. 39, No pneumoniae Keratitis in Rabbit Ormerod D, Hertzmark E, Gomez DS, et al. Epidemiology of microbial keratitis is Southern California. Ophthalmology. 1987;94: Upadhyay MP, Karmacharya PCD, Koirala S, Tuladhar NR. Epidemiologic characteristics, predisposing factors, and etiologic diagnosis of corneal ulceration in Nepal. Am J Ophthalmol. 1991;111: Carmichael T, Walpert M, Koornhof IL. Corneal ulceration at an urban African hospital. Br J Ophthalmol. 1985;69: Osato MS, Jensen HG, Trousdale MD, et al. The comparative in vitro activity of ofloxacin and selected ophthalmic antimicrobial agents against ocular bacterial isolates. Am J Ophthalmol. 1989; 108: Hyndiuk RA, Eiferman RQ, Caldwell DR, et al. Comparison of ciprofloxacin ophthalmic solution 0.3% to fortified tobramycincefazolin in treating bacterial corneal ulcers. Ophthalmology. 1996;103: Smith JA, O'Brien TP, Hahn TW, et al. Comparative efficacy of topical fluoroquinolones and penicillin G against experimental Streptococcus keratitis in rabbits. ARVO Abstract Invest Ophthalmol & Vis Set. 1994;35: Barth G. Animal models of bacterial corneal ulcers. In: Tabbara KF, Cello RM, eds. Animal Models of Ocular Disease. Springfield, IL: Charles C. Thomas Publishers; 1984: WaitzJA, Weinstein MJ. Recent microbiological studies with gentamicin. / Infect Dis. 1969;119: Meyers BR, Hirschman S. Tobramycin. In vitro antibacterial spectrum of a new aminoglycoside. / Clin Pharmacol. 1972;12: Glasser DB, Gardner S, Ellis JG, Pettit TL. Loading doses and extended dosing intervals in topical gentamicin therapy. Am J Ophthalmol. 1985;99: Yee YC, Thornsberry C, Brown SD, Bouchillon SK, Marier JK, Rich I. A comparative study of the in-vitro activity of Cefepime and other anti-microbial agents against penicillin-susceptible and penicillin-resistant Streptococcus pneumoniae. J Antimicrob Chemother. 1993;32(supplB): Hwang DG, Kenyon KR. Base ocular infection treatment on microbiologic findings. Ocular Surg News. 1966;l4:l, 34.