An experimental model of mycobacterial infection under corneal flaps

In order to develop a new experimental animal model of infection with Mycobacterium chelonae in keratomileusis, we conducted a doubleblind prospective study on 24 adult male New Zealand rabbits. One eye of each rabbit was submitted to automatic lamellar keratotomy with the automatic corneal shaper under general anesthesia. Eyes were immunosuppressed by a single local injection of methyl prednisolone. Twelve animals were inoculated into the keratomileusis interface with 1 μl of 106 heat-inactivated bacteria (heat-inactivated inoculum controls) and 12 with 1 μl of 106 live bacteria. Trimethoprim drops (0.1%, w/v) were used as prophylaxis for the surgical procedure every 4 h (50 μl, qid). Animals were examined by 2 observers under a slit lamp on the 1st, 3rd, 5th, 7th, 11th, 16th, and 23rd postoperative days. Slit lamp photographs were taken to document clinical signs. Animals were sacrificed when corneal disease was detected and corneal samples were taken for microbiological analysis. Eleven of 12 experimental rabbits developed corneal disease, and M. chelonae could be isolated from nine rabbits. Eleven of the 12 controls receiving a heat-inactivated inoculum did not develop corneal disease. M. chelonae was not isolated from any of the control rabbits receiving a heat-inactivated inoculum, or from the healthy cornea of control rabbits. Corneal infection by M. chelonae was successfully induced in rabbits submitted to keratomileusis. To our knowledge, this is the first animal model of M. chelonae infection following corneal flaps for refractive surgery to be described in the literature and can be used for the analysis of therapeutic responses. Correspondence


Introduction
Non-tuberculous mycobacteria comprise more than 90 species and are ubiquitous microorganisms widely distributed in the environment.They are classified as rapidly growing mycobacteria, which grow in less than 7 days, and slowly growing mycobac-teria, requiring culture for periods of 2-4 weeks (1).
Mycobacterium fortuitum, M. abscessus and M. chelonae are rapidly growing mycobacteria recognized as pathogenic to the eye.Turner and Stinson (2) described the first corneal infection caused by M. fortuitum after removal of a foreign body in 1965.
Keratitis caused by M. chelonae in patients undergoing laser in situ keratomileusis (LASIK) has only been described recently (3,4).
Infectious keratitis is a severe complication of LASIK that can potentially reduce the best final visual acuity.An incidence of infectious keratitis of 1:5000 was reported among patients submitted to LASIK, although the real incidence is unknown (5).With the growing number of refractive surgeries, an increase in the incidence of keratitis caused by non-tuberculous mycobacteria is expected.
M. chelonae is an opportunistic pathogen and in most cases non-tuberculous mycobacterial keratitis is related to exogenous inoculation of the microorganism.Since the LASIK interface confines the causative agent to replication under the suboptimal conditions of the stroma (6), it could be questioned whether keratomileusis itself would favor this infection.
Most patients with this infection have a predisposing injury (including surgery) and there is frequently a latent period of 2-8 weeks after corneal insult before keratitis appears (7).A clinical presentation of infectious crystalline keratopathy with an indolent, slowly progressive course and with graywhitish, stellate, round or needle-like opacities can also be observed (6).Severe necrosis may sometimes be found (2).
Treatment of infectious keratitis may be difficult and response to therapy is slow due to delayed diagnosis, inadequate drug penetration and resistance to most conventional antibiotics (3,7,8).
We describe a new experimental model of M. chelonae keratitis in the eyes of rabbits undergoing automatic lamellar keratotomy.This model can be used for studies on mycobacterial keratitis, as well as for the analysis of therapeutic responses.

Material and Methods
Surgery, inoculations, clinical evaluation, and cultures were performed in a prospective and double-blind fashion.

Animals
Three experiments, each involving eight adult male New Zealand white rabbits each weighing 3.5 to 4.0 kg, were performed.Rabbits were maintained according to the Institutional Guidelines and the Statement for the Use of Animals in Ophthalmologic and Vision Research.Rabbits underwent previous biomicroscopy examination.
Bacteria in the initial log phase (4 days of culture in 7H9-OADC at 30°C in a shaking incubator) were prepared for inoculation.Turbidity of the bacterial suspension was visually compared to tube 5 of McFarland's scale (1.5 x 10 9 bacteria/ml) and absorbance at 600 nm was determined with a spectrophotometer (Hitachi ® -U2000).One aliquot of each suspension was plated onto LB agar immediately after inoculation of the rabbits and the plates were incubated at 30°C.The bacterial concentration of the inoculum was determined on the basis of the number of colony forming units.A control inoculum was prepared in the same way, but the bacteria were heat-inactivated for 20 min at 80°C.After inoculation of the heat-inactivated control, 1 µl of the suspension was plated onto LB agar to confirm bacterial inactivity.

Surgical procedure
Animals were anesthetized by intramuscular injections of xylazine (5 mg/kg) and ketamine (25 mg/kg).The left eye also received topical anesthesia with 0.5% proparacaine hydrochloride.Local antisepsis was performed with 10% povidone-iodine (on the fur) and 5% (on the eyes).Fur and lashes were isolated with a sterile plastic sheet.The eye was proptosed and rinsed with balanced saline solution.A line was traced on the corneal surface using a gentian violet pencil in order to aid the repositioning of the keratomileusis flap.A nasal-based corneal flap with a diameter of 9.5 mm and a thickness of 160 µm was created using an automated microkeratome -Automatic Corneal Shaper (ACS No. 196, Chiron-Bausch & Lomb, Irvine, CA, USA).The flap was lifted using a spatula and the posterior face of the disk and the bed was dried with a polyvinyl acetate sponge.

Infection of the operated eyes
One microliter containing approximately 10 6 colony forming units of live or heatinactivated M. chelonae was placed into the keratomileusis interface.The suspensions were randomly distributed in a blind fashion during the experiment.After disk reposition and surface drying, subconjunctival injection of 20 mg methyl prednisolone acetate was performed to promote local immunosuppression.One drop of a wide spectrum antibiotic (0.1% trimethoprim (w/v) without preservatives) was applied immediately after the surgical procedure and every 4 h during the postoperative follow-up period for antibiotic prophylaxis.Susceptibility tests confirmed that the M. chelonae strain used in this study was resistant to this antibiotic at this concentration.
Animals were sacrificed when clinical signs of disease were detected (corneal infiltrates, edema, ulcer).Specimens were processed for laboratory work-up on the same day.

Cornea laboratory work-up
Removal of the corneal button was performed under aseptic conditions.The eye was proptosed, and before trephination the corneal flap was lifted and the stromal bed was scraped with a disposable blade for smears and for culture analysis.The flap was repositioned and a corneal trephination was performed using a disposable trephine 12.0 mm in diameter up to 2/3 of the corneal thickness.The anterior chamber was approached with a disposable blade and entrance was completed up to its total thickness with scissors.The button was placed in 2.0 ml 7H9-OADC culture medium and immediately processed.
One animal in each experiment was chosen in a random fashion to have its contralateral cornea (without surgery and without disease) recovered to serve as negative control (healthy contralateral eye of infected rabbits).

Specimen processing
All procedures were carried out under a laminar flow, using sterile gloves.All instru-ments were changed between samples in order to avoid cross-contamination.Corneal buttons were cut up into small pieces, placed in a porcelain container and then macerated in 2.0 ml 7H9-OADC.One aliquot of the suspension was diluted 1:10 in 7H9-OADC.Non-diluted and diluted aliquots were then plated onto LB agar containing antimicrobial agents to avoid secondary contamination (vancomycin hydrochloride (500 mg, 6 µg/ml; Eurofarma, São Paulo, SP, Brazil), polymixin B sulfate (5000 IU, 15 µg/ml; Sigma-Aldrich Co.), and amphotericin B (Fungizone ® , 50 mg, 5 µg/ml; Bristol-Myers Squibb) and incubated at 30ºC until the appearance of colonies.Non-diluted and diluted aliquots were maintained in 7H9-OADC.The presence of mycobacteria was confirmed by Ziehl-Neelsen staining and PCR-restriction enzyme analysis.

Results
The initial biomicroscopic corneal aspect was that of a mild lamellar keratitis (Figure 1A).Clinical findings evolved to the appearance of small and fine dots ("sands of Sahara" aspect) (Figure 1B) or of a "cracked windshield" lesion (Figure 1C) without surrounding edema.A corneal abscess with feathery margins was also observed under a slit lamp (Figure 1D).Disease started on the 5th postoperative day.Eleven experimental rabbits developed clinical corneal disease and one died before developing corneal disease.Eleven control rabbits receiving heatinactivated inoculum did not develop corneal disease.One rabbit in this group had a secondary infection with Gram-positive cocci and fungi.No healthy cornea control rabbit developed clinical disease.M. chelonae was isolated from 9 of 11 experimental rabbits, but not from any of the heat-inactivated inoculum control rabbits nor from any healthy cornea control rabbits.
Table 1 shows the clinical and laboratory results obtained in three series of experiments.A, Corneal interface with an initial, mild lamellar keratitis.B, Small and fine dots ("sands of Sahara" appearance).C, "Cracked windshield" appearance.D, Corneal abscess with feathery margins.

Discussion
LASIK is accepted as an effective surgical procedure for the correction of refractive errors (10).The rapid recovery of vision and predictability are advantages of this technique.However, complications such as infectious keratitis, although believed to be rare, have been reported in the literature and are considered to be sight-threatening (5,11).Mycobacteria, previously regarded as unusual organisms, have been recently reported as causative agents of outbreak infections following LASIK (8,12).The occurrence of keratitis in clusters following LASIK, an elective surgery with good prognosis, may be considered to be a very severe complication.
Available data about infection with the M. chelonae-abscessus and M. fortuitum groups are confusing, since these bacteria were previously described as belonging to the same subspecies or as being biovariants (13).With the recent introduction and evolution of molecular techniques, there is strong evidence that these are, in fact, two separate species (14).However, when analyzing previous studies, it is difficult to recognize and establish which species was, in fact, responsible for the disease, particularly in reports written before 1990 (14).
In the past, most experimental studies involving infection by atypical mycobacteria in rabbit eyes were performed using different strains of M. fortuitum.This species was probably chosen because of its strong pathogenicity to man and/or its frequent involvement in disease at that time, or even due to the lack of sophisticated methods of identification to allow separation of groups into different species.
Rabbits have been frequently used as experimental animals in ophthalmology (15) because of their eye structure (similar to the human eye), cornea size, and capability of being examined by slit lamp.In addition, the rabbit cornea can be submitted to keratomil-eusis.There are some difficulties in producing mycobacterial keratitis in this animal model, possibly due to their natural resistance.It has been known that rabbits are in fact resistant to infection by atypical mycobacteria which are pathogenic to man (16,17).According to Kushner et al. (18), non-tuberculous mycobacteria are virulent to mice, less so to rabbits and will not infect guinea pigs.Several attempts to develop an animal model of non-tuberculous mycobacterial keratitis have been unsuccessful.In 1966, Levenson and Harrison (16) reported an apparent high degree of resistance of the rabbit cornea to infection by M. fortuitum isolates from their patients.
Turner (17) reported some of the causative factors for the formation of corneal ulcer by M. fortuitum in rabbits.Using different methods to diminish corneal resistance prior to inoculation, as well as different techniques of inoculation, he observed that it was necessary to modify host resistance or to increase the virulence of the bacilli.He reported that pretreatment with topical corticosteroids 2 weeks prior to inoculation facilitated the development of keratitis.In 1992, Paschal et al. (19) observed that a single subconjunctival injection of methylprednisolone, an intermediate-acting corticosteroid, at the time of mycobacteria inoculation was enough to cause immunosuppression and induce indolent keratitis, and that treatment with corticosteroids before inoculation was not necessary.This approach was followed in the present study.
An important aspect of this animal model is related to the inoculum.In pilot studies, it was observed that the viability of this clinical strain of M. chelonae was reduced 2 h after thawing (data not shown).For this reason, inoculum suspensions were prepared 1 h before inoculation from freshly cultivated bacteria.
In order to avoid waste during flap reposition, we inoculated 10 6 bacteria/µl.Turner Since trauma and surgery seem to be factors predisposing to M. chelonae infection (24), in the present study we investigated the possibility that during LASIK there is mechanical trauma and inoculation of bac-teria into the interface may occur.In previous reports, inoculation was carried out by intrastromal injection with a Hamilton syringe, but keratomileusis was not simulated.
The clinical appearance of the keratitis observed in this animal model was similar to that described in human disease (7,8): well demarcated or irregular edges with a snowflake or "cracked windshield" appearance, along with multiple, fine, small opacities in some animals.However, epithelial defects, described by Broadway et al. (7) as being universal, were not observed in this study, or in studies by other authors (8).
In conclusion, although infectious keratitis after LASIK is rare, M. chelonae is the most frequently reported species among the non-tuberculous mycobacteria involved in keratitis.It is a severe condition, in which diagnosis may be delayed and treatment is difficult.
We describe here an experimental animal model with great potential of producing infectious keratitis by M. chelonae following keratomileusis.Regardless of some differences between disease in the rabbit model and in humans, this model can be used in studies on the natural course and treatment of keratitis, in order to obtain a better prognosis for this severe and sight-threatening condition.

Figure 1 .
Figure 1.Clinical aspect of corneal disease by Mycobacterium chelonae in rabbits submitted to automatic lamellar keratotomy.A, Corneal interface with an initial, mild lamellar keratitis.B, Small and fine dots ("sands of Sahara" appearance).C, "Cracked windshield" appearance.D, Corneal abscess with feathery margins.

Table 1 .
Clinical and laboratory results obtained with the model of Mycobacterium chelonae infection of rabbits described in the present study.