Rationale for Adjunctive Topical Corticosteroids in Bacterial Keratitis

Bacterial keratitis is a major cause of vision loss worldwide.¹ Despite the development of newer broad-spectrum antibiotics, the scope of the problem appears to have grown in the United States, where the incidence of ulcerative keratitis was found in a community-based study to have increased by 435% from 1950 to 1988.² This rise may have been secondary to prior underreporting of bacterial keratitis but notably also occurred during a period of growth of 1 important risk factor: contact lens wear.³

Data that are more recent indicate that bacterial keratitis remains a significant public health problem, particularly among contact lens wearers. In 1990, the estimated total US incidence of bacterial keratitis was 27 000 cases.² Forster⁴ reviewed 28 years (1969-1997) of ocular microbiology files in Miami and noted an incidence rate of 3.1 cases of suspected bacterial keratitis per 1000 clinic visits. A more recent study from Hong Kong identified an annual incidence of microbial keratitis of 0.63 per 10 000 people and 3.4 per 10 000 for contact lens wearers (daily wear lenses, 3.09 per 10 000; extended wear lenses, 9.30 per 10 000).⁵

Despite the early and aggressive use of broad-spectrum antibiotics, the long-term clinical and economic repercussions of bacterial keratitis are profound. Nearly 100 000 Americans have vision-limiting corneal opacities in the aftermath of bacterial keratitis.⁶ Not only can bacterial keratitis cause serious visual disability, but it also represents a significant financial burden. In 1996, McDonnell⁷ estimated that the average medical expenditures to evaluate and treat 1 corneal ulcer successfully was $300. This figure includes the cost of corneal cultures but does not include physicians' fees or surgical intervention for treatment failures. In the United States and Canada, almost 1% of corneal transplantations are performed to treat corneal perforation or sight-limiting opacities due to prior bacterial keratitis, adding significantly to the economic cost of this condition.⁸

Typically, the corneal opacification and scarring seen with bacterial keratitis are caused by both the bacteria and the host inflammatory response induced by the infection.⁹ The use of topical corticosteroids in bacterial keratitis as a means to reduce the inflammatory response and associated visual sequelae has been debated. On careful review of the literature, we believe that the judicious use of corticosteroids can improve outcomes via their potent anti-inflammatory effects when used adjunctively in the treatment of bacterial keratitis. Furthermore, and perhaps more importantly, we feel that the available literature supports the need for a properly executed clinical trial to provide guidance and insight into this controversy, thereby definitively ending the debate.

Most frequently, microbial keratitis is caused by bacteria, but corneal infections can also be caused by fungi and parasites. A 29-year review of suspected infectious keratitis (1969-1997) in Miami revealed a positive culture rate of 49%. The vast majority of isolates (82%) demonstrated a bacterial etiology, followed by fungal (16%) and parasitic (2%) organisms.⁴ In a study by Gudmundsson et al,¹⁰ the bacterial pathogens encountered were Staphylococcus aureus(28%), Pseudomonas aeruginosa(14%), coagulase-negative Staphylococcus(14%), diphtheroids (14%), Streptococcus pneumoniae(12%), and polymicrobial (15%). Similar figures were also seen in a review of 87 cases of microbial keratitis in South Australia.¹¹

Factors that predispose individuals to the development of ulcerative keratitis include prior topical corticosteroid use, contact lens wear, topical antibiotics, trauma, penetrating keratoplasty, bullous keratopathy, frequent administration of topical glaucoma medications, exposure, entropion, corneal scars, and immunosuppression. Furthermore, breakdowns in host defense mechanisms, dry ocular surfaces, and surface irritants can lead to epithelial compromise and opportunistic infection.^{2 ,10} Such predisposing factors have been identified in 84% to 97% of patients with suppurative keratitis.^{10 - 11}

Miedziak et al¹² noted several risk factors for poor outcomes in bacterial keratitis that required eventual penetrating keratoplasty. These included older age, delay in referral to a corneal specialist, treatment with steroids before initial examination, history of ocular surgery, poor visual acuity at initial examination, and lesion characteristics, including large size, presence of perforation or descemetocele, limbal involvement, and hypopyon. Interestingly, patients treated with topical corticosteroids by their primary ophthalmologist were referred later than those not given steroids. This may confound the correlation between poor outcome and prior corticosteroid use. An additional study by Morlet et al¹³ found that patients older than 60 years with large (>5-mm²) culture-positive ulcers were most likely to have primary treatment failures that resulted in increased ulcer size, perforation, and/or resistant organisms.

The pathogenesis and clinical impact of bacterial keratitis can be viewed as a 2-pronged attack against the cornea. The bacteria and its virulence factors set the stage for corneal injury, often in a relatively compromised cornea with underlying risk factors, as mentioned, while the host-derived inflammatory response to the bacteria further undermines corneal integrity in attempts to rid the eye of infection.

The initial event in microbial invasion typically involves adherence of bacteria to injured corneal tissue,^{14 - 15} intact but immature tissue,^{16 - 17} or exposed stroma.^{14 ,18} However, several species of bacteria, including Neisseria gonnorrhoeae,¹⁹ N meningitidis,²⁰ Corynebacterium diphtheriae,²¹ Haemophilus influenzae,²² Listeria monocytogenes,²³ and Shigella,²⁴ have been noted to directly penetrate intact corneal epithelia.^{25 - 26} Some bacteria, such as P aeruginosa, use pili as a virulence factor to attach to corneal epithelial glycoproteins and flagella for motility, which facilitates propagation within the host.²⁷ Microbial formation of a glycocalyx provides a protective environment, facilitating bacterial growth and aggregation, especially in the initial stages of infection.²⁸

Bacterial invasion into and through the epithelial barrier has been demonstrated to occur within an hour of exogenous microbial application to a corneal wound.²⁵ Various bacterial constituents allow for penetration into the corneal stroma. The capsular polysaccharide of gram-negative bacteria may prevent activation of the alternate complement pathway, while bacterial proteases, such as the elastase and alkaline protease, degrade basement membrane laminin, collagen, and extracellular matrix.^{29 - 30} Bacteria subsequently track through the stroma, propagating the infectious process; viable bacteria have been located by special stains to be primarily at the margins or deepest aspects of corneal infiltrates.²⁶ Bacterial lipopolysaccharide or endotoxin, an intrinsic component of the outer cell membrane of gram-negative bacteria, is released on death of the microbe and may induce the greatest tissue destruction in bacterial keratitis, as it is a potent trigger for the host response.³¹

The presence of bacteria in the cornea initiates a cascade of host-derived responses that aim to eradicate the infection. Tumor necrosis factor (TNF) and IL-1 (interleukin 1), produced by corneal epithelial cells and keratocytes as well as invading T cells and macrophages, are among the earliest cytokines produced to answer the bacterial challenge; IL-1 is a potent chemo-attractant for neutrophils, while TNF stimulates immunocompetent cells and induces IL-6 (a B-cell stimulator and acute phase protein) as well as further IL-1 production.³² The combination of these cytokines, along with the upregulated endothelial leukocyte-adhesion molecule 1 and intercellular adhesion molecule 1 (ICAM-1), facilitate neutrophil diapedesis through limbal blood vessels. Endothelial cell adhesion molecule 1 and ICAM-1 receptors on vascular endothelial cells allow neutrophil adherence before diapedesis into tissues.²⁶ Most neutrophils migrate to the infected cornea via the limbal vasculature, while others may enter through a breach in the epithelium via the tear film.³³

Two bacterial virulence factors that tremendously augment this acute inflammatory host response are bacterial endotoxin and small formyl peptide molecules, which are related to the tripeptide N-formyl-methionyl-leucyl-phenylalanine, which is released by bacteria. The lipopolysaccharides of endotoxin, released on bacterial death, bind macrophage and lymphocyte receptors and lead to upregulation of cytokines (IL-1, IL-6, and TNF), lipids (prostaglandin A₂, thromboxane E₂, and platelet-activating factor), and toxic oxide metabolites, which leads to secondary corneal damage. Platelet-activating factor, in particular, induces the activation of various metalloproteinases, which cause inflammatory necrosis and stromal thinning. Endotoxin also exerts its effects by promoting massive neutrophil infiltration into the cornea, contributing to corneal opacification, neutrophil degranulation of inflammatory mediators, and induction of the alternate complement pathway.^{25 - 26 ,31} N-formyl-methionyl-leucyl-phenylalanine has a similar effect on inflammatory induction by enhancing neutrophil activation, chemotaxis, and degranulation.

In a study by Gray and Kreger,³⁴ rabbit corneal response to P aeruginosainfection was documented with gross, light microscopic, and electron microscopic examination. They documented the following sequence of events: degeneration of the epithelial cells, polymorphonuclear leukocyte infiltration, loss of epithelium, degeneration and loss of keratocytes and endothelium, loss of the characteristic weblike pattern of the proteoglycan ground substance, dispersal of ultrastructurally normal collagen fibrils, extensive accumulation followed by degeneration of polymorphonuclear leukocytes, and accumulation of plasma proteins and fibrin in the necrotic area. While there was no loss of collagen in corneas infected by P aeruginosa, there was collagen loss as well as ultrastructurally abnormal collagen fibrils in corneas injected with Clostridium histolyticumcollagenase.³⁴

The goals of treatment in bacterial keratitis are 2-fold. The first goal is to eliminate the infectious organism, thereby limiting the direct toxic effects on cells and tissues. The second is to curb the inflammatory response, which can cause further destruction of corneal cells and tissue. Corticosteroids have a potent anti-inflammatory effect and have been used for this purpose in corneal disease processes, such as the treatment of corneal haze following refractive procedures and the treatment of interstitial keratitis caused by herpes simplex virus.^{35 - 37} However, their use in the treatment of bacterial keratitis has been consistently debated out of concern that the relative immunosuppression induced by corticosteroids will not only decrease inflammation but may also allow for increased bacterial survival and proliferation.

Corticosteroids can be an important adjunct to antibiotic therapy in the treatment of bacterial keratitis by decreasing the amount of inflammation that can lead to subsequent corneal destruction and scarring. A study by Steuhl et al³⁸ demonstrated that rabbits immunized to the 3 key exoenzymes of P aeruginosastill had a level of corneal damage comparable with that of the control, nonimmunized rabbits. Effective immunization of the rabbits leading to an appropriate serum antibody response was confirmed via several methods: (1) the Ouchterlony method; (2) specific radioimmunoassay; and (3) antibody detection in tears. Although the nonimmunized group experienced more intense corneal opacification in the first 24 hours after inoculation with each P aeruginosastrain, there were no differences noted in corneal status (complete loss of transparency) between the immunized and nonimmunized rabbits after 48 hours. This finding highlights the role of the intense host response to certain virulent bacteria in affecting the final outcome of bacterial keratitis and underscores the potential benefit and justification for anti-inflammatory adjunctive therapy in the treatment of this condition.

Corticosteroids affect the inflammatory response through several mechanisms important in the treatment of destructive corneal disease.³⁹ They have been found to reduce the accumulation of prostaglandin E in the eye,^{40 - 41} which has been identified as a mediator of ocular inflammation.⁴² This results in decreased release of vasoactive amines (heparin, serotonin, eosinophil chemotactic factor, bradykinin, etc) from mast cells and basophils. Furthermore, corticosteroids have been shown to constrict blood vessels, reducing vascular permeability and postinflammatory neovascularization.^{43 - 45} The constriction of blood vessels may act to reduce the number of inflammatory cells that can invade corneal tissue, thereby reducing the amount of damage produced by the inflammatory response. Through their inhibitory effect on neovascularization, topical steroids can help to prevent the ingrowth of blood vessels into the cornea that can be sight-limiting after resolution of an acute infection.

In Leibowitz and Kupferman's studies^{46 - 47} on the effect of topical corticosteroids on rabbit corneal inflammatory responses to intrastromal injections of clove oil, topical steroids decreased stromal inflammation by inhibiting the polymorphonuclear leukocyte invasion of the cornea. Clinical and histologic evidence of the anti-inflammatory effect was demonstrated by suppression of the corneal infiltrate, decreased anterior chamber response, and reduction in the number of inflammatory cells invading the cornea.⁴⁸ The effect can be seen rapidly after the administration of topical corticosteroid, with statistically significant inhibition of tear polymorphonuclear leukocyte response within 5 hours of administration.⁴⁹

Both HLA-DR and ICAM-1 are increased in inflamed tissue and play major roles in mediating the host inflammatory response; they are upregulated by the cytokines TNF-α and interferon γ, which are increased in inflammatory responses. When used in the management of ocular surface inflammatory disease, corticosteroid use led to significantly reduced HLA-DR expression on conjunctival cells.⁵⁰ Cytokine-induced expression of ICAM-1 and HLA-DR in the eye was found to be significantly downregulated by corticosteroids.⁵¹

A recent study by Vemuganti et al⁵² demonstrated that infectious keratitis leads to apoptotic cell death of keratocytes in the corneal stroma. Expression of HLA-DR has also been shown to be positively correlated with expression of Fas-Fas ligand antigens and apoptotic markers.⁵³ Therefore, the effect of corticosteroids in downregulating HLA-DR may also inhibit keratocyte apoptosis induced by corneal bacterial infections.

Consideration of the use of corticosteroids in bacterial keratitis must also take into account the potential issues related to specific formulations, concentrations, dosage frequency, route of administration, and timing of administration after onset of infection. McDonald et al⁵⁴ studied the effect of various steroid formulations (dexamethasone alcohol, dexamethasone acetate, and dexamethasone phosphate) on corneal wound healing in rabbits as determined by the anterior chamber pressure necessary to rupture a 6-mm limbal wound previously closed by 2 silk sutures. They noted a linear association of wound tensile strength with time in healthy eyes. No inhibition of healing was noted after 3 days of treatment with any corticosteroid. Alcohol-based and phosphate-based corticosteroids inhibited corneal wound healing at much lower concentrations than did acetate-based corticosteroids. Inhibition from dexamethasone acetate was not seen until 6 days of treatment with 2 drops administered 5 times daily.

Sugar and Chandler⁵⁵ noted that dexamethasone, 0.1%, administered 3 times daily to rabbit partial-thickness linear corneal wounds sutured with two 10-0 nylon sutures causes a decrease in tensile wound strength when administered during the first postoperative week. However, no reduction in corneal tensile wound strength was noted when the corticosteroids were initiated at the third postoperative week. Rabbits were euthanized on postoperative day 21, and tensile strength was determined by the weight necessary to disrupt the wound after suture removal. In a prospective randomized human trial of 40 patients, dexamethasone, 0.1%, administered 4 times daily to patients with corneal ulcers did not adversely affect corneal healing.⁵⁶

The acetate derivatives of various corticosteroids have been shown to be the most effective commercially available anti-inflammatory agents. Increased frequency of application as well as topical instillation (vs subconjunctival) results in a greater reduction in corneal inflammation.⁵⁷ Thus, appropriate choice of corticosteroid formulation and timing of administration may be important factors for maximizing the therapeutic effects of steroids while minimizing potentially negative iatrogenic effects, such as impaired epithelial and stromal wound healing.

Further insights into this debate may be gained from the effective use of topical corticosteroids to treat stromal inflammation in different clinical contexts. Topical corticosteroid use after excimer laser keratectomy results in fewer activated keratocytes and subepithelial fibrous tissue, decreased polymorphonuclear leukocyte accumulation, and less subepithelial haze.^{35 - 36 ,58} Following superficial keratectomy in rabbits, the use of betamethasone decreased corneal haze via its inhibitory effects on keratocyte proliferation and extracellular matrix deposition.⁵⁹ In infectious keratitis, topical steroids have a known benefit for reducing the inflammation that occurs in herpes simplex virus–mediated stromal keratitis.³⁷

A recent 50-year review of publications by Wilhelmus⁶ did not demonstrate a clear-cut beneficial effect of topical corticosteroids on the course of bacterial keratitis. While human studies are limited, many animal studies suggest that topical corticosteroids may have a beneficial effect. Stern and Buttross⁶⁰ advocated the use of topical corticosteroids for bacterial keratitis if appropriate guidelines are followed.

We reviewed a number of animal studies (rats, rabbits, and guinea pigs) of bacterial keratitis involving P aeruginosa, Staphylococcus, S pneumoniae, and various other species.^{61 - 72} Wilhelmus⁶ also comprehensively summarized 24 studies in which corticosteroids and antibiotics were concomitantly started after bacterial inoculation of the cornea, typically within 48 hours. He noted a beneficial effect of topical corticosteroids in 37.5% of studies, a neutral effect in 50%, and an adverse effect in 12.5% (3 series). There were no studies that demonstrated an adverse effect of corticosteroid treatment in cases of bacterial keratitis involving S pneumoniaeor S aureus. In addition, many rabbit studies have demonstrated the efficacy of bactericidal antibiotics in killing the most virulent bacteria, including S aureusand P aeruginosa, when used simultaneously with corticosteroids.^{61 ,63 ,70 - 71}

There may be a prevailing bias against using topical corticosteroids in Pseudomonalkeratitis. However, available evidence to support this bias is inconclusive. Of the 19 studies on P aeruginosaWilhelmus⁶ reviewed, 3 reported adverse outcomes, while 6 studies had beneficial outcomes and 10 had neutral outcomes. However, all experiments with adverse outcomes were flawed for various reasons. One study used an antibiotic to which the P aeruginosastrain was resistant,⁷³ while another study used an inadequate antibiotic dosage.⁷⁴ A third study used a long-acting, subconjunctival injection of a corticosteroid that did not affect the primary healing phase.⁶⁴ The adverse outcome in this case, recurrence of the keratitis, did not occur until several days after discontinuation of antibiotics.

Only a few human studies that evaluated the effect of topical corticosteroids in bacterial keratitis have been conducted.^{56 ,75} Aronson and Moore⁷⁵ provided a descriptive analysis of various causes of central stromal keratitis, including bacterial, viral, and fungal etiologies. They described several patients who did not significantly improve with an antibiotic regimen but subsequently experienced rapid clinical recovery after the initiation of corticosteroids.

Only 1 prospective, randomized trial on human subjects exists, which was performed by Carmichael et al⁵⁶ in South Africa. Patient stratification was skewed in this study: a larger number of patients with deep ulcers and hypopyons received the steroid treatment. Nevertheless, visual acuity outcomes and corneal healing rates were similar between the groups,⁵⁶ suggesting a possible outcome benefit for worse ulcers treated with corticosteroids in addition to antibiotics.

Available evidence indicates that the judicious use of topical corticosteroids combined with the appropriate antibiotic in the treatment of bacterial keratitis can be an effective adjunctive agent for limiting permanent corneal damage. We would like to propose the following important principles for the successful use of corticosteroids in the context of bacterial keratitis:

Bacterial keratitis is a leading cause of visual morbidity and a major public health burden. The long-term sequelae of sight-limiting scarring and penetrating keratoplasty can be difficult to manage and disabling for the patient. Based on the pathogenesis of bacterial keratitis, the use of adjunctive corticosteroids in specific clinical circumstances can be justified. The human evidence, however, is limited at present. A recent Cochrane Systematic Review identified only 1 randomized clinical trial on the adjunctive use of corticosteroids for the treatment of bacterial keratitis, which demonstrated major methodological inadequacies.⁷⁶ A randomized, prospective, multi-center clinical trial is necessary to better understand the impact of topical corticosteroid treatment in the management of bacterial keratitis. This would be challenging, as multiple variables would need to be controlled, including type of organism, severity of infection, antibiotic selection, length of time before treatment, duration of corticosteroid treatment, location of ulcer, and additional patient factors, such as immune status and presence of other medical and ocular conditions.

The debate over the use of topical corticosteroids in bacterial keratitis has continued for decades, and a resolution based on solid scientific grounds would allow physicians to help patients using evidence-based medicine. Previously, similar discussions occurred regarding the use of topical corticosteroids in the treatment of herpes simplex virus stromal keratitis. The Herpetic Eye Disease Study conclusively ended this uncertainty. The best way to end the present debate is to perform a properly executed clinical trial investigating the use of corticosteroids in bacterial keratitis. We expect that such a study would provide definitive evidence to guide clinicians in this contentious matter for the ultimate benefit of the many patients who develop this difficult clinical entity.

Correspondence: Albert S. Jun, MD, PhD, The Wilmer Eye Institute, Maumenee Bldg 317, 600 N Wolfe St, Baltimore, MD 21287 (aljun@jhmi.edu).

Submitted for Publication: November 1, 2005; final revision received February 4, 2008; accepted March 12, 2008.

Financial Disclosure: None reported.

Additional Information: Drs Hindman and Patel contributed equally to this work.