Archipelago keratitis: a clinical variant of recurrent herpetic keratitis?

OBJECTIVE: To describe archipelago keratitis, a presumed clinical variant of herpetic epithelial keratitis. DESIGN: Case series. PARTICIPANTS: A series of 6 patients with an unusual form of superficial keratitis. METHODS: History, including age, gender, clinical evolution, and treatment; slit-lamp biomicroscopy findings; in vivo confocal microscopy findings; and corneal epithelial scrapings were analyzed. MAIN OUTCOME MEASURES: Clinical ocular examination, a diagnostic workup including corneal scraping for herpesvirus polymerase chain reaction, in vivo confocal microscopy, and therapeutic outcome. RESULTS: The authors describe a series of 6 patients with keratitis consisting of foci of epithelial erosions associated with subepithelial nummular inflammatory infiltrates and disposed in a radial, centripetal, archipelagolike pattern originating from the limbus. All the patients had a past history of herpetic epithelial keratitis, herpetic vesicles on the ipsilateral lid, or both. Polymerase chain reaction-based screening for herpes simplex virus 1 and 2 in corneal scrapings demonstrated positive results in 2 patients. In vivo corneal confocal microscopy revealed focal areas of hyperreflective epithelial cells and hyperreflective subepithelial dendritic structures overlying activated keratocytes. All the patients improved with oral valacyclovir treatment followed by topical steroid therapy. CONCLUSIONS: Archipelago keratitis may be a new clinical variant of herpetic keratitis, reflecting herpetic dissemination from the limbus to the center of the cornea.

Successful treatment of Paecilomyces lilacinus keratitis in a patient with a history of herpes simplex virus keratitis.

PURPOSE: To report a case of Paecilomyces lilacinus keratitis, initially misdiagnosed as Penicillium sp., in a patient with a long-standing history of herpes simplex virus (HSV) keratitis. METHODS: A retrospective case report. RESULTS: A 62-year-old man developed P. lilacinus keratitis. He was treated with topical steroids for immune stromal keratitis secondary to HSV before developing the fungal keratitis. Initial corneal cultures were positive for Penicillium sp., but subsequent cultures identified P. lilacinus to be the causative organism. The patient later developed an anterior chamber abscess. Three penetrating keratoplasties, as well as intravitreal injection of amphothericin B, topical miconazole, subconjunctival miconazole, and systemic fluconazole, were required to eradicate the infection. CONCLUSION: To our knowledge, this is a first report of P. lilacinus keratitis in a patient with a previous history of HSV keratitis. The causative organism was initially reported as Penicillium sp. on 2 occasions, before the correct diagnosis was made. Paecilomyces keratitis progressed to an anterior chamber abscess in this eye. Aggressive treatment, including a therapeutic penetrating keratoplasty, intravitreal amphothericin B injection, topical miconazole, and systemic fluconazole can be successful in eradicating this extremely difficult-to-treat infection.

Immunohistochemical identification of trigeminal ganglion neurons that innervate the mouse cornea: relevance to intercellular spread of herpes simplex virus.

Inoculation of the scarified cornea with herpes simplex virus (type 1) leads to herpetic infection of trigeminal ganglion cells. A recent study of the susceptibility of ganglion cells revealed that there may be at least four populations of trigeminal ganglion cells that are infectable by herpes. Two classes were identified by their neuropeptide content: Substance P or calcitonin gene-related peptide. One class was identified by its affinity for a monoclonal antibody, SSEA-3. The fourth class was recognized by its common affinity for both the monoclonal antibody LD2 and for the lectin Bandeiraea simplicifolia isolectin. However, there has been no direct evidence of which types are infected directly as a result of retrograde transport from the corneal site and which may be infected by cell-to-cell spread. The aim of this study was to determine which classes of neurons, which are known to become infected with HSV after ocular inoculation, supply corneal innervation. We have identified four classes of trigeminal ganglion neurons that supply axons to the central cornea of the mouse, on the basis of their ability to transport Fluoro-Gold retrograde from axons in the central corneal epithelium and stroma. About 40% of the neurons that innervate the cornea contain Substance P or calcitonin gene-related peptide; about 60% of the neurons that innervate the cornea react with the monoclonal antibody SSEA-3. About 36% of all neurons in the whole ophthalmic division react with the LD2 or Bandeiraea simplicifolia isolectin, and Fluoro-Gold labels only 2% of them.(ABSTRACT TRUNCATED AT 250 WORDS)

A quantitative assay of retrograde transported HSV in the trigeminal ganglion.

The relationship between the dose of Herpes simplex virus type 1 (HSV) inoculated in the cornea and the amount of actively replicating virus recovered from mouse trigeminal ganglion cells 5 d after corneal scratch and inoculation was investigated with a tissue culture plaque assay. A dose response curve of productive viral replication was obtained. The estimated dose of HSV that produces half-maximal recovery of virus within the ganglion was 9.15 x 10(3) plaque forming units per eye, and the maximal amount of HSV recovered was 1.34 x 10(4) pfu per ganglion. This definition of infectivity as a function of dose will be useful for studying the effects of potential inhibitors of the binding, uptake, and transport of HSV by productively or latently infected trigeminal neurons.

Ocular infection with herpes simplex virus in several strains of rat.

PURPOSE: To assess the suitability of the rat for studies of ocular infection with herpes simplex virus (HSV). METHODS: LEW, AO, DA, PVG, and (DAxLEW)F1 x LEW backcross generation rats, 7 to 9 weeks of age, were inoculated with HSV-1 McKrae. The course of primary disease was assessed by clinical observation using a slit lamp. Infectious virus was assayed in ocular and nervous tissue, and the incidence of latent infection was determined. RESULTS: LEW and AO strains were the most susceptible. All LEW rats died after an inoculum of 4 x 10(2) plaque-forming units (pfu) and developed severe corneal disease and uveitis. In contrast, all PVG rats survived 10(4) pfu, 60% survived 4 x 10(4) pfu, and eye disease was restricted to epithelial lesions, sometimes accompanied by mild stromal haze. This resolved, even in animals that developed central nervous system disease. The DA strain showed intermediate susceptibility. Resistance was dominant because disease in backcross generation (DA x LEW)F1 x LEW rats resembled that of the DA rather than the LEW strain. Resistance appeared to be linked to coat color (P < 0.001) rather than to major histocompatibility complex (MHC) type. Chronic stromal disease did not occur in survivors (DA, PVG, and hybrid strains only). CONCLUSIONS: The susceptibility of rat strains to infection of the cornea with HSV varies, and, as with mice, resistance seems to be controlled by non-MHC genes. Rats may prove useful for immunologic studies. Virus reactivation will be the subject of a future report.

Herpes simplex virus type 1 transcription is not detectable in quiescent human stromal keratitis by in situ hybridization.

PURPOSE: The purpose of this study was to determine whether herpes simplex virus (HSV) transcripts are present in the corneas of patients with chronic herpetic stromal keratitis. METHODS: Corneal buttons from patients with a history of stromal keratitis, but no ongoing active disease, together with positive and negative control tissues, were analyzed by in situ hybridization using single-stranded RNA probes for all three classes of viral lytic cycle transcripts as well as for the latency-associated transcripts (LATs). Tissues also were screened for presence of HSV genomic DNA using the polymerase chain reaction (PCR). RESULTS: HSV DNA was detected in 7 of 13 quiescent corneas by PCR, but no viral transcripts were detected in any of these corneas by in situ hybridization. CONCLUSIONS: At the level of detection afforded by in situ hybridization, HSV persistent in scarred human corneas after stromal keratitis appears to be transcriptionally dormant. This contrasts with the situation in neurons of latently infected sensory ganglia, in which LATs are present at high levels.

Improvement of HSV-1 necrotizing keratitis with amniotic membrane transplantation.

PURPOSE: Stromal herpes simplex virus keratitis (HSK) is an immune-mediated disease. Previous studies have indicated that T cells, neutrophils, and macrophages contribute to the tissue damage in HSK. It has been shown that human amniotic membrane promotes epithelial wound healing and has diverse anti-inflammatory effects. In this study, the effect of amniotic membrane transplantation (AMT) on corneal wound healing and on inflammation in mice with necrotizing HSK was examined. METHODS: BALB/c mice were corneally infected with 10(5) plaque-forming units (PFU) of HSV-1 (KOS strain). In 16 mice that exhibited severe ulcerating HSK, the cornea was covered with a preserved human amniotic membrane as a patch. Corneas in 16 infected mice remained uncovered and served as a control. On days 2 and 7 after surgery, the amniotic membrane was removed (eight mice in each group), the HSV-1-infected cornea was evaluated clinically, and the eye was enucleated. Tissue sections were analyzed histologically for epithelialization and cellular infiltration and immunohistochemically with anti-CD3 mAb to T cells, anti-CD11b mAb to both macrophages and neutrophils, or anti-F4/80 mAb to macrophages. RESULTS: Profound regression of corneal inflammation and rapid closure of epithelial defects were observed clinically within 2 days in the amniotic membrane-covered eyes, whereas HSV-1 keratitis and ulceration progressed in all mice in the control group (P < 0.001). Histologically, corneal edema and inflammatory infiltration, and immunohistochemically the number of CD3(+), CD11b(+), and F4/80(+) cells in the cornea were markedly decreased at 2 and 7 days after amniotic membrane application, compared with the uncovered control corneas (P < 0.001). CONCLUSIONS: AMT promotes rapid epithelialization and reduces stromal inflammation and ulceration in HSV-1 keratitis. AMT in mice with HSV necrotizing stromal keratitis appears to be a useful model for investigating the effect and the action mechanism of human amniotic membrane.

Persistence of herpes simplex virus DNA in rabbit corneal cells.

Corneal cell cultures were established from the corneas of rabbits killed during a period of latency 118 d after ocular infection with the RE strain of herpes simplex virus (HSV). DNA was isolated from frozen cell pellets of 42 cell cultures that did not develop viral cytopathic effects during 44 d in culture. Using the polymerase chain reaction (PCR) to amplify HSV thymidine kinase (TK) gene sequences, HSV-specific DNA was detected in 15 of 42 culture-negative cell cultures. Subsequent reamplification, using nested primers that were complementary to HSV TK sequences internal to the orginal primers, resulted in eight additional culture-negative samples showing positive hybridization for HSV TK DNA. Twenty three of the 42 virus culture-negative corneal cell cultures tested by PCR were found to contain HSV genetic material. Detailed examination of the clinical histories of the eyes from which the corneal cultures were obtained showed no correlation between increased frequency or severity of epithelial disease, stromal disease, or virus shedding and more frequent isolation of virus or detection of HSV-specific DNA. These studies document that HSV DNA residues in the corneas of HSV-infected rabbits up to 118 d post-infection. About 10% of the eyes contained virus that could be reactivated in culture, whereas an additional 55% of the eyes contained DNA sequences homologous to a portion of the HSV TK gene.

Staining characteristics and antiviral activity of sulforhodamine B and lissamine green B.

PURPOSE: Fluorescein and rose bengal are dyes used routinely in the examination of the ocular surface. As part of an ongoing search for a superior ophthalmic dye with optimal specificity and sensitivity and a lack of interference with subsequent viral cultures, and as part of studies that use chemical dyes to understand better the pathophysiology of ocular surface disorders, the staining characteristics and antiviral activity of sulforhodamine B and lissamine green B were investigated. METHODS: Staining of rabbit corneal epithelial cell cultures by sulforhodamine B and lissamine green B was compared to that of fluorescein and rose bengal. Diffusion of each dye through a collagen gel was measured. Uptake of lissamine green B by herpes simplex virus type 1 (HSV-1)-infected Vero cell cultures was compared at several times postinfection. The effect of sulforhodamine B and lissamine green B on HSV-1 plaque formation in Vero cells was determined. The cellular toxicity of sulforhodamine B and lissamine green B in vitro was examined by a quantitative 14C-amino acid uptake assay and by a qualitative cell viability assay. Finally, the effect of sulforhodamine B and lissamine green B on viral replication was compared in vivo with that of rose bengal in a rabbit model of herpetic epithelial keratitis. RESULTS: Rose bengal vividly stained cell monolayers of explant cultures of rabbit corneal epithelium. By light microscopy, sulforhodamine B and lissamine green B, like fluorescein, did not stain the epithelial cells, but did stain the corneal explant stroma. Pretreatment of epithelial cells with 0.25% trypsin for 5 minutes failed to induce dye uptake; however, pretreatment with 0.5% Triton X-100 for 5 minutes resulted in nuclear staining by lissamine green B, but not sulforhodamine B. When added to a collagen gel, the relative diffusion rate was fluorescein > lissamine green B > sulforhodamine B > rose bengal. By spectrophotometric analysis, HSV-1-infected and uninfected Vero cells bound equivalent amounts of lissamine green B until late in infection, when infected cells took up more dye (P < 0.001). A direct neutralization assay showed that 0.06% lissamine green B or 0.5% sulforhodamine B reduced HSV-1 plaque formation in Vero cells by greater than 50%, when present at the time of viral adsorption. By a quantitative 14C-amino acid uptake assay, lissamine green B was toxic to Vero cells in a dose-dependent manner, whereas sulforhodamine B was relatively nontoxic at the concentrations tested. By a cell viability assay, however, neither dye showed significant cellular toxicity. In a rabbit model of herpetic epithelial keratitis, rose bengal significantly reduced viral replication and recovery, whereas sulforhodamine B and lissamine green B had no effect. CONCLUSIONS: Neither sulforhodamine B nor lissamine green B stain healthy, normal cells. Lissamine green B stains membrane-damaged epithelial cells, but sulforhodamine B does not. Both sulforhodamine B and lissamine green B stain corneal stroma. Lissamine green B inhibits HSV-1 plaque formation at low concentrations of dye in vitro, which correlates with suppression of cellular metabolism as demonstrated by a 14C-amino acid uptake assay, but does not affect cell viability. Neither sulforhodamine B nor lissamine green B inhibit viral replication or recovery in vivo.

Corneal Langerhans cell dynamics after herpes simplex virus reactivation.

PURPOSE: The authors investigated the progressive changes in the distribution of corneal Langerhans cells (LC) after reactivation of latent herpes simplex virus type 1 (HSV-1) in mice. METHODS: After corneal inoculation of National Institutes of Health inbred mice with HSV-1 and the establishment of latency, viral reactivation was induced by irradiating the ocular surface with 250 mJ/cm2 of ultraviolet B (UV-B) light. RESULTS: Subsequent viral replication in the cornea was followed by the migration of the LC toward the paracentral and central corneal epithelium. These areas are normally devoid of LC. The number of LC in the paracentral and central regions of the eye reached a peak at day 14 post-UV-B irradiation. After UV-B irradiation of mice latently infected with HSV-1, the development of corneal stromal opacification and neovascularization closely followed the migration of LC toward the central cornea and paralleled the influx of T-cells into the corneal stroma. This pattern was not observed in irradiated uninfected mice. CONCLUSIONS: LC migrate centrally in the corneal epithelium after viral reactivation. There is a direct correlation between the number of LC in the cornea and the degree of persistent stromal opacification.

Deletion of the herpes simplex virus type 1 ribonucleotide reductase gene alters virulence and latency in vivo.

The role of herpes simplex virus type 1 (HSV-1)-encoded ribonucleotide reductase (RR) has been investigated in mice and guinea pigs using a mutant from which 90% of the large subunit of the enzyme was deleted. The RR mutant was extremely impaired in its ability to induce external vaginal lesions or to cause death in mice following intracerebral, intraperitoneal, or intravaginal inoculation, or in guinea pigs following intraperitoneal or intravaginal inoculation. The RR mutant replicated poorly in the vagina of mice and guinea pigs when compared with the parental virus. Neither infectious nor latent virus was recovered from the trigeminal ganglia of mice or from the dorsal root ganglia of mice and guinea pigs after inoculation with the RR mutant. Using the polymerase chain reaction, RR mutant DNA was, nevertheless, detected in the dorsal root ganglia of guinea pigs. These studies suggest that HSV-1 RR is essential for virulence and may also play a role in the recovery of reactivatable latent virus from ganglia in both mice and guinea pigs.

Evaluation of 1-(2'-deoxy-2'-fluoro-beta-D-arabinofuranosyl)-5-ethyluracil in a rabbit model of herpetic keratitis.

The nucleoside analog 1-(2'-deoxy-2'-fluoro-beta-D-arabinofuranosyl)-5- ethyluracil (FEAU) was tested in a rabbit model of acute herpetic keratitis and its effectiveness compared with that of acyclovir (ACV). FEAU or ACV was applied topically 3 times daily, beginning 3 days post-HSV-1 inoculation and continued for a period of 7 days. FEAU at a concentration of 1% (w/v) or 3% ACV resulted in significant lessening of the severity of corneal lesions, conjunctivitis, iritis, and corneal clouding at 24 to 48 h after beginning chemotherapy. No toxic reaction was observed in any rabbit eyes treated with either FEAU or ACV. The duration of virus shedding into tear film and colonization of the trigeminal ganglia, however, were not reduced by either FEAU or ACV treatment begun 3 days post-inoculation. Fifty percent effective dose (ED50) of FEAU determinations performed on isolates from tear film and on the virus inoculum in secondary rabbit kidney cultures yielded a range of 4.6-7 microM, with two in vitro resistant isolates having ED50S of greater than or equal to 1500 microM of FEAU. Fifty percent cell growth inhibition for FEAU was 3000 microM at 72 h.

Effects of phosphonylmethoxyalkyl-purine and -pyrimidine derivatives on TK+ and TK- HSV-1 keratitis in rabbits.

The phosphonylmethoxyalkyl derivatives HPMPA [(S)-9-(3-hydroxy-2-phosphonylmethoxypropyl)adenine], HPMPC [(S)-1-(3-hydroxy-2-phosphonylmethoxypropyl)cytosine] and PMEA [9-(2-phosphonylmethoxyethyl)adenine] were evaluated as 0.2% eyedrops for their efficacy in the treatment of experimental herpes simplex virus type 1 (HSV-1) keratitis in the rabbit model. BVDU 0.2% eyedrops were used as the reference treatment. HPMPA, HPMPC, PMEA and BVDU eyedrops showed a rapid and highly significant healing effect (P less than 0.005) on keratitis caused by TK+ HSV-1 (McIntyre strain) when compared with placebo eyedrops, whereas BVDU treatment did not affect the course of TK- HSV-1 (VMW-1837) keratitis. HPMPA and HPMPC treatment again caused a highly significant healing (P less than 0.005, compared with placebo eyedrops). Although PMEA eyedrops were less effective than HPMPA or HPMPC eyedrops, the effect of PMEA eyedrops was significantly (P less than 0.05) different from the effect of either BVDU or placebo eyedrops.

Potentiating effect of ribavirin on the anti-herpes activity of acyclovir.

The combined antiviral effects of acyclovir (ACV) and ribavirin (Rbv) on herpes simplex virus type 1 (HSV-1) and pseudorabies virus (PRV) in cell cultures and on experimental HSV-1 keratitis in rabbits were studied. The antiviral activity in vitro was based on cytopathogenicity inhibition and yield reduction. The combination of the two drugs exhibited synergy as evaluated graphically (isobolograms). Rbv also potentiated the antiviral effect of ACV in vivo, in the experimental HSV-1 keratitis model in rabbits. This was evident from both the severity of corneal lesions and virus shedding in the tear film. The potentiating effect of Rbv on the anti-HSV-1 activity of ACV in vitro was reversed by guanosine.

Herpetic keratitis: persistence of viral particles despite topical and systemic antiviral therapy. Report of two cases and review of the literature.

OBJECTIVE: First, to characterize the histologic features of corneal buttons taken from two patients with chronic active herpetic stromal keratitis. Both eyes had suffered frequent and prolonged viral epithelial recurrences despite topical and systemic antiviral therapy and developed uniquely rapid deposition of chalklike stromal deposits. Second, to determine the clinical outcome of surgical intervention in eyes with such a pattern of herpetic disease. DESIGN: Patients received topical antiviral medication and 200 to 400 mg of acyclovir five times daily for 2 or 5 months until penetrating keratoplasty. They received tapered doses of acyclovir after surgery. Corneal buttons were evaluated with light microscopy and electron microscopy. RESULTS: Light microscopy of the specimens revealed calcium in the area of the chalklike deposits and a few cocci in the deep stroma. Electron microscopy showed numerous herpetic viral particles at various stages of maturity, including completely enveloped organisms, in the basal cells and keratocytes, and a few cocci in basal cells. Apart from one minor recurrence of a dendritic ulcer, both patients were free of herpetic disease at 13 and 22 months, required little to no medication, and had clear grafts. CONCLUSIONS: Rapid calcium deposition in herpetic corneas may indicate disease of sufficient severity to warrant surgical intervention for removal of a stromal viral reservoir. Such intervention can stop further recurrences of keratitis that is poorly controlled by antiviral therapy. Such chronically diseased eyes may also harbor unsuspected bacterial infection.

Herpes simplex virus: molecular biology and the possibility of corneal latency.

Herpes simplex virus (HSV) is known to be latent in ganglionic neurons. Over the past eight years, a series of reports have described the isolation of HSV after organ culture of human corneas that had been removed in the course of penetrating keratoplasty. None of the corneas showed any clinical signs of active herpetic disease immediately before keratoplasty. Studies in rabbits and mice confirmed that HSV can be recovered from corneas by organ culture long after primary infection has subsided. Recently, sophisticated techniques of molecular biology, such as specific DNA or RNA probes, have been used to detect HSV nucleic acids in the cornea. The crux of the matter is whether the virus recovered from or detected in the cornea is 1) truly latent in cell populations that are nonneuronal; 2) resident in the cornea, replicating at a slow rate; or 3) newly arrived in the cornea following ganglionic reactivation. The evidence suggests that a guarded case can be made for limited HSV latency within corneal cells. HSV corneal latency would allow for reactivation, replication, and the immune response to occur in the absence of ganglionic HSV reactivation. Such a localized phenomenon has not, however, been demonstrated to occur clinically.

The proportion of galanin-immunoreactive neurons in mouse trigeminal ganglia is transiently increased following corneal inoculation of herpes simplex virus type-1.

To investigate whether neurobiological functions are modified by viral infection, we inoculated mouse corneas with herpes simplex virus type 1 (HSV-1) and examined neuronal galanin content in trigeminal ganglia at selected survival times. HSV-1 antigen appeared in neurons at day 3, peaked at day 6 and disappeared by day 11. Increased galanin positivity was first seen at day 6, peaked at day 10 and approached control values by day 21. This result provides further evidence that the biological program of peripheral sensory neurons is modified by herpes-virus infection.

Demonstration of herpes simplex virus DNA in idiopathic corneal endotheliopathy.

A 56-year-old man developed idiopathic corneal endotheliopathy. The lesion consisted of severe stromal edema at the lower half of the cornea along with a number of associated keratic precipitates and steadily progressed to the upper half of the cornea. By polymerase chain reaction, herpes simplex virus DNA was demonstrated in the aqueous humor of this patient. Corneal stromal edema was resolved in response to treatment with topically applied and systemic acyclovir. Herpes simplex virus DNA was repeatedly demonstrated in the aqueous humor when the endothelial lesion recurred later. This evidence strongly indicates that this unique endothelial disorder is of viral origin.

Detection of herpes simplex virus DNA in human tear film by the polymerase chain reaction.

We investigated the use of the polymerase chain reaction for detecting genomes of herpes simplex virus, varicella-zoster virus, and cytomegalovirus from tear film of patients with clinically diagnosed herpes simplex virus keratitis. Using the polymerase chain reaction with a herpes simplex virus detection sensitivity adjusted to 1.0 plaque-forming units/ml, we detected herpes simplex virus genomic sequences in 12 of 12 epithelial keratitis specimens, two of six stromal keratitis specimens, but in none of 20 normal specimens. Neither varicella-zoster virus nor cytomegalovirus genomic sequences were detected in any sample. These results suggest that polymerase chain reaction quickly performed with reduced sensitivity is useful as a diagnostic tool for confirming clinical observations.

Reactivation of latent herpes simplex virus by excimer laser photokeratectomy.

We tested whether excimer laser photorefractive and phototherapeutic keratectomy may reactivate latent herpes simplex and cause recurrent keratitis in mice. Two of ten latently infected mice that were treated with ten excimer laser pulses to the corneal epithelium shed herpes simplex virus type 1, as did four of ten mice that were treated with 50 excimer laser pulses. Ocular shedding of herpes simplex virus was detected in four of ten mice that were treated with ethylenediamine-tetraacetic acid (EDTA) scraping of the corneal epithelium without laser keratectomy, and in six of ten mice on which combined EDTA-facilitated epithelial removal was performed followed by the application of ten excimer laser pulses. In both EDTA-treated groups, viral shedding was prolonged and 18 of 20 mice developed marked corneal opacification or neovascularization, or both. Corneal photoablation with the excimer laser may induce reactivation of latent herpes simplex virus, even in mice with clear and smooth-appearing corneas, and should be considered in the differential diagnosis of humans with persistent corneal epithelial defects after refractive or therapeutic excimer procedures.