Article Text

Download PDFPDF

Established and emerging ancillary techniques in management of microbial keratitis: a review
  1. Dana Robaei1,2,3,
  2. Nicole Carnt1,
  3. Stephanie Watson1,4
  1. 1Save Sight Institute, University of Sydney, Sydney, New South Wales, Australia
  2. 2Department of Ophthalmology, Westmead Hospital, Sydney, New South Wales, Australia
  3. 3Westmead Millennium Institute for Medical Research, Westmead, New South Wales, Australia
  4. 4Corneal Unit, Sydney Eye Hospital, Sydney, New South Wales, Australia
  1. Correspondence to Dr Dana Robaei, Save Sight Institute, University of Sydney, 8 Macquarie St, Sydney, NSW 2000, Australia; dana.robaei{at}


Microbial keratitis is a sight-threatening condition and an ocular emergency, because of the potential for rapid progression. Intensive topical antimicrobials are the mainstay and the gold standard of treatment for microbial keratitis. However, despite appropriate diagnosis and therapy, treatment failure is still common, and can result in significant morbidity due to corneal perforation and/or scarring. For this reason, clinicians continue to seek novel treatment techniques in order to expand the armamentarium of tools available to manage microbial keratitis, and in doing so improve clinical outcomes. In this review, we examine the evidence for some established, as well as a few emerging ancillary techniques used to manage microbial keratitis. These include topical corticosteroids, corneal collagen cross-linking, intrastromal antimicrobials, amniotic membrane transplantation and miscellaneous other techniques. Of these, collagen cross-linking shows some promise for selected cases of infectious keratitis, although more research in the area is required before it is accepted as mainstream treatment for this potentially blinding condition.

  • Cornea
  • Ocular surface
  • Treatment other
  • Infection

Statistics from


Microbial keratitis is a sight-threatening condition and an ocular emergency, because of the potential for rapid progression.1 Intensive topical antimicrobials (encompassing antibiotics, antifungals and antiseptics) are the mainstay and the gold standard of treatment for microbial keratitis,2 especially in the first phase, in which the goal is to achieve ocular sterilisation. The second phase of microbial keratitis management encompasses measures which promote corneal healing, culminating in an eye that is comfortable, fully epithelialised and uninflamed.2 However, despite appropriate diagnosis and therapy, treatment failure is still anecdotally common, and can occur during both the sterilisation, and the healing phase, resulting in reduced final vision from corneal scarring, and/or the need for additional interventions should corneal perforation occur.

Failure to eradicate the causative organism is an important first aspect of treatment failure, which can be caused by diagnostic delay,3 poor compliance, antimicrobial resistance,4 ,5 polymicrobial aetiology,6 poor antimicrobial penetration into the cornea and/or poor bioavailability.7 Failure to promote corneal healing in the second phase constitutes another important aspect of overall treatment failure. Measures to promote healing include appropriate treatment of inflammation, corneal exposure and ocular surface dryness, as well as minimising corneal toxicity from frequent topical antimicrobial dosing.8 For these reasons, clinicians continue to seek novel treatment techniques in order to expand the armamentarium of tools available to manage microbial keratitis, both in the initial sterilisation phase and in the subsequent healing phase, and in doing so improve visual outcomes and limit such complications as persistent epithelial defects, corneal melting and perforation.

In this review, we examine the evidence for some established, as well as a few emerging ancillary techniques used to manage microbial keratitis. These include topical corticosteroids, corneal collagen cross-linking (CXL), intrastromal antimicrobials, amniotic membrane (AM) transplantation and miscellaneous other techniques. We have not included strategies that are related to management of viral keratitis (eg, due to herpes simplex, herpes zoster or adenovirus), as this vast topic is outside the scope of our review.

Topical corticosteroids

The benefit of adjunctive topical corticosteroids for management of bacterial keratitis has long been a subject of debate in the ophthalmic literature, while corticosteroid use is generally regarded as being deleterious in cases of fungal keratitis,9 and Acanthamoeba keratitis (AK). In the latter group, exposure of Acanthamoeba trophozoites and cysts to corticosteroids is thought to increase the pathogenicity of the organisms,10 and commencement of corticosteroids prior to initiation of antiamoebic therapy has been associated with poorer visual outcomes.3 Nonetheless, there are specific indications for corticosteroids in the management of AK, and these include corneal neovascularisation and scleritis in the setting of adequate antiamoebic therapy.11

The potential benefits of corticosteroids in reducing corneal scarring and improved patient comfort in bacterial keratitis need to be balanced against excessive suppression of the immune response and resultant potentiation of infection.12 ,13 Until recently, the debate was only supported by three small clinical trials, all of which were underpowered to address the question with certainty.14–16 Table 1 summarises these clinical trials, as well as two retrospective case series,17 ,18 which have reported on topical corticosteroids in management of microbial keratitis.

Table 1

The role of topical corticosteroids in management of microbial keratitis

The Steroids for Corneal Ulcer Trial (SCUT) attempted to definitively assess the effect of adjunctive topical corticosteroids on clinical outcomes in patients with bacterial keratitis. SCUT was a randomised, placebo-controlled, double-masked multicentre clinical trial comparing clinical outcomes of bacterial keratitis treated with topical moxifloxacin 0.5% and topical prednisolone phosphate 1% with those of moxifloxacin 0.5% and placebo.19 All participants had at least 48 h of topical moxifloxacin before being randomised to either placebo drops or topical corticosteroids. The timing of commencement of topical corticosteroids was at the discretion of the treating clinician, with a time lag ranging from 48 h to ≥4 days after commencement of topical moxifloxacin. At 3 months, SCUT found no difference in best-spectacle-corrected visual acuity (BSCVA) or epithelial healing in the two groups. However, a subgroup analysis demonstrated benefit of corticosteroid treatment in patients with severe keratitis, that is, those with central corneal ulceration and worst visual acuity at enrolment. The authors of SCUT also published their findings on patients diagnosed with Nocardia keratitis alone, demonstrating that topical corticosteroid use in this subgroup resulted in an enlargement of infiltrate/scar size at 3 months, and advising caution about the use of corticosteroids in these patients.20 Importantly, the susceptibility of Nocardia organisms to the trial antibiotic, moxifloxacin, varied greatly by the strain of Nocardia, ranging from 45% for N pneumonia isolates, to 100% susceptibility for N. farcinica. Indeed, amikacin may be a better choice of antibiotic for Nocardia keratitis, with 98% of the SCUT strains showing susceptibility to it.20

While being an important addition to the ophthalmic literature, SCUT leaves certain questions about adjunctive corticosteroid use unanswered. First, because 97% of patients were enrolled from India, and almost half of them being agricultural workers, the spectrum of causative bacterial organisms was not typical of those found in Western countries. The most common cultured organism in SCUT was Streptococcus pneumonia, accounting for 49% of all isolates.19 This is in contrast to studies from Western countries, where Staphylococcus species are the most prevalent.21–23 In addition, Nocardia species were isolated from 11% of all positive bacterial cultures, while in other previous series, Nocardia is documented as a rare cause of bacterial keratitis.24 ,25 Also, given that few SCUT participants were contact lens wearers, the role of topical corticosteroids in this, largest at-risk group in Western countries, remains unclear.

Second, the impact of topical corticosteroids on visual outcome in bacterial keratitis may depend on the timing of introduction of corticosteroids. A recent analysis of the SCUT data revealed that participants who were treated with topical corticosteroids within 2–3 days of antibiotic therapy had approximately 1-line better visual acuity at 3 months than those given placebo (p=0.01). In patients who had ≥4 days of antibiotic therapy before corticosteroid treatment, the effect was not significant (p=0.14).26 In this subanalysis, ulcer severity did not influence whether there was an improvement with earlier steroid use. Given that these findings were derived from non-prespecified subanalysis, their confirmation is needed in studies designed specifically to examine the optimal timing of steroid therapy in bacterial keratitis. In general terms, non-prespecified or unplanned post-hoc analyses run the risk of finding apparent differences that may be due to nothing more than chance or a coincidence.27

Yet another important variable may be the particular virulence mechanism of the causative organism. Cytotoxicity, that is, the ability of an organism to kill host cells, and invasiveness, that is, the ability of an organism to sequester itself intracellularly, may play differing roles in the pathogenesis of microbial keratitis, and these differing strains may in turn respond differentially to topical corticosteroids. Borkar et al28 recently examined differences in the cytotoxicity and invasiveness of Pseudomonas isolates cultured from patients in the SCUT trial. Percentage cytotoxicity for each corneal isolate was determined by measuring the amount of lactate dehydrogenase (LDH) released into the media by dead or dying host cells after bacterial exposure using a cytotoxicity detection kit (LDH Cytotoxicity Detection Kit; Roche Diagnostics). Values were compared with a positive control (Pseudomonas aeruginosa clinical isolate 6206). Percentage invasiveness was determined by comparing growth from the intracellular lysate of each isolate with growth from P aeruginosa clinical isolate 6294, which was used as a positive control. The authors demonstrated that treatment with topical corticosteroids was associated with significantly more improvement in visual acuity in the invasive subgroup (p=0.04), but was associated with less improvement in vision in the cytotoxic subgroup (p=0.07).28

Finally, if topical corticosteroids can be shown to improve patient comfort, it may still be a worthwhile adjunct to the treatment regime even if final visual acuity is not improved. SCUT did not address this issue of quality of life (QOL), leaving a gap in the literature, which needs to be addressed by future clinical trials.

Corneal collagen cross-linking

Since the publication of the first clinical study of CXL in 2003,29 this treatment is now an established modality for the safe and effective treatment of progressive keratoconus.30 In 2000, Schnitzler et al,31 reported four cases of non-infectious corneal melts that were successfully treated with CXL. A similar report in the veterinary literature also supports the use of CXL for corneal melting in infectious and non-infectious corneal conditions in cats and dogs.32

While riboflavin photosensitisation using ultraviolet light has long been recognised to cause inactivation of pathogens,33 it has only in recent years, been tried as treatment of microbial keratitis in humans. To better distinguish the use of CXL for the treatment of infectious keratitis from that used for progressive keratoconus, the term photoactivated chromophore for infectious keratitis (PACK)-CXL was created at the ninth cross-linking congress in Dublin, Ireland, in 2013.

Table 2 summarises current published studies of PACK-CXL for management of microbial keratitis. Iseli et al34 first reported using PACK-CXL in four cases of microbial keratitis, two due to non-tuberculous Mycobacterium, one due to Fusarium and the last case being culture negative. Since then there have been several case reports,35–37 or small retrospective case series,38–41 examining the role of PACK-CXL in the management of microbial keratitis. These reports have all suffered from significant methodological limitations, which limit their usefulness. Being retrospective studies, they are limited by the accuracy of medical records. In addition, several of the reported cases were culture negative,36 ,38 ,39 casting doubt on the accuracy of infectious aetiology. In all cases, topical antimicrobials had been used before and after the cross-linking procedure, which in the absence of a control group limits any conclusions that can be drawn from these studies.

Table 2

Studies of corneal collagen cross-linking for management of microbial keratitis

Makdoumi et al42 subsequently published a prospective, non-randomised case series, which examined the role of PACK-CXL as the sole treatment for 16 cases of bacterial keratitis, 13 of which had positive corneal cultures. Prior topical antibiotics had not been used in any of the cases, but two cases required supplemental topical antibiotics or other additional therapy to achieve complete epithelial healing after the cross-linking procedure. The indication for supplemental topical antibiotics was progression of the infectious process, as evidenced by increasing corneal infiltrate. A second prospective case series has also been reported by Price et al,43 although concurrent topical antimicrobials were used in all cases.

Vajpayee et al44 recently published a retrospective case–control study, in which they compared the role of PACK-CXL in patients with culture-positive fungal keratitis (n=41). Both groups received topical antifungals, but in addition, cases (n=20) also received PACK-CXL on the day of presentation. Average healing time and best-corrected final visual acuity were similar in the two groups. There were also no statistically significant differences in the rates of complications, namely corneal perforation, need for tectonic keratoplasty or corneal neovascularisation. These findings led the authors to conclude that PACK-CXL had no additional advantage for management of moderate fungal keratitis.

To date, there have been two prospective randomised controlled clinical trials examining the role of PACK-CXL in management of infectious keratitis. Said et al45 recently published their data on 40 patients with infectious keratitis and concurrent corneal melting, half of whom underwent antimicrobial therapy alone (control group), while the other half had PACK-CXL in addition to antimicrobial therapy (treatment group). Both groups consisted of a heterogeneous patient population with bacterial, as well amoebic and fungal keratitis cases present. Both groups also included a few cases of culture-negative keratitis. While the average time to epithelialisation was 7 days shorter in the PACK-CXL group, this difference was not statistically significant. An interesting finding was that of the lower complication rate in the PACK-CXL group. Twenty-one per cent of patients in the control group experienced a severe complication such as corneal perforation or recurrence of infection; none did so in the treatment group.

While this study provides useful information about the safety of PACK-CXL as an adjunctive treatment in the management of a corneal melt in infectious keratitis, the heterogeneity of its patient population means that it is not sufficiently powered to examine the efficacy of PACK-CXL in bacterial, fungal or AK as separate entities. It should also be noted that PACK-CXL effects are limited to the anterior cornea, and therefore may not be suitable for cases of filamentary fungal keratitis where deep penetration into the stroma occurs.

The second randomised clinical trial of PACK-CXL was published by Bamdad et al.46 They reported findings on 32 participants with culture-proven bacterial keratitis. Cases (n=16) underwent collagen cross-linking at day 1, followed by medical therapy, while controls (n=16) underwent medical therapy alone. At day 14, cases had significantly smaller corneal infiltrates and epithelial defects compared with controls; both were statistically significant. The main weaknesses of this study were twofold. First, baseline clinical characteristics of cases and controls were not described, bringing the comparability of the two groups into question. Second, final visual outcomes were not reported, so the clinical significance of the intervention remains uncertain.

Given the shortcomings described above, the use of PACK-CXL in the management of microbial keratitis can only be considered experimental at this stage, although future large-scale clinical trials of single entity microbial keratitis will shed further light on the exact role of PACK-CXL in the management of this potentially blinding condition. It is worth noting that the use of this new technology poses significant costs for clinics and patients. This is particularly true of low-income and middle-income countries, where the cost and availability of a particular procedure may limit its generalised use. Therefore, it is important that future studies clearly define the role of PACK-CXL in the management of microbial keratitis, that is, whether it is to aid corneal sterilisation in the initial phase, or to limit corneal inflammation and secondary corneal melting in the subsequent, healing phase. These studies should be adequately powered to establish the effectiveness of this technique and may require a pilot phase focused on the main outcome to determine a suitable sample size. Finally, they should include final visual acuity and health-related QOL as primary outcomes, so there is clear benefit to patients that outweigh the additional cost of PACK-CXL.

Intrastromal antimicrobials

Intrastromal injection of antimicrobials is a form of targeted drug delivery that provides antimicrobials at high concentrations directly to the site of the infection. This mode of drug delivery has gained recent momentum, especially in the management of fungal keratitis, which in some countries accounts for up to 40% of all cases of infectious keratitis.47 Currently available topical antifungal agents are limited by poor bioavailability and limited ocular penetration, especially in deeper lesions.7 Intrastromal delivery of antifungal agents holds promise for circumventing some of the above limitations. To this end, voriconazole is the most commonly reported antifungal for intrastromal delivery. However, the body of clinical evidence for its use is not strong, and consists mainly of single case reports48–50 and small, or uncontrolled interventional case series.51–55 Other agents, namely caspofungin and amphotericin have also been reported for intrastromal use, but again only in case report format.56 ,57 Table 3 summarises published studies of intrastromal antifungals for fungal keratitis.

Table 3

Studies of intrastromal antifungal agents for fungal keratitis

Intrastromal delivery of antifungals can be carried out either in the clinic at the slit lamp or in the operating theatre. Under strict aseptic technique, the drug is loaded into a 1 mL tuberculin syringe, fitted with a 26-gauge or 30-gauge needle and delivered to the cornea. The needle, with its bevel facing away from the surgeon, is inserted obliquely into clear cornea, to reach adjacent to the infiltrate at the mid-stromal level. This process is repeated at multiple sites around the fungal infiltrate to form a deposit of the drug around the circumference of the lesion.

To date, there has been a single randomised clinical trial comparing the efficacy of intrastromal voriconazole with topical voriconazole.58 In this study, 40 patients with culture-proven fungal keratitis were randomised to receive either topical voriconazole therapy or intrastromal injections of voriconazole. Patients in both groups had already been on 2 weeks of topical natamycin and had shown no clinical improvement. This was continued throughout the clinical trial until epithelial healing was achieved. Intrastromal voriconazole did not offer any beneficial effect over topical therapy. In fact, patients in the intrastromal group had significantly worse best-spectacle-corrected vision after treatment, compared with the topical group, although the authors attribute this difference to more ulcers involving the visual axis in the intrastromal group.

In short, there is currently insufficient evidence to support the use of intrastromal antifungal agents for the management of fungal keratitis. It may be that the above study had insufficient power to examine the question at hand, and to this end, future larger clinical trials are warranted to re-examine the premise of intrastromal agents for management of fungal keratitis.

Amniotic membrane transplantation

Transplantation of human AM has been used in ocular surface reconstruction procedures in patients with corneal stem cell deficiency,59 as a substitute for conjunctival grafts in pterygium surgery,60 and for promotion of healing in acute chemical injuries.61

The AM's antibacterial,62 anti-inflammatory,63 and antifibroblastic64 properties are well recognised, and may play a role in the management of microbial keratitis. AM contains growth factors like epidermal growth factor, transforming growth factor (TGF) α, TGF β1, β2 and β3, human chorionic growth factor and β fibroblast growth factor help support the growth of progenitor cells.65 Cytokines and protease inhibitors (eg, interleukin (IL) 4, IL-6 and macroglobulin 2) are also present in AM. These factors inhibit inflammatory cells by their antiprotease activity.66 AM also contains collagen type IV, V and VII, in addition to laminin and fibronectin. The latter aids epithelial cell migration, adhesion and differentiation. Laminin helps in adhesion of epithelial cells to stroma, and type V collagen helps to anchor epithelial cells to stroma.67 Finally, AM provides mechanical support by protecting the regenerating epithelium from the frictional movement of eyelids.68

Monitoring of the effect of the AM on ocular surface healing usually takes the patient's symptoms into account, with reduced pain and improved comfort reported shortly after transplantation.69 With time, the ocular surface shows less inflammation, and the final impact of the AM on epithelial healing can be assessed once the membrane has dissolved.

There are several small retrospective70–72 and prospective68 ,73 ,74 case series reporting the use of human AM in the management of both bacterial and fungal keratitis; these, together with a single randomised controlled trial comparing AM with a conjunctival flap for management of microbial keratitis, are summarised in table 4.

Table 4

Published studies of AM transplantation for microbial keratitis

The majority of these studies lacked a control group, limiting any conclusions that can be made, either for or against AM transplantation. Kheirkhah et al69 conducted a prospective interventional case series with retrospective controls, comparing the role of AM at day 2–3 postantibiotic therapy with antibiotic therapy alone for Pseudomonas keratitis. The entire cornea and limbus were covered by a single layer of cryopreserved AM, epithelial-side up, sutured at 2 mm posterior to the limbus by a continuous 10–0 nylon suture (overlay technique). The healing time of the epithelial defect was determined using fluorescein staining after dissolution of the AM. Concurrent antibiotics were not used with the AM, and once dissolved the AM was not replaced. The authors demonstrated equal healing time and BSCVA in cases and controls, but better uncorrected visual acuity and less corneal scar density in the AM group. Less corneal haze and neovascularisation has also been demonstrated in a rat model of Staphylococcus aureus keratitis, which randomised eyes to antibiotic therapy with and without AM, and demonstrated better outcomes in the former group.76

Abdulhalim et al75 recently reported the results of a randomised clinical trial comparing the clinical outcomes of AM grafting with that of bipedicle conjunctival flap for infectious keratitis due to bacteria, fungi and Acanthamoeba. They found no significant difference in final visual acuity or time to re-epithelialisation between the two groups, although the study may have been underpowered to detect a significant difference.

In summary, there is some evidence to suggest that AM transplantation may be a useful adjunct to antimicrobial therapy in the management of recalcitrant microbial keratitis, although further studies are required to determine the optimum timing of transplantation, specifically whether AM transplantation is efficacious both in the acute setting, as well as after the sterilisation phase, for delayed epithelial healing. As an extension of this point, it would also be useful to determine whether bacterial, fungal and amoebic keratitis would benefit equally from this adjuvant procedure. In considering whether to use AM, it should be also noted that its application generally requires a surgical procedure and can leave residual opacity affecting vision. Further, as a biological product, it carries a small risk of infection, as well as potential batch variability.

Novel methods of ocular drug delivery

In order to overcome the issues of poor drug bioavailability and inadequate drug penetration into the cornea, drug molecules can be modified and/or packaged. Excellent recent reviews of this topic have been published.77 ,78 Here, we detail in general terms three novel drug delivery devise, namely iontophoresis, contact lenses and collagen shields.

In iontophoresis, a small electrical charge drives charged drug molecules across a biological membrane towards an opposite charge.79 The technique has been used transcorneally and transclerally to deliver therapeutic drugs to the cornea, aqueous and vitreous; for the cornea, an eye cup or hydrogel delivery vehicle is used. Clinical trials of iontophretic delivery of corticosteroids for corneal graft rejection indicates the procedure is an alternative to frequent dosing or intravenous pulsing of corticosteroids.80 Although there is some fear of electrical burns,81 the results are promising, and this technique is likely to develop further for management of recalcitrant bacterial keratitis.

Contact lenses impregnated with drugs provide a mechanism of bandaging the ocular surface and delivering drugs to the ocular surface; however, drugs are released rapidly and could have a toxic effect with unsustainable bioavailability.82 These systems have been shown to work well in animal models; however, tear film composition and blink rates differ from humans and it is yet to be seen how well they work in vivo.

Bovine or porcine collagen shields cross-linked with ultraviolet light during manufacture can also be loaded with drugs. Collagen shields have been used as anti-infective drug delivery devices in animal models. Compared with topical fortified drops and standard hydrogel contact lenses, collagen shields appear equivalent to slightly superior.83 Mendicute et al84 reported the adjunctive use of collagen shields soaked in 0.5% amphotericin B in three patients treated with topical therapy for Aspergillus keratitis; however, others have shown concern regarding oxygen permeability of the collagen shields when worn overnight,85 as well as the possible risk of ocular surface toxicity.86

Collagen shields, contact lenses and iontophoresis all show promise for improved delivery of drugs in microbial keratitis. The evidence for these novel methods of drug delivery is still weak, necessitating well-designed controlled clinical trials that investigate the safety and efficacy of these techniques in humans.

Miscellaneous techniques

In addition to the above, a number of other techniques have, anecdotally, been reported to be useful in the management of microbial keratitis. These include doxycycline, an oral antibiotic that also has inhibitory properties against matrix metalloproteinases and neutrophils, and is postulated to stabilise corneal melting in the setting of a number of corneal conditions including microbial keratitis. There is, however, sparse evidence in the literature supporting its impact in human studies.87

The use of cryotherapy in the management of AK has also been described, though not extensively studied. In vitro studies of AK have shown that cryotherapy kills trophozoites but not cysts, unless combined with medical treatment.88 ,89 It may rarely have a role for persistently culture-positive AK unresponsive to medical management.11

Finally, there are two case reports describing neodymium-doped yttrium aluminium garnet (Nd:YAG) photodisruption of bacterial biofilm in cases of crystalline keratopathy.90 ,91 These alone are insufficient evidence to support the routine use of Nd:YAG laser in this context, and topical antibiotics remain the gold standard of treatment for this condition.


Intensive topical antimicrobial treatment is, and will for the foreseeable future, remain the mainstay of treatment for infectious keratitis. With the exception of topical corticosteroids and AM transplantation for selected cases of bacterial keratitis, the literature currently does not provide firm support for other ancillary techniques in the management of microbial keratitis. However, some of the examined techniques, such as CXL, hold promise as adjunctive therapy, and warrant further investigation.



  • Contributors All authors have contributed to the inception of this paper, the writing, critical reviews of the drafts and final approval of the manuscript.

  • Funding DR is supported by a National Health and Medical Research Council (NHMRC) Early Careers fellowship (APP1073846). NC is supported by an NHMRC Early Careers CJ Martin Fellowship (APP1036728). SW is supported by an NHMRC Career Development Fellowship (APP1050524).

  • Competing interests None declared.

  • Provenance and peer review Not commissioned; externally peer reviewed.

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.