Commentary: NSAID-Induced Corneal Melt: Clinical Importance, Pathogenesis and Risk Mitigation

Basil Rigas1*, Wei Huang2,3, Robert A. Honkanen2

1Departments of Preventive Medicine, Stony Brook University, Stony Brook, NY, 11794-8175; USA

2Departments of Ophthalmology, Stony Brook University, Stony Brook, NY, 11794-8175; USA

3Second Xiangya Hospital, Central South University, Changsha, Hunan, CN, China

Corneal melt, the loss of corneal epithelium accompanied by thinning or even perforation of the corneal stroma, is a potentially serious eye condition. The association of corneal melt with topical use of non-steroidal anti-inflammation drugs (NSAIDs) has received recent attention not only because of its clinical importance, but also because of its impact on the therapeutic use of NSAIDs in the eye. NSAIDs are used frequently to treat ocular inflammation and/or pain1.

In our publication2, we evaluated the literature on NSAID-induced corneal melt (NICM), assessed NICM as a clinical entity (its association with NSAIDs was initially doubted), examined its pathogenesis and recommended measures to mitigate its impact on ocular health.

The first description of NICM was reported in 1999 based on a survey of members of the American Society of Cataract and Refractory Surgery3. The finding was so dramatic that shortly thereafter a large pharmaceutical company, Alcon, suspended the distribution of its ocular diclofenac4. Multiple case reports and small series followed this seminal observation. By now, all but one of the NSAIDs used in the eye have been associated with corneal melt and its existence is no longer in doubt4-16.

Several questions regarding NICM remained, however, not fully addressed. Most prominent among them is the one concerning the true incidence of NICM. The highest reported incidence was 7.5%17, but this is probably an overestimate. Considering all reports, the true incidence of NICM remains unknown and given the size of studies that are required for such a determination it will likely be so in the foreseeable future. For clinical purposes, one should consider the incidence of NICM ‘fairly low’ but certainly a real possibility, especially in the right clinical setting. Equally uncertain is the effect of NSAID dose and duration of treatment on NICM. Reports in the literature indicate that it can occur as early as within 3 days and as late as 17 months after the initial administration of ocular NSAIDs. Finally, notwithstanding the fact that one of the ocular NSAIDs has not been associated with NICM, it is clinically prudent to consider all ocular NSAIDs as potential inducers of NICM.

What is clinically relevant is that administration of ocular NSAIDs often requires an additional ‘trigger’ to induce corneal melt. Conditions predisposing to NICM include: diabetes mellitus, systemic immune diseases and ocular surface diseases that compromise the cornea, as well as recent ocular surgery18. Notable among the immune systemic diseases are rheumatoid arthritis, Sjögren’s syndrome and rosacea. Keratoconus and dry eye disease are ocular surface diseases that predispose to NICM. In fact, dry eye disease is considered a contraindication to the administration to NSAIDs to the eye19. Finally, surgery of the anterior chamber of the eye can make the ocular surface susceptible to NICM.

A point that lingers in the literature is whether vitamin E, when part of the formulation of ocular NSAIDs, is responsible for the corneal melt. Based on some in vitro studies and likely over-interpretation of results, it was proposed that it was this vitamin that caused corneal melt and not the NSAID in those preparations. Numerous reports have confirmed that vitamin E does not cause corneal melt. A compelling argument was provided, for example, by reports of corneal melt associated with diclofenac and other NSAIDs that did not contain vitamin E, that establishing the NSAID as the offending agent20.

The pathogenesis of corneal melt has been surmised from the available literature, by necessity often not thorough or exhaustive. It appears that corneal melt begins with a corneal epithelial defect, which, if not corrected, is followed by a breakdown of the stroma (hydrolysis of collagen fibers), resulting in corneal thinning, descemetocele and, in some cases, corneal perforation. Infiltrating inflammatory cells may contribute to the early corneal destruction. The suppression of the corneal epithelium propagates the damage.

Of the various signaling molecules in this setting, eicosanoids and matrix metalloproteinases (MMPs) are the major players. Eicosanoids, such as prostaglandin E2 (PGE2), are cytoprotective to the cornea21. Since NSAIDs inhibit their biosynthesis, their absence accelerates corneal melting, contributing to the destruction of the epithelial layer. It was reported that NSAIDs delay corneal wound healing by reducing production of 12-HHT (12-hydroxyheptadecatrienoic acid), a COX product22. Studies with transgenic mice identified this pathway as the main target in diclofenac-induced delayed corneal wound healing. MMPs, which can degrade components of the extracellular matrix, participate in corneal matrix remodeling23. Enhanced expression of MMPs -1, -2, -8 and -9 in the cornea has been implicated in NICM17, 24. Their activity erodes the stroma, which, when prolonged, culminates in true perforation.

We proposed a two-stage mechanism of NICM. It begins with the epithelial stage, followed by the stromal stage. The initiating event is the corneal epithelial defect induced by exposure to the NSAIDs of an already compromised cornea. The rapid reduction of PGE2 levels prevents the repair of the mucosa. Infiltrating leukocytes compound the damage and activation of MMPs disrupts tight junctions of epithelial cells enhancing the damage. The net result is to advance the cornea to the stromal stage, which is dominated by activated MMPs; MMPs hydrolyze the collagen fibrils of the stroma below the denuded epithelium. Advancing lysis of collagen reaches Descemet’s membrane producing a demescetocele. Further action by the MMPs completes the process by generating a perforation.

A recent report by us indicated that, unlike conventional NSAIDs, the so-called modified NSAIDs (phosphosulindac being a prime example), are extremely unlikely to have corneal melt as a side effect25. There are two reasons for this conclusion. First, phosphosulindac, not a cyclooxygenase (COX) inhibitor, does not affect PGE2 levels in the cornea. Second, phosphosulindac suppresses the expression and activity of MMPs. This dual effect protects the cornea instead of damaging it. In stark contrast, conventional NSAIDs such as diclofenac and ketorolac, two NSAIDs frequently implicated in NICM, suppress PGE2 almost completely and have no effect on MMPs.

Many clinicians view NICM as Damocles’ sword when prescribing ocular NSAIDs, especially in vulnerable patients or for prolonged periods of time. Our assessment of the existing data helped us formulate the following suggestions to help mitigate the risk of NICM.

1. Awareness of this side effect and of its severity is the first step towards risk control.

2. Risk factors should inform the decision to prescribe ocular NSAIDs. Diabetes, systemic immune diseases, and ocular surface diseases that compromise the cornea are clear risk factors. Ophthalmic surgery ought to be considered a risk factor. Some ocular surface diseases, such as dry eye disease are considered relative, and for most experts absolute, contraindications to the use of ocular NSAIDs.

3. The frequency and duration of administration of ocular NSAIDs should be restricted to the minimum required; their open-ended administration should be avoided.

4. Patients receiving topical ophthalmic NSAIDs should be monitored closely, especially those with risk factors for corneal melt or after ocular surgery.

5. When corneal melt is diagnosed, NSAID eye drops should be discontinued and aggressive and timely treatment be instituted.

The clinical importance of NICM creates a veritable challenge for ocular pharmacology. This problem can be addressed by developing approaches to both rapidly and effectively treat NICM or agents that control pain and inflammation but devoid of this side effect. It is hoped that the therapeutic conundrum in the use of ocular NSAIDs will be resolved soon. Until then, sensible awareness of this entity should serve our patients well.

This work was supported in part by funds awarded by the Office of Scientific Affairs, Stony Brook University, and a research grant from Medicon Pharmaceuticals, Inc. The authors declare no conflict of interest except for BR who has equity positions in Medicon Pharmaceuticals, Inc. and Apis Therapeutics, LLC.

  1. Semeraro F, Russo A, Gambicorti E, et al. Efficacy and vitreous levels of topical NSAIDs. Expert Opin Drug Deliv. 2015; 12(11): 1767-82. Epub 2015/07/16. doi: 10.1517/17425247.2015.1068756. PubMed PMID: 26173446.
  2. Rigas B, Huang W, Honkanen R. NSAID-induced corneal melt: Clinical importance, pathogenesis, and risk mitigation. Surv Ophthalmol. 2019. Epub 2019/07/16. doi: 10.1016/j.survophthal.2019.07.001. PubMed PMID: 31306671.
  3. Congdon NG, Schein OD, von Kulajta P, et al. Corneal complications associated with topical ophthalmic use of nonsteroidal antiinflammatory drugs. J Cataract Refract Surg. 2001; 27(4): 622-31. Epub 2001/04/20. PubMed PMID: 11311634.
  4. Flach AJ. Corneal melts associated with topically applied nonsteroidal anti-inflammatory drugs. Trans Am Ophthalmol Soc. 2001; 99: 205-10; discussion 10-2. PubMed PMID: 11797308; PMCID: PMC1359011.
  5. Lin JC, Rapuano CJ, Laibson PR, et al. Corneal melting associated with use of topical nonsteroidal anti-inflammatory drugs after ocular surgery. Archives of ophthalmology (Chicago, Ill : 1960). 2000; 118(8): 1129-32. Epub 2000/08/02. PubMed PMID: 10922213.
  6. Asai T, Nakagami T, Mochizuki M, et al. Three cases of corneal melting after instillation of a new nonsteroidal anti-inflammatory drug. Cornea. 2006; 25(2): 224-7. Epub 2005/12/24. PubMed PMID: 16371788.
  7. Isawi H, Dhaliwal DK. Corneal melting and perforation in Stevens Johnson syndrome following topical bromfenac use. J Cataract Refract Surg. 2007; 33(9): 1644-6. Epub 2007/08/28. doi: 10.1016/j.jcrs.2007.04.041. PubMed PMID: 17720085.
  8. Bekendam PD, Narvaez J, Agarwal M. Case of corneal melting associated with the use of topical nepafenac. Cornea. 2007; 26(8): 1002-3. Epub 2007/08/28. doi: 10.1097/ICO.0b013e3180cfe626. PubMed PMID: 17721306.
  9. Wolf EJ, Kleiman LZ, Schrier A. Nepafenac-associated corneal melt. J Cataract Refract Surg. 2007; 33(11): 1974-5. doi: 10.1016/j.jcrs.2007.06.043. PubMed PMID: 17964407.
  10. Feiz V, Oberg TJ, Kurz CJ, et al. Nepafenac-associated bilateral corneal melt after photorefractive keratectomy. Cornea. 2009; 28(8): 948-50. doi: 10.1097/ICO.0b013e318199e9cc. PubMed PMID: 19654513.
  11. Mian SI, Gupta A, Pineda R, 2nd. Corneal ulceration and perforation with ketorolac tromethamine (Acular) use after PRK. Cornea. 2006; 25(2): 232-4. Epub 2005/12/24. PubMed PMID: 16371790.
  12. Mohamed-Noriega K, Butron-Valdez K, Vazquez-Galvan J, et al. Corneal Melting after Collagen Cross-Linking for Keratoconus in a Thin Cornea of a Diabetic Patient Treated with Topical Nepafenac: A Case Report with a Literature Review. Case Rep Ophthalmol. 2016; 7(1): 119-24. Epub 2016/06/14. doi: 10.1159/000444437. PubMed PMID: 27293413; PMCID: PMC4899655.
  13. Prasher P. Acute corneal melt associated with topical bromfenac use. Eye Contact Lens. 2012; 38(4): 260-2. Epub 2011/12/16. doi: 10.1097/ICL.0b013e318235c506. PubMed PMID: 22169877.
  14. Khalifa YM, Mifflin MD. Keratitis and corneal melt with ketorolac tromethamine after conductive keratoplasty. Cornea. 2011; 30(4): 477-8. Epub 2010/11/26. doi: 10.1097/ICO.0b013e3181ef6ec7. PubMed PMID: 21099410.
  15. Gokhale NS, Vemuganti GK. Diclofenac-induced acute corneal melt after collagen crosslinking for keratoconus. Cornea. 2010; 29(1): 117-9. Epub 2009/11/13. doi: 10.1097/ICO.0b013e3181a06c31. PubMed PMID: 19907307.
  16. Di Pascuale MA, Whitson JT, Mootha VV. Corneal melting after use of nepafenac in a patient with chronic cystoid macular edema after cataract surgery. Eye Contact Lens. 2008; 34(2): 129-30. Epub 2008/03/11. doi: 10.1097/ICL.0b013e31814510a8. PubMed PMID: 18327052.
  17. Hargrave SL, Jung JC, Fini ME, et al. Possible role of the vitamin E solubilizer in topical diclofenac on matrix metalloproteinase expression in corneal melting: an analysis of postoperative keratolysis. Ophthalmology. 2002; 109(2): 343-50. Epub 2002/02/05. PubMed PMID: 11825822.
  18. Vivino FB, Minerva P, Huang CH, et al. Corneal melt as the initial presentation of primary Sjogren's syndrome. J Rheumatol. 2001; 28(2): 379-82. Epub 2001/03/15. PubMed PMID: 11246683.
  19. Guidera AC, Luchs JI, Udell IJ. Keratitis, ulceration, and perforation associated with topical nonsteroidal anti-inflammatory drugs. Ophthalmology. 2001; 108(5): 936-44. PubMed PMID: 11320025.
  20. Johnson CC. Central corneal melting associated with reformulated generic diclofenac in a patient with inferior fornix foreshortening. Ophthalmol Eye Dis. 2011; 3: 21-4. Epub 2011/01/01. doi: 10.4137/OED.S7300. PubMed PMID: 23861620; PMCID: PMC3661476.
  21. Shim J, Park C, Lee HS, et al. Change in prostaglandin expression levels and synthesizing activities in dry eye disease. Ophthalmology. 2012; 119(11): 2211-9. doi: 10.1016/j.ophtha.2012.05.038. PubMed PMID: 22858125; PMCID: PMC3689228.
  22. Iwamoto S, Koga T, Ohba M, et al. Non-steroidal anti-inflammatory drug delays corneal wound healing by reducing production of 12-hydroxyheptadecatrienoic acid, a ligand for leukotriene B4 receptor 2. Sci Rep. 2017; 7(1): 13267. Epub 2017/10/19. doi: 10.1038/s41598-017-13122-8. PubMed PMID: 29038497; PMCID: PMC5643301.
  23. Peng WJ, Yan JW, Wan YN, et al. Matrix metalloproteinases: a review of their structure and role in systemic sclerosis. J Clin Immunol. 2012; 32(6): 1409-14. doi: 10.1007/s10875-012-9735-7. PubMed PMID: 22767184./li>
  24. O'Brien TP, Li QJ, Sauerburger F, et al. The role of matrix metalloproteinases in ulcerative keratolysis associated with perioperative diclofenac use. Ophthalmology. 2001; 108(4): 656-9. Epub 2001/04/12. PubMed PMID: 11297478.
  25. Honkanen RA, Huang L, Xie G, et al. Phosphosulindac is efficacious in an improved concanavalin A-based rabbit model of chronic dry eye disease. Transl Res. 2018; 198: 58-72. Epub 2018/04/28. doi: 10.1016/j.trsl.2018.04.002. PubMed PMID: 29702077.

Article Info

Article Notes

  • Published on: December 03, 2019


  • Pathogenesis

  • Mitigation


Dr. Basil Rigas
Stony Brook University, HSC T17-080, Stony Brook, NY 11794-8175, USA; Telephone No: 631-444-9538; Fax No: 631-444-9553