17 Flach Final 11/9/01 11:23 AM Page 205

CORNEAL MELTS ASSOCIATED WITH TOPICALLY APPLIED NONSTEROIDAL

ANTI-INFLAMMMATORY DRUGS*

BY Allan J. Flach, MD

ABSTRACT

Purpose: Topically applied nonsteroidal anti-inflammatory drugs (NSAIDs) are frequently used to prevent miosis during
cataract surgery, to treat ocular allergies, to prevent excessive postoperative inflammation following cataract surgery, and
to treat cystoid macular edema following cataract surgery. They have also been used to control pain and photophobia fol-
lowing radial keratotomy and excimer laser photorefractive keratectomy. During August of 1999, severe complications fol-
lowing topical NSAID use, including corneal melting, were reported by members of the American Society of Cataract and
Refractive Surgery (ASCRS) responding to a survey distributed in letters from ASCRS to its members. The purpose of
this report is to review 11 cases of corneal melting in patients treated with topical NSAIDs, with special attention to the
observed toxicity and its relationship to dose and duration of treatment, coexistent disease and therapies, and the indica-
tion for treatment. The goal of this study is to identify factors useful in minimizing the occurrence of corneal toxicity.

Methods: The medical records and/or histories of 11 patients with corneal melting associated with the use of topical
NSAIDs are reviewed, with special attention to the indication for treatment, the dose and duration of treatment, and
coexistent diseases and medical treatments. In addition, the relationship between NSAID treatment and surgery and
between NSAID treatment and onset and extent of corneal toxicity are described.

Results: Each of the 11 patients appeared to suffer severe corneal toxicity following the topical use of 0.5% diclofenac
ophthalmic solution. Generic diclofenac (Falcon) (Alcon Laboratories, Inc, Fort Worth, Texas) was associated with 7 and
Voltaren (Ciba Vision, Atlanta, Georgia) with 4 of these cases. Duration of treatment prior to corneal melting varied from
6 days to 17 months. Associated ocular and systemic diseases and their respective treatments complicate the analysis of
these cases. In addition, the indication for treatment with topical NSAIDs was frequently unclear.

Conclusions: The inconsistent and variable dose-toxicity relationships suggest that coexistent factors other than a sim-
ple drug toxicity are implicated, if not causative, in NSAID-associated corneal melting. These cases demonstrate the
importance of making a clinical diagnosis before treatment and of following the clinical course of patients carefully dur-
ing treatment.

Tr Am Ophth Soc 2001;99:205-212

INTRODUCTION were reported by members of the American Society of

Cataract and Refractive Surgery (ASCRS) responding to a
Topically applied nonsteroidal anti-inflammatory drugs survey distributed in letters from ASCRS to its mem-
(NSAIDs) are frequently used to prevent miosis during bers.3,4 This led to a recall of Falcon, a generic form of
cataract surgery, to treat ocular allergies, to prevent exces- diclofenac ophthalmic solution (Alcon Laboratories, Inc,
sive postoperative inflammation, and to treat cystoid mac- Fort Worth, Texas).5 Some have concluded that the avail-
ular edema following cataract surgery. These agents have ability of generic diclofenac was the sole reason that
also been used to control pain and photophobia following corneal toxicity was observed.6 However, the potential
radial keratotomy and excimer laser photorefractive kera- importance of completing a careful review of all of these
tectomy.1,2 During August of 1999, severe complications reported cases before concluding that an isolated drug
following topical NSAID use, including corneal melting, toxicity explains the appearance of these severe corneal
toxicities has been recently emphasized.7 The purpose of
*From the Department of Veteran Affairs, San Francisco, and the this report is not to substitute itself for a complete analy-
Department of Ophthalmology, University of California, San Francisco, sis of the cases of corneal melting, but only to provide an
School of Medicine. Supported by a Merit Review grant from the interim review of 11 cases of corneal melting in patients
Department of Veterans Affairs, San Francisco; a grant from That Man
May See, Inc; and a departmental core grant from the National treated with topical NSAIDs, with special attention to the
Institutes of Health, Department of Ophthalmology, University of observed corneal toxicity and its relationship to dose and
California, San Francisco, School of Medicine. duration of treatment, coexistent diseases and therapies,

Tr. Am. Ophth. Soc. Vol. 99, 2001 205

17 Flach Final 11/9/01 11:23 AM Page 206

Flach
and indication for treatment. The goal of this report is to 3 months following cataract surgery. She was treated with
help identify factors potentially useful in minimizing the Falcon for 10 days, and after a corneal infiltrate with 80%
occurrence of corneal toxicity while we await a more thor- tissue loss was observed, she eventually perforated.
ough examination of the factors associated with these tox- Culture revealed group B streptococcus.
icities.
CASE 2

METHODS A 66-year-old woman with a history of dry eyes was treat-

ed with Voltaren and apraclonidine hydrochloride
The medical records and histories of 11 patients with (Iopidine) (Alcon Laboratories, Inc) following cataract
corneal melting associated with the use of topical NSAIDs surgery. After 4 days of treatment, she complained of a
were reviewed, with special attention to the indication for foreign body sensation in the eye. The eye was red and
treatment, the dose and duration of treatment, and coex- photophobic, and she stated that the Voltaren burned
istent diseases and medical therapies. These 11 cases con- more upon instillation. She was told to refrigerate the
sist of 5 published cases,8 3 cases reported as a poster pres- Voltaren to reduce the burning sensation and to continue
entation,9 and 3 cases from the author’s referral practice. treatment. She presented 29 days after surgery with 50%
Seven cases mentioned at the 104th Annual Meeting of tissue loss. She had reduced values on the Schirmer test.
the American Academy of Ophthalmology (AAO) are also
CASE 3
included.6
A 77-year-old man was treated with Voltaren and
tobramycin-dexamethasone drops (TobraDex) (Alcon
RESULTS Laboratories, Inc) following cataract surgery. Although he
had normal examination results 1 week after surgery, he
Each of the 11 patients presented with severe corneal tox- presented with corneal perforation in the area of surgery
icity and a history of treatment with 0.5% diclofenac oph- 18 days after surgery. He had reduced values on Schirmer
thalmic solution. Generic diclofenac (Falcon) was associ- tests (12 mm and 8 mm) and diminished corneal sensation.
ated with 7 and Voltaren (Ciba Vision Corporation,
Atlanta, Georgia) was associated with 4 of these cases. A CASE 4
summary of the 11 cases is provided in Table I. A brief A 71-year-old diabetic man with systemic hypertension
description of each case follows: was treated with Falcon and 1% prednisolone given 6
times daily following cataract surgery. He experienced dis-
CASE 1 comfort and hyperemia on postoperative day 7, and he
A 76-year-old woman with a history of dry eye (Schirmer noted decreased vision on postoperative day 9.
test results, 2 mm and 5 mm) developed a red, painful eye Perforation occurred on postoperative day 11.

TABLE I. SUMMARY OF CASES

CASE(AGE/SEX) TREATMENT CORNEAL NSAID OTHER INDICATION FOR TREATMENT

DURATION PERFORATION (REGIMEN) MEDICATIONS (CULTURED ORGANISM)

1. 76 F 10 days Yes Falcon (QID) Tears Unknown (streptococci)

2. 66 F 29 days No (keratitis) Voltaren (QID) Glaucoma medications, Unknown (NP)
tears
3. 77 M 18 days Yes Voltaren (QID) Dexamethasone, Postsurgery
tobramycin, tears Inflammation (NP)
4. 71 M 11 days Yes (Falcon) (QID) Prednisolone Unknown (no growth)
5. 79 M 17 days Yes Falcon (QID) Glaucoma medications, Unknown (NP)
prednisolone
6 27 M 5 days (6 hr) Yes Falcon (QID) Rimexolone, Ciloxan Unknown (NP)
7. 47 F 4 days No (descemetocele) Falcon (QID) Corticosteroid Unknown (NP)
8. 80 M 10 mo No (descemetocele) Voltaren (QID) None Unknown (NP)
7 mo No (descemetocele) Falcon (QID) None Unknown (NP)
9. 65 F 5 mo No (keratitis) Voltaren (QID) Flarex, Alcaine Corneal abrasion? (NP)
10. 71 F 5 days No (keratitis) Voltaren (TID) Econopred, Ciloxan Dellen? (NP)
11. 77 F 14 days Yes Falcon (6/day) Polymixin B sulfate, Unknown (NP)
neomycin,
dexamethasone

NSAID, nonsteroidal anti-inflammatory drug; NP, not performed.

206
17 Flach Final 11/9/01 11:23 AM Page 207

Corneal Melts Associated With Topically Applied Nonsteroidal Anti-InfLammmatory Drugs

CASE 5 fluorometholone, and a bandage contact lens for 2 weeks.
A 79-year-old man underwent argon laser trabeculoplasty Voltaren was discontinued, after which she developed bul-
and was treated with 1% prednisolone with good results. lous keratopathy. Despite intermittent debridement, epi-
His eye became painful and developed an anterior cham- lation, and treatment with Voltaren and Flarex, her cornea
ber reaction 3 weeks following surgery. Falcon was added continued to exhibit a superficial punctate keratitis. She
to a regimen of brimonidine tartrate ophthalmic solution eventually underwent a penetrating keratoplasty with
(Alphagan) (Allergan, Inc, Irvine, California), dorzo- good results.
lamide (Trusopt) (Merck & Co, West Point,
Pennsylvania), timolol maleate (Timoptic) (Merck & Co), CASE 10
and latanoprost (Xalatan) (Pharmacia Corp, Kalamazoo, A 71-year-old woman was treated with Voltaren, pred-
Michigan). Increased pain, photophobia, and hyperemia nisolone acetate (Econopred) (Alcon Laboratories, Inc),
developed over 2 weeks, and he presented with 99% tis- and ciprofloxacin following cataract surgery. She present-
sue loss and a descemetocele 17 days after initiation of ed on postoperative day 5 with ocular pain, and a dellen
treatment with Falcon. was observed during examination. Voltaren was discontin-
ued, and goniosol hydroxypropyl methylcellulose (Ciba
CASE 6 Vision) was added to the Econopred treatment. Following
A 27-year-old man presented 5 days following excimer a poor response to patching, she underwent conjunctival
laser surgery complaining of pain. No corneal thinning grafting with good results.
was observed on examination. He was using rimexolone
(Vexol) (Alcon Laboratories, Inc), ciprofloxacin (Ciloxan) CASE 11
(Alcon Laboratories, Inc), and Falcon ophthalmic med- A 77-year-old woman with an eye that had been irritated
ications. Falcon was discontinued, but he continued to use for many months following a complicated cataract surgery
rimexolone and ciprofloxacin. He returned in 6 hours with presented with increased pain and redness in that eye and
a corneal perforation. an associated “injection of the upper tarsus” of unknown
origin. She was treated with Voltaren every 4 hours. She
CASE 7 returned in 2 weeks using Falcon and a steroid-antibiotic
A 47-year-old woman with a history of radial keratotomy eye drop. She eventually suffered corneal melt with cen-
20 years previously returned for excimer laser surgery. tral perforation.
She received postoperative treatment with Falcon, fluo-
rometholone acetate (Flarex) (Alcon Laboratories, Inc), OTHER CASES
and ciprofloxacin. She returned on postoperative day 4 In addition to these 11 cases, corneal melting in 7
complaining of pain. Her medical regimen was discontin- “healthy, asymptomatic eyes” following refractive or
ued, and cephazolin and tobramycin eye drops were pre- cataract surgery in patients treated with Falcon were
scribed. The cornea continued to melt, and a topical cor- reported at the 104th Annual Meeting of the AAO.6
ticosteroid was added to the regimen. She required a pen- Although detailed clinical descriptions were not provided
etrating keratoplasty 5 months later. for these cases, at least 2 of the patients were said to have
a history of punctal plug insertion, which suggested clini-
CASE 8 cally significant dry eyes. In addition, it was noted that all
An 80-year-old man developed cystoid macular edema 5 7 cases had occurred in one practice, while the nation’s
months after cataract surgery. He was treated with “top 15 presribers of Voltaren” have yet to report a severe
Voltaren for 10 months without toxic effects. Falcon was case of corneal toxicity with use of topical NSAIDs. The
substituted, and after 7 months he presented with pain. speaker concluded that use of Falcon was the cause of all
Corneal thinning was observed, and a descemetocele was of the cases of corneal melting observed.
noted after 48 hours.
DISCUSSION
CASE 9
A 65-year-old woman with trichiasis and a history of Corneal complications related to topical NSAID use are
cataract surgery 3 years previously underwent a YAG cap- uncommon. Superficial punctate keratitis, corneal infil-
sulotomy during which she suffered a corneal abrasion. trates, and epithelial defects have been reported following
After treatment with Voltaren without patching, she the use of these anti-inflammatory agents.10-13 These find-
developed a recurrent corneal erosion. She was treated ings are not surprising, because most topically applied
with intermittent patching, Voltaren, proparacaine medications, particularly those with preservatives, are
hydrochloride (Alcaine) (Alcon Laboratories, Inc), associated with potential corneal toxicity.14,15 However, the

207
17 Flach Final 11/9/01 11:23 AM Page 208

Flach
reports of corneal melting associated with topical NSAID corneal melting in this small series of 11 cases.
treatment are surprising and of great interest. Because Coexistent local and systemic medical treatments
both infectious and noninfectious corneal melting disor- complicate the analysis of these cases of corneal toxicity.
ders have many different causes, careful examination of For more than 2 decades, corticosteroids have been rec-
patients is important before a drug toxicity is identified as ognized as a cause of corneal toxicity. In fact, 25 cases of
the cause in all of these cases.16 corneal perforation reminiscent of these cases of corneal
Generic diclofenac (Falcon) may be the sole reason melting have been reported by a single observer.21
that corneal melting occurred in 2000. Unfortunately, this Therefore, the use of corticosteroids by 8 of 11 of these
conclusion is supported only by anecdotal presentations patients may be important. In further support of this pos-
that include a minimum of data with limited analysis and sibility, it is of note that in case 7, a descemetocele formed
little or no discussion or consideration of complicating fac- despite discontinuation of Falcon and during use of only a
tors and alternative explanations.6 The 7 cases of corneal corticosteroid. In addition, patient 5, who eventually per-
melts in Falcon-treated patients that were presented at forated, was using not only a corticosteroid but also mul-
the AAO annual meeting6 are difficult to discuss because tiple medications, some of which predispose to dry eye
the associated environmental factors and clinical descrip- (hydrochlorothiazide and timolol) and others that have
tions were not provided. Furthermore, it appears that significant potential for inducing corneal toxicity (dorzo-
potentially important coexistent ocular disease was largely lamide, timolol, brimonidine, and latanoprost).
ignored during the review of these cases. For example, 2 A touchstone for the determination of pharmacologic
patients requiring punctal plugs were included as toxic disease has been proposed.22 I advocate use of the
“healthy, asymptomatic patients,” ignoring the fact that following Koch-type postulates for a toxic etiology:
patients with Sjögren’s syndrome can develop sterile • The clinical signs of toxicity must be reproducible in
corneal ulcerations and perforations without any medical experimental animals.
treatment or surgical procedure.17 In addition, patients • The toxic dose-response may show normal scatter of
with mild and clinically insignificant keratitis sicca have random distribution, but no patient must get toxic
developed severe penetrating and perforating ulcers fol- effects from doses differing by several orders of mag-
lowing cataract surgery without any associated medical nitude.
treatments.18 Finally, it is impressive that there appears to • Cessation of dosage should be followed by a decrease
be an unbalanced geographic distribution of these cases of in toxicity.
corneal melting. An asymmetric distribution of an The corneal toxicity reported in these 11 cases does
observed drug toxicity can reflect production or manufac- not fulfill these criteria.
turing problems in a specific lot of drug.19 Therefore, it is Corneal melting has not been reproduced in experi-
of paramount importance to complete a careful review of mental animals with use of topically applied, commercially
all of the reported cases of corneal toxicity before con- available, brand-name NSAIDs. To the contrary, well-
cluding that an isolated drug toxicity explains the appear- designed laboratory studies suggest that these topically
ance of these severe corneal toxicities.7 administered NSAIDs may be beneficial in protecting ani-
This review of 11 cases of corneal toxicity observed in mals from corneal melting.20 In addition, carefully
patients using topically applied diclofenac does not pro-
vide compelling evidence of an isolated drug toxicity. The TABLE II: POTENTIAL CAUSES OF ACUTE CORNEAL MELTING
potential causes of acute corneal melting suggest that
many cases are unrelated to medical treatment, as sum- Herpes simplex keratitis
Mooren’s ulcer
marized in Table II.20 There is little evidence that these Rheumatoid arthritis
potential causes were carefully excluded from these 11 Bacterial keratitis
cases. A clinical diagnosis and therefore an indication for Keratoconjunctivitis sicca
anti-inflammatory treatment were lacking in 8 of 11 cases. Erythema multiforme
Alkali burn
It is particularly impressive how seldom an infectious Anterior-segment dysgenesis
cause was ruled out despite the presence of an uncom- Herpes zoster
fortable red eye of uncertain origin (9 of 11 patients). Neuroparalytic keratitis
Three patients (cases 1, 2, and 3) had dry eyes. A deficient Wound melt/keratoplasty
Pemphigoid
tear film has been associated with corneal melting11,17,18 In Rosacea keratitis
addition, abnormal tear production may contribute to Thermal burn
enhanced corneal toxicity from topical therapy, particular- Vernal keratoconjunctivitis
ly if preservatives are present. Therefore, coexistent dis-
Information from Kenyon.20
eases may have contributed to any or all of the observed

208
17 Flach Final 11/9/01 11:23 AM Page 209

Corneal Melts Associated With Topically Applied Nonsteroidal Anti-InfLammmatory Drugs

controlled, prospective, double-masked, randomized clini- therapeutic efforts of her surgeon and the resultant excel-
cal studies in hundreds of patients indicate that topically lent visual outcome, but instead gravitated toward a
applied brand-name NSAIDs appear safe for use in potential toxicologic explanation that was at the time pop-
patients.1,2 Therefore, it is unlikely that there is something ularized in the tabloids and debated on the Internet.
distinctive about the diclofenac molecule that predisposes to Overall, this review of 11 cases of corneal toxicity
severe corneal toxicity. It remains to be seen how carefully observed in patients using topically applied diclofenac
the generic NSAID Falcon has been tested for the potential reveals an inconsistent and variable dose-toxicity relation-
of inducing a corneal melt in experimental animals. ship. Furthermore, all of the cases are complicated by
An incredibly inconsistent dose-toxicity relationship coexistent diseases and medical therapies. This makes it
exists within the group of 11 cases described here. In case difficult to establish a definitive diagnosis for the observed
8, for example, the patient was treated with Voltaren for corneal melting. These cases underscore the importance
10 months and Falcon for 7 months for cystoid macular of making a clinical diagnosis before initiating nonspecific
edema following cataract surgery without apparent toxici- anti-inflammatory treatment and the need for careful fol-
ty. Then, for unexplained reasons, the patient developed low-up of patients after surgical procedures.
corneal thinning with descemetocele formation over 48 While we await a definitive analysis of all the report-
hours while using Falcon. This history of prolonged expo- ed cases of corneal melting associated with topical NSAID
sure without corneal toxicity followed by an acute corneal use, it seems prudent to keep in mind an admonishment
melt is not typical of an isolated drug toxicity. from Sir William Osler concerning the potential toxicity of
Patient 9 was referred to the author as a possible case medications, quoted by Dr Fred Wilson II in his
of NSAID-induced corneal melt. This case clearly demon- American Ophthalmological Society thesis:23
strates how resistant a cornea can be to the potential
corneal toxicity of Voltaren. The patient was treated for In the fight which we have to wage incessantly against igno-
various indications, including corneal abrasion, recurrent rance and quackery among the masses and follies of all sorts
among the classes, diagnosis, not drugging, is our chief
corneal erosion, and bullous keratopathy with Voltaren
weapon of offence. Lack of systematic personal training in
therapy. Despite concurrent treatment with Flarex and the methods of the recognition of disease leads to the mis-
Alcaine, with and without patching, bandage contact lens application of remedies, to long courses of treatment when
application, and intermittent debridement, no significant treatment is useless, and so directly to that lack of confi-
thinning of the cornea was noted during 5 months of treat- dence in our methods which is apt to place us in the eyes of
ment. This is remarkable. The coexistent corneal treat- the public on a level with empirics and quacks.
ments could have in and of themselves caused consider-
able corneal damage, but little was observed. CONCLUSIONS
Finally, in case 6, the patient had a normal corneal
examination 5 days after excimer surgery only to have per- The inconsistent and variable dose-toxicity relationships
foration 6 hours later while using Falcon. It is difficult to reflected in these 11 cases of corneal melting in patients
find any drug or chemical toxicity reported within the lit- using topical diclofenac suggest that coexistent factors
erature with such a rapid course. Therefore, a comparison other than a simple drug toxicity are implicated, if not
of cases 6, 8, and 9 makes clear an unusual inconsistency causative, in these toxicities. Concurrent ocular and sys-
of observed corneal toxicity with topically administered temic disease in many of these patients, as well as their
diclofenac. This is unlike an isolated drug toxicity. use of medications including corticosteroids, complicates
Case 10 demonstrates the importance of careful post- this analysis. The occurrence of corneal melting can be
operative examinations. This patient was referred to the minimized by attempting to make a definitive diagnosis
author because she was unhappy despite a good visual before initiating anti-inflammatory treatment and by care-
outcome. She believed that she had suffered an “eye melt- ful follow-up examinations of patients after surgery.
ing” due to Voltaren use. Five days following cataract sur-
gery, she had a painful eye and was treated with Voltaren REFERENCES
and Econopred. She was carefully examined, and a dellen
1. Flach AJ. Cyclo-oxygenase inhibitors in ophthalmology. Surv
was recognized and appropriately treated. Fortunately,
Ophthalmol 1992;36:259-284.
the patient was not simply treated aggressively for a resist- 2. Flach AJ. Duane’s Foundations of Clinical Ophthalmology. 2nd ed.
ant postoperative inflammatory response without benefit Philadelphia, Pa: JB Lippincott; 1994;1-32.
of an examination. It is likely that the careful follow-up 3. NSAID Adverse Reaction Report. (Press release): American
Society of Cataract and Refractive Surgery, Aug 3, 1999.
care that she received helped her to avoid severe corneal
4. NSAID Update. (Press release): American Society of Cataract and
damage related to an improperly treated dellen. It is iron- Refractive Surgery, American Society of Ophthalmic
ic that the patient ignored the extensive diagnostic and Administrators, Aug 11, 1999.

209
17 Flach Final 11/9/01 11:23 AM Page 210

Flach
5. Rosenthal KJ. ASCRS inquiry of NSAID problem leads to drug autoimmune disease and co-existent bacterial ulcers in
recall. Ophthalmol Times 1999;Nov 15. some of these patients. Both the brand (Voltaren)
6. Gayton JL. Primary management problems symposium:
Complications of NSAIDs. Presented at the 104th Annual Meeting diclofenac and the generic brand have been implicated,
of the American Academy of Ophthalmology, Oct 24, 2000. about 60% of the cases having been treated with the
7. Flach AJ. Topically applied nonsteroidal anti-inflammatory drugs generic brand.1
and corneal problems: An interim review and comment. Corneal melting or keratolysis in the absence of treat-
Ophthalmology 2000;107:1224-1226.
8. Lin JC, Rapuano CJ, Laibson PR, et al. Corneal melting associated ment with NSAIDs, occurs in the association with autoim-
with use of topical nonsteroidal anti-inflammatory drugs after ocu- mune disease, keratoconjunctivitis sicca, diabetes and
lar surgery. Arch Ophthalmol 2000;118:1129-1132. steroid use (which occurred in a number of these
9. Apte RS, Hargrave SL, June JC, et al. Matrix metalloproteinases patients). Recently analysis of tissue removed from a
and post-operative corneal melts. Scientific Poster 231. Presented
at the 104th Annual Meeting of the American Academy of patient suffering a corneal melt associated with NSAID
Ophthalmology, Oct 24, 2000. use suggests that this keratolysis was mediated by matrix
10. Gills JP. Voltaren associated with medication keratitis. (Letter) J metalloproteinases (MMPs). MMP is a family of degrada-
Cataract Refract Surg 1994;20:110. tive enzymes. They degrade collagen I, II, III, V, VII, in
11. Sher NA, Krueger RR, Teal R. Role of topical corticosteroids and
nonsteroidal anti-inflammatory drugs in the etiology of stromal addition to basement membrane, laminin, and proteogly-
infiltrates after excimer photorefractive keratectomy. J Refract cans. In a case report by O’Brien et al, MMP8 was iden-
Corneal Surg 1994;10:587-588. tified.2 The source of this is thought to be neutrophils and
12. Probst LEV, Machat JJ. Corneal subepithelial infiltrates following epithelium; this enzyme degrades collagens I, II and III.
photorefractive keratectomy. (Letter) J Cataract Refract Surg
1996;22:281. In addition, MMP2 and MMP9 have been identified after
13. Shimazaki J, Saito H, Yang HY, et al. Persistent epithelial defect refractive surgery and in patients with keratoconjunctivi-
following penetrating keratoplasty: An adverse effect of diclofenac tis sicca.3
eyedrops. Cornea 1995;14:623-627. It seems likely that possible triggers of MMPs
14. Wilson FM. Adverse external ocular effects of topical ophthalmic
medications. Surv Ophthalmol 1979;24:57-88. include: keratoconjunctivitis sicca, ocular surface disease,
15. Burstein NL. Corneal cytotoxicity of topically applied drugs, vehi- bacterial infections, NSAIDs, preservatives in topical
cles and preservatives. Surv Ophthalmol 1980;25:15-30. medications, and surgery. In a predisposed ocular sur-
16. Donzis PB, Mondino BJ. Management of noninfectious corneal face, the use of NSAIDs may substantially increase the
ulcers. Surv Ophthalmol 1987;32:94-110.
17. Pfister RR, Marphy GE. Corneal ulceration and perforation asso- risk of a clinically significant episode of keratolysis.
ciated with Sjögren’s syndrome. Arch Ophthalmol 1980;98:89-94.
18. Radtke N, Meyers S, Kaufman HE. Sterile corneal ulcers after REFERENCES
cataract surgery in keratoconjunctivitis sicca. Arch Ophthalmol
1978;96:51-52.
19. O’Day DM. Value of a centralized surveillance system during a 1. Lin JC, Rapuano CJ, Laibson PR, et al. Corneal melting associated
national epidemic of endophthalmitis. Ophthalmology with use of topical non-steroidal anti-inflammatory drugs after ocu-
1985;92:309-315. lar surgery. Arch Ophthalmol 2000;118(8):1129-1132.
20. Kenyon KR. Inflammatory mechanisms in corneal ulceration. 2. O’Brien TP, Li QJ, Sauerburger F, et al. The role of matrix metal-
Trans Am Ophthalmol Soc 1985;83:610-663. loproteinases in ulcerative keratolysis associated with perioperative
21. Thygeson P. Controversies in Ophthalmology. Philadelphia, Pa: diclofenac use. Ophthalmol 2001;108: 656-659
WB Saunders; 1977;450-469. 3. Sack RA, Sathe S, Bialon A. Gelatinases, origins, and associated
22. Potts AM. Tobacco amblyopia. Surv Ophthalmol 1973;17:313-339. proteins in normal and dry eye tear fluids. Invest Ophth Vis Sci
23. Wilson FM. Adverse external ocular effects of topical ophthalmic 2001;42(4)(Supple); S263
therapy: An epidemiologic, laboratory, and clinical study. Trans Am
Ophthalmol Soc 1983;81:854-965. [Editor’s note] DR OLIVER SHINE discussed the paper he
co-authored that analyzed 140 patients with corneal toxic-
ity associated with the use of NSAID’s, and the problem
DISCUSSION
of multiple events and underlying diseases which could
affect the cornea in these patients. DR TAYLOR ASBURY
DR MICHAEL A. LEMP. Dr Flach has called our attention, mentioned that Dr Philips Thygeson was one of the first
by way of a retrospective series of 11 patients, to the con- physicians to describe corneal melting associated with the
temporary vexing problem of keratolysis associated with use of topical medicines (corticosteroids).
the use of non-steroidal anti-inflammatory topical med-
ications. He points out the compounding variables in DR ALLAN J. FLACH. First I want to thank Dr Michael
these cases and suggests that co-existing factors rather Lemp for agreeing to discuss my paper and also Drs Tuck
than a simple drug toxicity are implicated. Asbury and Oliver Shine for their interest and comments.
Indeed, the conditions associated with these, and Dr Asbury reminds us that it was Dr Philips Thygeson
other previously recorded cases include cataract surgery, who first emphasized the dangers of corneal melts associ-
dry eye in a significant number of these patients, ated with corticosteroids and that he was elected to the

210
17 Flach Final 11/9/01 11:23 AM Page 211

Corneal Melts Associated With Topically Applied Nonsteroidal Anti-InfLammmatory Drugs

AOS in 1936. I would like to add that Dr Thygeson, while the geographical distribution and specific origin of the 34
not physically as active as he would like to be, is still intel- cases of severe corneal toxicity. As mentioned in my pres-
lectually active at 97 years of age. Although he must lift entation this afternoon, we know at least 7 of these cases
himself from his bed with block and tackle, he continues of severe corneal melting are reported from a single sur-
to have many interests including the history of American geon’s practice.5 In addition, less than 2% of the entire
Indians of the southwestern United States. In fact, I membership (over 5000 ophthalmologists) of the
recently sent him an article concerning Geronomo as American Society of Cataract and Refractive Surgery
reflected in the writings of Edgar Rice Burroughs. reported corneal problems with NSAIDs.2 This unequal
One of my greatest fears is to prepare to present a distribution of cases of corneal melts may reflect a differ-
paper and submit a copy of the manuscript, only to have a ence between individual batches or lots of a given drug or
similar paper appear in print just prior to my presentation. other important localized differences in technique or
Today my greatest fear has been fully realized. Not only practice. This deserves further study to help understand
1 but 3 papers have been recently published on corneal the origin and pathogenesis of the observed corneal melts.
melts associated with topical NSAIDs.1,2,3 Dr Lemp has It does not seem appropriate to analyze cases of
mentioned one of them and Dr Shine is a co-author on uncomplicated keratitis and corneal melts with equal
another.1,2 Fortunately, my paper is reasonably consistent attention and emphasis as was done in this study because
with all 3 of them in terms of its observations and conclu- keratitis is such a common finding during any eyedrop
sions. Dr Lemp has summarized the results of O’Brien et treatment as mentioned within my paper. Of greatest
al’s case study of 1 case in which the matrix metallopro- importance and interest are the 34 cases of severe corneal
teinases (MMPs) may have been implicated in the toxicity. Therefore, the report of this study would bene-
observed corneal melting reported in a patient using top- fit greatly from a more careful description and discussion
ical NSAIDs.1 This report compliments a poster presen- of each of the 34 severe cases of corneal toxicity with spe-
tation provided by Apte et al at the 104th Annual Meeting cial attention to the working diagnosis and indication for
of the American Academy of Ophthalmology on this same treatment with a topical NSAID in each case, to the pres-
subject as is mentioned within the text of my presenta- ence of coexistent ocular and systemic disease and coexis-
tion.4 While both paper and poster share the common tent medical treatments, in particular the use of corticos-
suggestion that the observed corneal melts are consistent teroids. This data is largely omitted from the publication
with enhanced MMP activity, both presentations agree of this study.2
that clear evidence for a definitive etiologic relationship The study has unexpected outcomes that must be
between NSAIDs and corneal melts is lacking and more explained or at least discussed. More specifically, the
study is indicated to clarify the role of MMPs and possi- study fails to find an association of corneal melts, or even
bility of NSAID associated corneal melts. keratitis, with dry eye. This is not consistent with past
Dr Shine has mentioned an analysis of 140 patients experience and our existing literature both of which clear-
demonstrating corneal toxicity associated with the use of ly identify dry eyes, even asymptomatic dry eyes, as pre-
topically applied nonsteroidal anti-inflammatory agents disposed to corneal melts with or without coexistent sur-
(NSAIDs) that he has recently co-authored.2 He has gery or medical therapy.6,7 It is also of concern that this
appropriately underscored the difficult issues complicat- study states within its discussion that postoperative sterile
ing the review of these cases including the presence of corneal ulcers are most often associated with Mooren’s
multiple compounding events, multiple chronic diseases ulcer or collagen vascular disease with no mention of the
and questions about the actual identity of the NSAID in potential association with dry eyes with this complication.
question. However, even within these limitations some The report concludes that there was no association of
distinct patterns found within the analysis speak to causal- increased toxicity with off-label use of NSAIDs. This con-
ity in association with the generic diclofenac, which is no clusion seems inconsistent with the study’s finding that the
longer on the market, and severe keratitis. Corneal toxic- more severe cases of corneal toxicity were more likely to
ity may have been occasionally associated with brand have other than cataract surgery and that nonsurgical
name diclofenac (Voltaren, Novartis) or ketorolac (Acular, cases tended to have much greater doses of NSAIDs than
Allergan), when used for long periods of time in patients surgical cases. Both these statements suggest off-label
with underlying pre-existing eye disease but it is not pos- use. These inconsistencies deserve discussion.
sible to know for certain these were actually the drugs dis- Although it is clear that the authors of this report of
pensed by the pharmacist. 140 patients devoted a great deal of time and effort to the
Since its publication I have had the opportunity to study the final publication provides less information and
review this study and, unfortunately, its design has sever- discussion of this important issue than it deserves.
al significant shortcomings.2 The analysis does not include The third paper recently published reports 16 cases

211
17 Flach Final 11/9/01 11:23 AM Page 212

Flach
of keratitis, ulceration and perforation associated with REFERENCES
topical NSAIDs.3 Some of these cases may be the same
cases included in previously discussed studies. However, 1. O’Brien TP, Li QJ, Sauerburger F, et al. The role of matrix metal-
all of the 16 cases had extenuating circumstances includ- loproteinases in ulcerative keratolysis associated with perioperative
ing dry eye, coexistent steroid treatment, rosacea and diclofenac use. Ophthalmology 2001;108:656-659.
2. Congdon NG, Schein OD, von Kulajta P, et al. Corneal complica-
often an unclear indication for the treatment with an anti- tions associated with topical ophthalmic use of nonsteroidal anti-
inflammatory agent. Therefore, the observations and inflammatory drugs. J Cataract Refract Surg 2001;27:622-631.
results from this study are consistent with the presenta- 3. Guidera AC, Luchs JI, Udell IJ. Keratitis, ulceration and perfora-
tion and conclusions that I have provided this afternoon. tion associated with topical nonsteroidal anti-inflammatory drugs.
Ophthalmology 2001;108:936-944.
Unfortunately, at present, no controlled study exists 4. Apte RS, Hargrave SL, June JC, et al. Scientific Poster 231: Matrix
that permits us to make any conclusions about the risks metalloproteinases and post-operative corneal melts 104th Annual
associated with topical NSAID use and severe corneal Meeting of the American Academy of Ophthalmology, 2000:
toxicity. Although it seems clear that there is a small but October 24.
5 Gayton JL. Primary management problems symposium:
definite incidence of corneal melting following cataract Complications of NSAIDs, 104th Annual Meeting of the American
surgery associated with dry eye and other predisposing Academy of Ophthalmology, 2000;October 24.
diseases, the question remains whether any specific drug 6. Pfister RR, Marphy GE. Corneal ulceration and perforation asso-
treatment or changes in our current surgical techniques or ciated with Sjogren’s Syndrome. Arch Ophthalmol 1980;98:89-94.
7 Radtke N. Meyers S, Kaufman HE. Sterile corneal ulcers after
regimens increase the likelihood of these corneal melts cataract surgery in keratoconjunctivitis sicca, Arch Ophthalmol
following contemporary anterior segment surgery. 1978;96:51-52.

212