- Email: [email protected]
Microbial Keratitis Complicating Penetrating Keratoplasty
Microbial I(eratitis Complicating Penetrating Keratoplasty LYE P. FONG, MD,t·2 L. DAVID ORMEROD, MD,t·3 KENNETH R. KENYON, MD,t·3 C. STEPHEN FOSTER, MD1•2
Abstract: A retrospective review of 68 consecutive episodes of microbial keratitis complicating 66 penetrating keratoplasties (PKs) showed major risk associations: suture-related problems (50%), contact lens wear (26%), previous herpes simplex infection (15%), graft failure (15%), and persistent epithelial defects (15%). Topical steroid (85%) and antibiotic (59%) usage were common iatrogenic factors. Half the infections occurred more than 1 year after grafting. Bacterial infections involving gram-positive organisms (59%) predominated, except for patients with extended-wear hydrophilic contact lenses, which usually involved gram-negative bacilli. The incidence of fungal infections (6%) was relatively low. Recommendations to minimize microbial keratitis include prompt attention to exposed, broken, or loose sutures, and preventive and therapeutic management of epithelial defects. The inadequacy of low-dose antibiotics in precluding microbial infection in many cases and the propensity to develop infections with resistant organisms suggest that guidelines for using postoperative and prophylactic topical antibiotics require reevaluation. [Key words: contact lens complications, endophthalmitis, keratoplasty, microbial keratitis, suture complications.] Ophthalmology 95: 1269-1275, 1988
Microbial keratitis after penetrating keratoplasty (PK) is an uncommon but serious complication. Although reported incidence rates are low, ranging from 1.8 1 to 4.9%,2 the visual sequelae of graft infection are frequently severe. Tuberville and Wood2 recorded a 46% rate of visual loss and only 14% achieved visual acuity greater than 6/60 in the study by Driebe and Stern. 3 This large retrospective study of microbial keratitis complicating PK was conducted to identify the associated risk factors, describe the microbiologic spectrum, and disOriginally received: February 16, 1988. Revision accepted: April 27, 1988. Cornea Service, Department of Ophthalmology, Massachusetts Eye and Ear Infirmary, Harvard Medical School, Boston. 2 Immunology and Uveitis Unit, Department of Ophthalmology, Massachusetts Eye and Ear Infirmary, Harvard Medical School, Boston. 3 Eye Research Institute, Boston. 1
Supported in part by a Hugh Noel Puckle Scholarship and grants EY06008 and EY05799 from the National Institutes of Health. Reprint requests to The Library, Eye Research Institute, 20 Staniford Street, Boston, MA 02114.
cuss the findings with respect to prevention, treatment, and complication management in corneal graft infections.
MATERIALS AND METHODS The medical records were reviewed from all corneal graft patients treated for microbial keratitis at the Massachusetts Eye and Ear Infirmary during an 8-year period (1978-1985). Cases caused by corneal exposure were excluded. Criterion for inclusion in this study was a discharge diagnosis of microbial keratitis in a keratoplasty patient that included either a positive corneal culture or a positive corneal smear Gram stain. Corneal scrapings, performed on all patients, were Gram stained and inoculated onto blood, chocolate and modified Sabouraud agar, and cooked meat broth using standard techniques. 4 ,5 Sabouraud agar was incubated at 25°C and the remainder at 35°C. Standard disc diffusion criteria for antibiotic sensitivity were used. 6 With the exception of fungi, cultures were considered positive 1269
OPHTHALMOLOGY
•
SEPTEMBER 1988 •
Table 1. Indications for Penetrating Keratoplasty in 66 Patients with Post keratoplasty Microbial Keratitis Indications Aphakic bullous keratopathy· Corneal scarringt Fuchs' endothelial dystrophyt Herpes simplex keratitis Keratoconus Trauma Corneal perforation (Sjogren's syndrome) Lattice corneal dystrophy Uveitic bullous keratopathy
No. of Patients
Prevalence (%)
19 13
29 20
11 10 5
17 15 8
2
3 2
4
1 1
6
2
* Including one pseudophakic bullous keratopathy. t Three patients with previous microbial keratitis, single cases with cicatricial pemphigoid, aniridia, sclerocornea, congenital leukoma, chlamydial infection, syphilitic interstitial keratitis, alkali burn, Stevens-Johnson syndrome, Goldenhar's syndrome, and a scarred lamellar keratoplasty (Salzmann's degeneration). t Not including aphakic cases.
when organisms were isolated from two or more media or from one culture medium if supported by an appropriate corneal smear Gram stain. A chart review was made using a standardized protocol with attention directed to the following: indications for PK; interval between PK and onset of microbial keratitis; microbiologic spectrum; complications; and detailed analysis of specific risk associations. Ulcer size was defined as small ( <2 mm greatest dimension), moderate (2-6 mm), and large (>6 mm). The location of the ulcers in donor or host cornea, or at the woundjunction, was recorded with details of the specific site.
RESULTS GENERAL EPIDEMIOLOGY
A total of 68 consecutive episodes of infection occurring in 66 patients after PK was reviewed. There were almost equal numbers of male and female patients. Right eyes were involved in 30 cases and left eyes in 38. The mean age was 61 years (range, 1-87 years). Thirtyseven patients (56%) were seen within 2 days of symptoms; nine individuals (13%) delayed attendance beyond 1 week. All patients were admitted and had a mean hospital stay of 10 days (range, 4-28 days). The indications for the previous corneal graft are shown in Table 1; the most common diagnoses were aphakic bullous keratopathy, miscellaneous corneal scarring, Fuchs' endothelial dystrophy, and herpes simplex keratitis. RISK FACfORS
Potential risk factors have been categorized under local ocular, topical medication, and systemic causes 1270
VOLUME 95
•
NUMBER 9
(Table 2) and identified under three major risk groups: suture (50%) and contact lens-related (26%) complications, and miscellaneous (24%). Forty-two (62%) of the infections occurred in aphakic patients. Two thirds (62%) of the suture-related infections involved abscesses around apparently secure sutures, although infections induced by loose or broken sutures and after suture removal also were seen (Table 3). Infections associated with sutures were more common within the palpebral fissure (Fig 1). Ten (7 with previous herpes simplex keratitis) of 18 episodes (56%) of contact lensrelated infections occurred in patients using therapeutic bandage lenses for persistent epithelial defects. Seven aphakic patients were using extended-wear and one patient a daily wear hydrophilic contact lens. Altogether, 14 of the 18 infections occurring in the contact lenswearing group were in aphakic patients. Of 16 patients in the miscellaneous group, 7 had recent graft rejections that had been treated with intensive topical corticosteroids, one Sjogren's syndrome, and one a wound dehiscence; 14 (88%) were receiving topical corticosteroid therapy. Except possibly for diabetes mellitus (12%) and chemotherapeutic immunosuppression (9%), systemic factors appeared insignificant. Two patients had recurrent ulcers. A Mycobacterium chelonei infection developed in one herpetic patient with a persistent epithelial defect, and a viridans streptococcal keratitis associated with bandage contact lens wear subsequently developed. The second patient had two infections involving different coagulase-negative staphylococcal biotypes complicating loose sutures. Two infections were in grafts performed for the sequelae of microbial keratitis occurring before the study period. The chronology of microbial keratitis after corneal transplantation is summarized for the three etiologic groups in Figure 2. Twenty-eight (41 %) infections occurred within 6 months of keratoplasty and 34 (50%) occurred within 1 year. The longest interval between surgery and infection was 13 years. Suture complications were the main causes of graft infections in the first year after keratoplasty. Eighty-five percent of 34 suturerelated infections occurred during this period, 38% within 2 months and 62% within 6 months of grafting. Problems related to contact lens wear, graft rejection, and corneal edema were the main late causes of infection; persistent epithelial defects were rarely associated at that time. The size and location of the microbial keratitis are shown in Table 4. Although suture-related infections occurred almost exclusively in the graft periphery, infections associated with bandage and extended-wear hydrophilic contact lens generally were localized to the center of the graft. Sutures were present in 10 of the 16 contact lens-infection grafts in which the suture status was known, but the sutures were not believed to be the source of infection in any of these cases. Infections generally remained confined by the grafthost interface; only 5 of 37 peripheral infections had relatively minor extensions of suppuration into the host corneas. Five infections, occurring at 7, 12, and 14
FONG et al
•
CORNEAL GRAFT INFECTIONS
Table 2. Predisposing Factors for Microbial Keratitis in 68 Corneal Graft Infections No. of Infections
Predisposing Factors
Suture Related (n = 34)
Contact Lens Related (n = 18)
Miscellaneous (n = 16)
Total (n = 68)
Prevalence (%)
0 0 7 6 0 2 1 1 1
34 18 11 10 10 4 3 3 1
50 26 16 15 15 6 4 4 1
Local ocular Suture-related problems Contact lens Recent graft rejection episode Corneal edema (graft failure) Persistent epithelial defect Trichiasis Mucosal scarring disorders* Wound dehiscence Sjogren's syndrome
34 0 3 2 (1)+ 1 1 2 0
(2)+ 18 1 2 10 1 2 (1)+ 0 0
Topical medication Corticosteroid Antibiotic Glaucoma medication(s) Antiviral
30 24 4 1
13 10 5 3
14 6 4 2
57 40 13 6
84 59 19 9
6 1 1 1
1 2 1 0
1 3 0 0
8 6 2 1
12 9 3 2
Systemic factors Diabetes mellitus Systemic immunosuppressiont Atopy Chronic alcoholism
* One cicatricial pemphigoid, one Stevens-Johnson syndrome, and one alkali burn patient. t Five oral corticosteroids ~ 10 mg daily (graft rejection, 2; herpes simplex, 2; asthma 1) and one azathioprine. + Associated factors.
Table 3. Analysis of Suture-related Factors in 34 Corneal Graft Infections Suture-associated Factors*
No. of Cases
Prevalence (%)
Suture abscess around secure suture(s) Loose suture(s)t Ruptured suture Recent suture removal
21 8 3 2
62 24 9 6
SUPERIOR
TEMPORAL
NASAL
* Usually 10-0 or 11-0 monofilament nylon. Comprises 11 interrupted and 11 continuous sutures; the suture type in 12 cases was unspecified. t May be underestimated in this retrospective series.
months and 3 and 9 years after transplantation, appeared to have arisen in the scar with approximately equal involvement of donor and host tissue. In three otherwise unremarkable cases, microbial keratitis was restricted to the host cornea abutting the wound margin.
INFERIOR
MICROBIAL PATHOGENESIS
Fig 1. Corneal segment involvement of suture-related infections complicating 32 corneal grafts. Notice that there was also one case of multifocal infection and one of extensive infection.
Of 68 infections, 66 (97%) were culture-positive; two had positive Gram stains confirming bacterial infection, although no organisms were isolated. Sixty-four (94%) cases were bacterial and four (6%) fungal in origin (Table 5). Gram-positive organisms were involved in
59% of the (culture-positive) infections, predominantly Staphylococcus aureus (21 %), coagulase-negative staphylococci (18%), Streptococcus pneumoniae (7%), and {j1271
OPHTHALMOLOGY
•
SEPTEMBER 1988
MISCELLANEOUS 10 5 til
;::
~
...><
0
It
CONTACT LENS· RELATED
....
'iii"
0
10
It
U
~
...
5
til
0
0
W
Q
0
SUTURE · RELATED
til
ii: w
"-
0
30
It W
'"~z
25 20
•
VOLUME 95
•
NUMBER 9
infections, and 50% of the miscellaneous group; the equivalent figures (including polymicrobial infections) for gram-negative bacillary infection were 22, 56, and 50%, respectively. Six often (60%) bandage contact lens infections were associated with gram-positive bacterial infection, and six of eight (75%) aphakic hydrophilic contact lens infections had gram-negative isolates. Polymicrobial infection was uncommon. All four fungal infections were suture related. There were no episodes of infectious crystalline keratopathy over this time period. In this study, the sensitivity of the Gram stain was 75% and the specificity of a positive Gram stain was 88%. Twenty-five (63%) of the 40 cases receiving topical antibiotics before the onset of infection had bacterial isolates that were resistant in vitro to the antibiotics. Polysporin (polymyxin (3 and bacitracin; Burroughs Wellcome, Research Triangle Park, NC), erythromycin, gentamicin, or chloramphenicol was commonly used for this purpose.
15 10
COMPLICATIONS 10.2 2·4
4· 6 6·8 8·10 10.1?
FIRST I
SECONO
MONTHS
2·4
4·6
6·8
8·10 I
YEARS
Fig 2. Time interval between penetrating keratoplasty and microbial infection.
hemolytic streptococci (6%). Of the gram-negative infections (38% of 68), Serratia marcescens (12%), Pseudomonas aeruginosa (10%), and Klebsiella sp (4%) were the most common isolates. Gram-positive cocci were isolated from 63% of the suture-related infections, 50% ofthe contact lens-related
The prevalence of major complications of microbial keratitis is presented in Table 6. Although limited by relatively small numbers, the rates for corneal perforation, enucleation/evisceration, wound dehiscence, failure of previously clear graft, and the requirement for emergency regraft suggest that the suture-related infections tended to have a worse prognosis than the other groups. Overall, one third of the patients required replacement of the corneal graft. Of previously clear grafts, 17% became edematous, either as a direct sequela of the infection or as a result of unrecognized, superimposed graft rejection. Endophthalmitis occurred in four
Table 4. Size and Location of Microbial Keratitis Complicating Corneal Grafts No. of Patients (%) Suture Related (n = 34)
Contact Lens Related (n = 18)
Miscellaneous (n = 16)
Size (mm) Small «2 in greatest dimension) Moderate (2-6) Large (>6)
14 (41) 18 (53) 2 (6)
7 (39) 7 (39) 4 (22)
6 (38) 9 (56) 1 (6)
27 (40) 34 (50) 7 (10)
Location Limited to donor cornea Central/paracentralPeripheral Overwhelming infection Equal donor and host corneal involvement Limited to host cornea
1 27t 1 3 2
4 8 1 2 1
20 (29) 37t (54) 3 (4) 5 (7) 3 (4)
- Not contiguous with wound edge. Remote from sutures in all but four. t All but four suture-related cases extending to wound edge. tlncluding five cases with minor extension across the wound into the host cornea.
1272
15 2 1 0 0
(n
Total = 68)
FONG et al
•
CORNEAL GRAFT INFECTIONS
Table 5. Microbial Etiology of 66 Episodes of Culture-positive Keratitis Complicating Corneal Grafts No. of Infections*
Organisms Gram-positive cocci Staphylococcus aureus Coagulase-negative staphylococci Streptococcus pneumoniae {j-hemolytic streptococci Viridans streptococci Gram-negative bacilli Serratia marcescens Pseudomonas aeruginosa Klebsiella sp Enterobacter aerogenes Mycobacterium chelonei Proteus mirabilis Morganella morgagnii Haemophilus inDuenzae Fungi Candida albicans Aspergillus sp Penicillium sp
Suture Related (n = 32)
Contact Lens Related (n = 18)
Miscellaneous (n = 16)
(n
Total = 66)
Prevalence (%)
10 4 4 2 1
1 6 (1) 0 1 1
3
2 1 1 1
14 12 (1) 5 4t 3
21 18 7 6 4
4 2 0 0 0 0 1 0
2 3 2 2 (1) 1 0 0 0
2 2 1 0 1 1 0 1
8 7 3t 2 (1) 2 1 1 1
12 10 4
3
0 0 0
0 0 0
1 (1) 1 (1)
1 (1) 1 (1)
3
3 1 1 1
3
4 1 1
* Polymicrobial isolates are in parentheses. tOne Lancefield group B, one group D, one group G, and one unclassified. t Two Klebsiella oxytoca and one Klebsiella pneumoniae. Table 6. Major Complications of Corneal Graft Infections No. of Patients (%)
Hypopyon Descemetocele Corneal perforation Glue and bandage contact lens Endophthalmitis Enucleation/evisceration Wound dehiscence* Failure of previously clear graftt Loss of light perception Emergency repeat PK Elective repeat PK Total repeat PK
Suture Related (n = 34)
Contact Lens Related (n = 18)
Miscellaneous (n = 16)
10 (29) 1 (3) 5 (15) 3 (9) 2 (6) 4 (12) 9 (26) 7 (21) 4 (12) 9 (26) 3 (9) 12 (35)
2 (11) 1 (6) 2 (11) 1 (6) 1 (6) 1 (6) 1 (6) 3 (17) 2 (11) 2 (11) 3 (17) 5 (28)
4 (25) 2 (13) 1 (6) 1 (6) 1 (6) 1 (6) 6 (38) 1 (6) 1 (6) 2 (13) 3 (19) 5 (31)
(n
Total = 68)
16 (24) 4 (6) 8 (12) 5 (7) 4 (6) 6 (9) 16 (24) 11 (16) 7 (10) 13 (19) 9 (13) 22 (32)
PK = penetrating keratoplasty. * Includes corneal perforations (but not central perforations) occurring at the PK wound edge. t Irreversible corneal edema.
patients (6%) and was associated with moderate-to-Iarge ulcers; two cultured ,a-hemolytic streptococci, one P. aeruginosa, and one Klebsiella oxytoca. Endophthalmitis complicated, respectively, a suture-related infection 6 weeks after a graft performed for Fuchs' dystrophy, a suture-related infection at 7 months in a graft under-
taken for corneal scarring, intensive topical steroid therapy for graft rejection 16 months after PK (aphakic bullous keratopathy), and bandage contact lens wear in a failed graft 4 years after surgery. Intraocular infection was confirmed at vitrectomy in three of these patients and by anterior chamber tap in the fourth. 1273
OPHTHALMOLOGY
•
SEPTEMBER 1988 •
DISCUSSION This study emphasizes the importance of exposed, loose, and broken sutures in the generation of postkeratoplasty microbial keratitis. Of our graft infections, 50% were suture related, compared with 46% in the report by Tuberville and Wood2 and 14% by Driebe and Stern. 3 Exposed sutures breach the epithelial surface and provide a direct pathway into the corneal stroma. Furthermore, abnormal mucus accumulation and debris may act as a nidus for microbial colonization. Operative factors, such as the failure to bury surgical knots and suture ends, improper suture tension or tying, and poor hostdonor alignment at the epithelial surface also may be important. Corneal deturgescence, wound remodeling,7.B and scar contracture occur during wound healing, and resultant forces may loosen or expose sutures. Suture biodegradation and rupture also can occur postoperatively. Because 19% of the cases ( 13 infections) occurred within 2 months postoperatively, early postoperative changes in wound configuration are considered major etiologic factors. Even though the superficial arc of the suture becomes rapidly buried, it is also possible that the overlying epithelium might sometimes remain unstable, with a propensity for exposure of the suture material. One exrlanation for the nasal and temporal predisposition of suture-related infections is the greater relative exposure in the interpalpebral fissure; another explanation might be the increased technical difficulty of suturing at these sites. A similar distribution has been noted for keratoplasty wound separations. 9 The much lower infection rates of superior and inferior sutures suggest that lid coverage may be a protective factor. Suture removal was the precipitating factor in two of our patients; this problem may increase because of the greater frequency of suture removal proposed for astigmatic control. 10 Although wound separation (and possible microbial infection) may occur with suture removal, particularly in the early postoperative period,9 suture manipulation alone may introduce bacteria from the ocular surface and induce keratitis or endophthalmitis. I1-13 We recommend a short course of topical bactericidal antibiotic, such as gentamicin, after suture removal. Most contact lens-associated microbial keratitis was related to the use of bandage lenses in treating persistent epithelial defects. These infections were characterized by a predominance of gram-positive bacteria (59%), notably S. aureus and coagulase-negative staphylococci. The gram-positive bacterial pattern in bandage lens wear was noted previously,14,15 and probably represents endogenous infection with ocular flora. In contrast, the microbial keratitis complicating aphakic extended-wear hydrophilic contact lenses was caused mainly by gram-negative organisms (75%), which compares with the 78% incidence reported by Alfonso et al 16 in their contact lens-associated microbial keratitis series and 75% in a recent study restricted to infections complicating the use 1274
VOLUME 95
•
NUMBER 9
of extended-wear cosmetic contact lenses. 17 Factors such as contact lens handling and contaminated contact lens cases/solutions probably are involved in explaining this different disease spectrum. No rigid contact lens wearers acquired infection in this study. It was notable that the graft infections associated with contact lens use largely occurred in the graft center, perhaps related to central corneal hypoxia under the contact lens. Despite the frequent concurrence of sutures, the cause of infection in these cases did not appear to involve an increased risk of suture exposure with contact lens wear. The distinct sites of involvement in suture-related and contact lens-related graft infections imply differing pathogenetic mechanisms. Although the use of aphakic contact lens correction may be diminishing after PK because of combined intraocular lens insertion, the complications of bandage contact lens wear will continue to be problematic. The risk of a persistent epithelial defect developing can be minimized by caring for the corneal epithelium during surgery, using preservative-free solutions to avoid medication toxicity, and aggressive lubrication. 14 Conservative occlusive therapy, consideration of early tarsorrhaphy, and the prophylactic correction of preexisting factors such as exposure, sicca, trichiasis, and external ocular inflammation are all important. 14,IB Clinical and experimental trials to assess the efficacy of topical fibronectin and epidermal growth factor to enhance epithelial healing l9-21 and botulinum toxin A-induced protective ptosis22 promise further therapeutic improvements. Fungal infections were uncommon (6%) in our series, reflecting the low incidence of fungal infection in the northeastern United States; a high incidence of fungal graft infection is recorded in some,I,21 but not in all,3 studies from the southern United States. All fungal infections were suture related: an important clinical point is that fungal infection may recede from the surface into the suture tract. Long-term prophylactic antibiotics were used in 59% of the corneal transplantation patients with microbial keratitis, but did not prevent infection. An important finding was that almost two thirds of the graft infections complicating the use of prophylactic antibiotics involved bacterial strains resistant to those antibiotics. Similar findings have been reported in other PK keratitis studies. 3,14,23 The possibility that the long-term use of topical antibiotics may alter the normal flora, select resistant strains of bacteria, or locally induce antimicrobial resistance requires urgent attention. Studies should be conducted to delineate the risk-benefit ratio of longterm topical antibiotic and/or corticosteroid regimens in ocular surface disease. Topical corticosteroid usage was a common factor (84%) in the graft infections, although the concurrence of multiple factors makes it difficult to interpret its specific role. However, corticosteroids impair host-defense and healing mechanisms and promote microbial superinfection24 ; they also may reduce the symptoms and signs of keratitis (as may poor graft reinnervation 25 ), encouraging more extensive disease to develop. There is
FONG et al
•
CORNEAL GRAFT INFECTIONS
no general consensus on the duration and intensity of postoperative steroids after corneal transplantation. 26 The contamination of topical medications also can be important. 27 At the time of the study (January 1978-December 1985), 2006 penetrating keratoplasties were performed at the Massachusetts Eye and Ear Infirmary. Sixty-six corneal grafts (recurrent infections in 2 grafts) were complicated by microbial keratitis over this period. An approximate incidence of 3.3% can therefore be calculated, which, because of delayed infection and the progressively increasing frequency of PK surgery at this institution, should probably be considered as a minimal estimate. The figure is comparable with previous reports,I,2,23 if somewhat less than another. 14 Although the incidence is low, the toll of ocular morbidity is disproportionately high. 2,3 Steroid-induced local immunosuppression and delayed wound healing,IO,28 compounded by the high incidence of suture-related infections, may lead to a loss of structural integrity and wound dehiscence (24%), perforation (12%), and the high rates for emergency regrafting (19%) and enucleation/evisceration (9%). The data suggest a worse prognosis for suture-related than contact lens-related graft infections, possibly related to their proximity to the graft edge and also to their occurrence earlier in the wound healing process. Endophthalmitis developed in a very large proportion of perforated PK infections as noted previously.29 This study has shown a major association of postkeratoplasty microbial keratitis to suture-related problems, which emphasizes the need for meticulous surgery, prompt patient recognition of new symptoms, and the urgent attention to exposed, loose, and broken sutures. The identification of patients at special risk, with persistent epithelial defects, contact lens wear, failed grafts, and graft rejection episodes also can improve preventive management. Only increased vigilance and patient education can prevent the disaster of microbial keratitis complicating corneal transplantation. Because of the high risk of visual loss, these patients should be admitted to the hospital, with the exception of patients with a small infection who are highly compliant; these individuals can be treated with subconjunctival or intensive topical antibiotics, or both, and seen every day. The inadequacy oflong-term antibiotics in preventing infections and the risk of inducing an antibiotic-resistant conjunctival flora necessitate reappraisal of the optimal use of postoperative and prophylactic antibiotics.
REFERENCES 1. Lamensdorf M, Wilson LA, Waring GO, et al. Microbial keratitis after penetrating keratoplasty. Ophthalmology 1982; 89(Sept. Suppl): 124. 2. Tuberville AW, Wood TO. Comeal ulcers in comeal transplants. Curr Eye Res 1981; 1:479-85. 3. Driebe WT Jr, Stem GA. Microbial keratitis following comeal transplantation. Comea 1983; 2:41-5. 4. Jones DB, Liesegang TJ, Robinson NM. Laboratory diagnosis of
ocular infections, Cumitech 13. Washington, DC: American Society for Microbiology, 1981. 5. Lennette EH, Balows A, Hausler WJ Jr, Truant JP. Manual of Clinical Microbiology. 3rd ed. Washington, DC: American Society for Microbiology, 1980. 6. National Committee for Clinical Laboratory Standards. Performance Standards for Antimicrobial Disc Susceptibility Tests (Approved Standard, M2A2). 2nd ed. Villanova, PA: The Committee, 1979. 7. Davison PF, Galbavy EJ. Connective tissue remodeling in comeal and scleral wounds. Invest Ophthalmol Vis Sci 1986; 27:1478-84. 8. Maurice DM. The biology of wound healing in the comeal stroma. Castroviejo Lecture. Comea 1987; 6:162-8. 9. Binder PS, Abel R Jr, Polack FM, Kaufman HE. Keratoplasty wound separations. Am J Ophthalmol1975; 80:109-15. 10. Binder PS. Selective suture removal can reduce postkeratoplasty astigmatism. Ophthalmology 1985; 92:1412-6. 11. Forstot SL, Abel R Jr, Binder PS. Bacterial endophthalmitis following suture removal after penetrating keratoplasty. Am J Ophthalmol 1975;80:509-12. 12. Weiss JL, Nelson JD, Lindstrom RL, Doughman DJ. Bacterial endophthalmitis following penetrating keratoplasty suture removal. Cornea 1984/1985; 3:278-80. 13. Gelender H. Bacterial endophthalmitis following cutting of sutures after cataract surgery. Am J Ophthalmol1982; 94:528-33. 14. Smith SG, Lindstrom RL, Nelson JD, et al. Comeal ulcer infiltrate associated with soft contact lens use following penetrating keratoplasty. Comea 1984; 3:131-4. 15. Ormerod LD, Smith RE. Contact lens-associated microbial keratitis. Arch Ophthalmol1986; 104:79-83. 16. Alfonso E, Mandelbaum S, Fox MJ, Forster RK. Ulcerative keratitis associated with contact lens wear. Am J Ophthalmol 1986; 101 :429-33. 17. Cohen EJ, Laibson PR, Arentsen JJ, Clemons CS. Comeal ulcers associated with cosmetic extended wear soft contact lenses. Ophthalmology 1987; 94:109-14. 18. Nelson JD. Epithelial problems. In: Brightbill FS, ed. Comeal Surgery: Theory, Technique, and Tissue. St Louis: Mosby, 1986; 292-304. 19. Singh G, Foster CS. Epidermal growth factor in alkali-bumed comeal epithelial wound healing. Am J Ophthalmol1987; 103:802-7. 20. Holmes V, Singh G, Jumblatt MM, et al. Comeal epithelial cell attachment to alkali-bumed keratectomy and basement membrane corneal substrates: a new model. ARVO Abstracts. Invest Ophthalmol Vis Sci 1986; 27(Suppl):31. 21. Nishida T, Nakagawa S, Manabe R. Clinical evaluation of fibronectin eyedrops on epithelial disorders after herpetic keratitis. Ophthalmology 1985; 92:213-6. 22. Adams GGW, Kirkness CM, Lee JP. Botulinum toxin A induced protective ptosis. Eye 1987; 1:603-8. 23. Harris DJ JR, Stulting RD, Waring GO III, Wilson LA. Late bacterial and fungal keratitis following comeal transplantation. Ophthalmology 1987; 94(Final Program Suppl):75. 24. Dannenberg AM Jr. The antiinflammatory effects of glucocorticosteroids. A brief review of the literature. Inflammation 1979; 3:329-43. 25. Rao GN, John T, Ishida N, Aquavello JV. Recovery of corneal sensitivity in grafts following penetrating keratoplasty. Ophthalmology 1985; 92:1408-11. 26. Sullivan WR. Routine postoperative management. In: Brightbill FS, ed. Comeal Surgery: Theory, Technique, and Tissue. St Louis: CV Mosby, 1986; 283-5. 27. Templeton WC III, Eilerman RA, Snyder JW, et al. Serratia Keratitis transmitted by contaminated eyedroppers. Am J Ophthalmol 1982; 93:723-6. 28. Gasset AR, Dohlman CH. The tensile strength of comeal wounds. Arch Ophthalmol 1968; 79:595-602. 29. Guss RB, Koenig S, De La Pen a W, et al. Endophthalmitis after penetrating keratoplasty. Am J Ophthalmol1983; 95:651-8.
1275
Abstract: A retrospective review of 68 consecutive episodes of microbial keratitis complicating 66 penetrating keratoplasties (PKs) showed major risk associations: suture-related problems (50%), contact lens wear (26%), previous herpes simplex infection (15%), graft failure (15%), and persistent epithelial defects (15%). Topical steroid (85%) and antibiotic (59%) usage were common iatrogenic factors. Half the infections occurred more than 1 year after grafting. Bacterial infections involving gram-positive organisms (59%) predominated, except for patients with extended-wear hydrophilic contact lenses, which usually involved gram-negative bacilli. The incidence of fungal infections (6%) was relatively low. Recommendations to minimize microbial keratitis include prompt attention to exposed, broken, or loose sutures, and preventive and therapeutic management of epithelial defects. The inadequacy of low-dose antibiotics in precluding microbial infection in many cases and the propensity to develop infections with resistant organisms suggest that guidelines for using postoperative and prophylactic topical antibiotics require reevaluation. [Key words: contact lens complications, endophthalmitis, keratoplasty, microbial keratitis, suture complications.] Ophthalmology 95: 1269-1275, 1988
Microbial keratitis after penetrating keratoplasty (PK) is an uncommon but serious complication. Although reported incidence rates are low, ranging from 1.8 1 to 4.9%,2 the visual sequelae of graft infection are frequently severe. Tuberville and Wood2 recorded a 46% rate of visual loss and only 14% achieved visual acuity greater than 6/60 in the study by Driebe and Stern. 3 This large retrospective study of microbial keratitis complicating PK was conducted to identify the associated risk factors, describe the microbiologic spectrum, and disOriginally received: February 16, 1988. Revision accepted: April 27, 1988. Cornea Service, Department of Ophthalmology, Massachusetts Eye and Ear Infirmary, Harvard Medical School, Boston. 2 Immunology and Uveitis Unit, Department of Ophthalmology, Massachusetts Eye and Ear Infirmary, Harvard Medical School, Boston. 3 Eye Research Institute, Boston. 1
Supported in part by a Hugh Noel Puckle Scholarship and grants EY06008 and EY05799 from the National Institutes of Health. Reprint requests to The Library, Eye Research Institute, 20 Staniford Street, Boston, MA 02114.
cuss the findings with respect to prevention, treatment, and complication management in corneal graft infections.
MATERIALS AND METHODS The medical records were reviewed from all corneal graft patients treated for microbial keratitis at the Massachusetts Eye and Ear Infirmary during an 8-year period (1978-1985). Cases caused by corneal exposure were excluded. Criterion for inclusion in this study was a discharge diagnosis of microbial keratitis in a keratoplasty patient that included either a positive corneal culture or a positive corneal smear Gram stain. Corneal scrapings, performed on all patients, were Gram stained and inoculated onto blood, chocolate and modified Sabouraud agar, and cooked meat broth using standard techniques. 4 ,5 Sabouraud agar was incubated at 25°C and the remainder at 35°C. Standard disc diffusion criteria for antibiotic sensitivity were used. 6 With the exception of fungi, cultures were considered positive 1269
OPHTHALMOLOGY
•
SEPTEMBER 1988 •
Table 1. Indications for Penetrating Keratoplasty in 66 Patients with Post keratoplasty Microbial Keratitis Indications Aphakic bullous keratopathy· Corneal scarringt Fuchs' endothelial dystrophyt Herpes simplex keratitis Keratoconus Trauma Corneal perforation (Sjogren's syndrome) Lattice corneal dystrophy Uveitic bullous keratopathy
No. of Patients
Prevalence (%)
19 13
29 20
11 10 5
17 15 8
2
3 2
4
1 1
6
2
* Including one pseudophakic bullous keratopathy. t Three patients with previous microbial keratitis, single cases with cicatricial pemphigoid, aniridia, sclerocornea, congenital leukoma, chlamydial infection, syphilitic interstitial keratitis, alkali burn, Stevens-Johnson syndrome, Goldenhar's syndrome, and a scarred lamellar keratoplasty (Salzmann's degeneration). t Not including aphakic cases.
when organisms were isolated from two or more media or from one culture medium if supported by an appropriate corneal smear Gram stain. A chart review was made using a standardized protocol with attention directed to the following: indications for PK; interval between PK and onset of microbial keratitis; microbiologic spectrum; complications; and detailed analysis of specific risk associations. Ulcer size was defined as small ( <2 mm greatest dimension), moderate (2-6 mm), and large (>6 mm). The location of the ulcers in donor or host cornea, or at the woundjunction, was recorded with details of the specific site.
RESULTS GENERAL EPIDEMIOLOGY
A total of 68 consecutive episodes of infection occurring in 66 patients after PK was reviewed. There were almost equal numbers of male and female patients. Right eyes were involved in 30 cases and left eyes in 38. The mean age was 61 years (range, 1-87 years). Thirtyseven patients (56%) were seen within 2 days of symptoms; nine individuals (13%) delayed attendance beyond 1 week. All patients were admitted and had a mean hospital stay of 10 days (range, 4-28 days). The indications for the previous corneal graft are shown in Table 1; the most common diagnoses were aphakic bullous keratopathy, miscellaneous corneal scarring, Fuchs' endothelial dystrophy, and herpes simplex keratitis. RISK FACfORS
Potential risk factors have been categorized under local ocular, topical medication, and systemic causes 1270
VOLUME 95
•
NUMBER 9
(Table 2) and identified under three major risk groups: suture (50%) and contact lens-related (26%) complications, and miscellaneous (24%). Forty-two (62%) of the infections occurred in aphakic patients. Two thirds (62%) of the suture-related infections involved abscesses around apparently secure sutures, although infections induced by loose or broken sutures and after suture removal also were seen (Table 3). Infections associated with sutures were more common within the palpebral fissure (Fig 1). Ten (7 with previous herpes simplex keratitis) of 18 episodes (56%) of contact lensrelated infections occurred in patients using therapeutic bandage lenses for persistent epithelial defects. Seven aphakic patients were using extended-wear and one patient a daily wear hydrophilic contact lens. Altogether, 14 of the 18 infections occurring in the contact lenswearing group were in aphakic patients. Of 16 patients in the miscellaneous group, 7 had recent graft rejections that had been treated with intensive topical corticosteroids, one Sjogren's syndrome, and one a wound dehiscence; 14 (88%) were receiving topical corticosteroid therapy. Except possibly for diabetes mellitus (12%) and chemotherapeutic immunosuppression (9%), systemic factors appeared insignificant. Two patients had recurrent ulcers. A Mycobacterium chelonei infection developed in one herpetic patient with a persistent epithelial defect, and a viridans streptococcal keratitis associated with bandage contact lens wear subsequently developed. The second patient had two infections involving different coagulase-negative staphylococcal biotypes complicating loose sutures. Two infections were in grafts performed for the sequelae of microbial keratitis occurring before the study period. The chronology of microbial keratitis after corneal transplantation is summarized for the three etiologic groups in Figure 2. Twenty-eight (41 %) infections occurred within 6 months of keratoplasty and 34 (50%) occurred within 1 year. The longest interval between surgery and infection was 13 years. Suture complications were the main causes of graft infections in the first year after keratoplasty. Eighty-five percent of 34 suturerelated infections occurred during this period, 38% within 2 months and 62% within 6 months of grafting. Problems related to contact lens wear, graft rejection, and corneal edema were the main late causes of infection; persistent epithelial defects were rarely associated at that time. The size and location of the microbial keratitis are shown in Table 4. Although suture-related infections occurred almost exclusively in the graft periphery, infections associated with bandage and extended-wear hydrophilic contact lens generally were localized to the center of the graft. Sutures were present in 10 of the 16 contact lens-infection grafts in which the suture status was known, but the sutures were not believed to be the source of infection in any of these cases. Infections generally remained confined by the grafthost interface; only 5 of 37 peripheral infections had relatively minor extensions of suppuration into the host corneas. Five infections, occurring at 7, 12, and 14
FONG et al
•
CORNEAL GRAFT INFECTIONS
Table 2. Predisposing Factors for Microbial Keratitis in 68 Corneal Graft Infections No. of Infections
Predisposing Factors
Suture Related (n = 34)
Contact Lens Related (n = 18)
Miscellaneous (n = 16)
Total (n = 68)
Prevalence (%)
0 0 7 6 0 2 1 1 1
34 18 11 10 10 4 3 3 1
50 26 16 15 15 6 4 4 1
Local ocular Suture-related problems Contact lens Recent graft rejection episode Corneal edema (graft failure) Persistent epithelial defect Trichiasis Mucosal scarring disorders* Wound dehiscence Sjogren's syndrome
34 0 3 2 (1)+ 1 1 2 0
(2)+ 18 1 2 10 1 2 (1)+ 0 0
Topical medication Corticosteroid Antibiotic Glaucoma medication(s) Antiviral
30 24 4 1
13 10 5 3
14 6 4 2
57 40 13 6
84 59 19 9
6 1 1 1
1 2 1 0
1 3 0 0
8 6 2 1
12 9 3 2
Systemic factors Diabetes mellitus Systemic immunosuppressiont Atopy Chronic alcoholism
* One cicatricial pemphigoid, one Stevens-Johnson syndrome, and one alkali burn patient. t Five oral corticosteroids ~ 10 mg daily (graft rejection, 2; herpes simplex, 2; asthma 1) and one azathioprine. + Associated factors.
Table 3. Analysis of Suture-related Factors in 34 Corneal Graft Infections Suture-associated Factors*
No. of Cases
Prevalence (%)
Suture abscess around secure suture(s) Loose suture(s)t Ruptured suture Recent suture removal
21 8 3 2
62 24 9 6
SUPERIOR
TEMPORAL
NASAL
* Usually 10-0 or 11-0 monofilament nylon. Comprises 11 interrupted and 11 continuous sutures; the suture type in 12 cases was unspecified. t May be underestimated in this retrospective series.
months and 3 and 9 years after transplantation, appeared to have arisen in the scar with approximately equal involvement of donor and host tissue. In three otherwise unremarkable cases, microbial keratitis was restricted to the host cornea abutting the wound margin.
INFERIOR
MICROBIAL PATHOGENESIS
Fig 1. Corneal segment involvement of suture-related infections complicating 32 corneal grafts. Notice that there was also one case of multifocal infection and one of extensive infection.
Of 68 infections, 66 (97%) were culture-positive; two had positive Gram stains confirming bacterial infection, although no organisms were isolated. Sixty-four (94%) cases were bacterial and four (6%) fungal in origin (Table 5). Gram-positive organisms were involved in
59% of the (culture-positive) infections, predominantly Staphylococcus aureus (21 %), coagulase-negative staphylococci (18%), Streptococcus pneumoniae (7%), and {j1271
OPHTHALMOLOGY
•
SEPTEMBER 1988
MISCELLANEOUS 10 5 til
;::
~
...><
0
It
CONTACT LENS· RELATED
....
'iii"
0
10
It
U
~
...
5
til
0
0
W
Q
0
SUTURE · RELATED
til
ii: w
"-
0
30
It W
'"~z
25 20
•
VOLUME 95
•
NUMBER 9
infections, and 50% of the miscellaneous group; the equivalent figures (including polymicrobial infections) for gram-negative bacillary infection were 22, 56, and 50%, respectively. Six often (60%) bandage contact lens infections were associated with gram-positive bacterial infection, and six of eight (75%) aphakic hydrophilic contact lens infections had gram-negative isolates. Polymicrobial infection was uncommon. All four fungal infections were suture related. There were no episodes of infectious crystalline keratopathy over this time period. In this study, the sensitivity of the Gram stain was 75% and the specificity of a positive Gram stain was 88%. Twenty-five (63%) of the 40 cases receiving topical antibiotics before the onset of infection had bacterial isolates that were resistant in vitro to the antibiotics. Polysporin (polymyxin (3 and bacitracin; Burroughs Wellcome, Research Triangle Park, NC), erythromycin, gentamicin, or chloramphenicol was commonly used for this purpose.
15 10
COMPLICATIONS 10.2 2·4
4· 6 6·8 8·10 10.1?
FIRST I
SECONO
MONTHS
2·4
4·6
6·8
8·10 I
YEARS
Fig 2. Time interval between penetrating keratoplasty and microbial infection.
hemolytic streptococci (6%). Of the gram-negative infections (38% of 68), Serratia marcescens (12%), Pseudomonas aeruginosa (10%), and Klebsiella sp (4%) were the most common isolates. Gram-positive cocci were isolated from 63% of the suture-related infections, 50% ofthe contact lens-related
The prevalence of major complications of microbial keratitis is presented in Table 6. Although limited by relatively small numbers, the rates for corneal perforation, enucleation/evisceration, wound dehiscence, failure of previously clear graft, and the requirement for emergency regraft suggest that the suture-related infections tended to have a worse prognosis than the other groups. Overall, one third of the patients required replacement of the corneal graft. Of previously clear grafts, 17% became edematous, either as a direct sequela of the infection or as a result of unrecognized, superimposed graft rejection. Endophthalmitis occurred in four
Table 4. Size and Location of Microbial Keratitis Complicating Corneal Grafts No. of Patients (%) Suture Related (n = 34)
Contact Lens Related (n = 18)
Miscellaneous (n = 16)
Size (mm) Small «2 in greatest dimension) Moderate (2-6) Large (>6)
14 (41) 18 (53) 2 (6)
7 (39) 7 (39) 4 (22)
6 (38) 9 (56) 1 (6)
27 (40) 34 (50) 7 (10)
Location Limited to donor cornea Central/paracentralPeripheral Overwhelming infection Equal donor and host corneal involvement Limited to host cornea
1 27t 1 3 2
4 8 1 2 1
20 (29) 37t (54) 3 (4) 5 (7) 3 (4)
- Not contiguous with wound edge. Remote from sutures in all but four. t All but four suture-related cases extending to wound edge. tlncluding five cases with minor extension across the wound into the host cornea.
1272
15 2 1 0 0
(n
Total = 68)
FONG et al
•
CORNEAL GRAFT INFECTIONS
Table 5. Microbial Etiology of 66 Episodes of Culture-positive Keratitis Complicating Corneal Grafts No. of Infections*
Organisms Gram-positive cocci Staphylococcus aureus Coagulase-negative staphylococci Streptococcus pneumoniae {j-hemolytic streptococci Viridans streptococci Gram-negative bacilli Serratia marcescens Pseudomonas aeruginosa Klebsiella sp Enterobacter aerogenes Mycobacterium chelonei Proteus mirabilis Morganella morgagnii Haemophilus inDuenzae Fungi Candida albicans Aspergillus sp Penicillium sp
Suture Related (n = 32)
Contact Lens Related (n = 18)
Miscellaneous (n = 16)
(n
Total = 66)
Prevalence (%)
10 4 4 2 1
1 6 (1) 0 1 1
3
2 1 1 1
14 12 (1) 5 4t 3
21 18 7 6 4
4 2 0 0 0 0 1 0
2 3 2 2 (1) 1 0 0 0
2 2 1 0 1 1 0 1
8 7 3t 2 (1) 2 1 1 1
12 10 4
3
0 0 0
0 0 0
1 (1) 1 (1)
1 (1) 1 (1)
3
3 1 1 1
3
4 1 1
* Polymicrobial isolates are in parentheses. tOne Lancefield group B, one group D, one group G, and one unclassified. t Two Klebsiella oxytoca and one Klebsiella pneumoniae. Table 6. Major Complications of Corneal Graft Infections No. of Patients (%)
Hypopyon Descemetocele Corneal perforation Glue and bandage contact lens Endophthalmitis Enucleation/evisceration Wound dehiscence* Failure of previously clear graftt Loss of light perception Emergency repeat PK Elective repeat PK Total repeat PK
Suture Related (n = 34)
Contact Lens Related (n = 18)
Miscellaneous (n = 16)
10 (29) 1 (3) 5 (15) 3 (9) 2 (6) 4 (12) 9 (26) 7 (21) 4 (12) 9 (26) 3 (9) 12 (35)
2 (11) 1 (6) 2 (11) 1 (6) 1 (6) 1 (6) 1 (6) 3 (17) 2 (11) 2 (11) 3 (17) 5 (28)
4 (25) 2 (13) 1 (6) 1 (6) 1 (6) 1 (6) 6 (38) 1 (6) 1 (6) 2 (13) 3 (19) 5 (31)
(n
Total = 68)
16 (24) 4 (6) 8 (12) 5 (7) 4 (6) 6 (9) 16 (24) 11 (16) 7 (10) 13 (19) 9 (13) 22 (32)
PK = penetrating keratoplasty. * Includes corneal perforations (but not central perforations) occurring at the PK wound edge. t Irreversible corneal edema.
patients (6%) and was associated with moderate-to-Iarge ulcers; two cultured ,a-hemolytic streptococci, one P. aeruginosa, and one Klebsiella oxytoca. Endophthalmitis complicated, respectively, a suture-related infection 6 weeks after a graft performed for Fuchs' dystrophy, a suture-related infection at 7 months in a graft under-
taken for corneal scarring, intensive topical steroid therapy for graft rejection 16 months after PK (aphakic bullous keratopathy), and bandage contact lens wear in a failed graft 4 years after surgery. Intraocular infection was confirmed at vitrectomy in three of these patients and by anterior chamber tap in the fourth. 1273
OPHTHALMOLOGY
•
SEPTEMBER 1988 •
DISCUSSION This study emphasizes the importance of exposed, loose, and broken sutures in the generation of postkeratoplasty microbial keratitis. Of our graft infections, 50% were suture related, compared with 46% in the report by Tuberville and Wood2 and 14% by Driebe and Stern. 3 Exposed sutures breach the epithelial surface and provide a direct pathway into the corneal stroma. Furthermore, abnormal mucus accumulation and debris may act as a nidus for microbial colonization. Operative factors, such as the failure to bury surgical knots and suture ends, improper suture tension or tying, and poor hostdonor alignment at the epithelial surface also may be important. Corneal deturgescence, wound remodeling,7.B and scar contracture occur during wound healing, and resultant forces may loosen or expose sutures. Suture biodegradation and rupture also can occur postoperatively. Because 19% of the cases ( 13 infections) occurred within 2 months postoperatively, early postoperative changes in wound configuration are considered major etiologic factors. Even though the superficial arc of the suture becomes rapidly buried, it is also possible that the overlying epithelium might sometimes remain unstable, with a propensity for exposure of the suture material. One exrlanation for the nasal and temporal predisposition of suture-related infections is the greater relative exposure in the interpalpebral fissure; another explanation might be the increased technical difficulty of suturing at these sites. A similar distribution has been noted for keratoplasty wound separations. 9 The much lower infection rates of superior and inferior sutures suggest that lid coverage may be a protective factor. Suture removal was the precipitating factor in two of our patients; this problem may increase because of the greater frequency of suture removal proposed for astigmatic control. 10 Although wound separation (and possible microbial infection) may occur with suture removal, particularly in the early postoperative period,9 suture manipulation alone may introduce bacteria from the ocular surface and induce keratitis or endophthalmitis. I1-13 We recommend a short course of topical bactericidal antibiotic, such as gentamicin, after suture removal. Most contact lens-associated microbial keratitis was related to the use of bandage lenses in treating persistent epithelial defects. These infections were characterized by a predominance of gram-positive bacteria (59%), notably S. aureus and coagulase-negative staphylococci. The gram-positive bacterial pattern in bandage lens wear was noted previously,14,15 and probably represents endogenous infection with ocular flora. In contrast, the microbial keratitis complicating aphakic extended-wear hydrophilic contact lenses was caused mainly by gram-negative organisms (75%), which compares with the 78% incidence reported by Alfonso et al 16 in their contact lens-associated microbial keratitis series and 75% in a recent study restricted to infections complicating the use 1274
VOLUME 95
•
NUMBER 9
of extended-wear cosmetic contact lenses. 17 Factors such as contact lens handling and contaminated contact lens cases/solutions probably are involved in explaining this different disease spectrum. No rigid contact lens wearers acquired infection in this study. It was notable that the graft infections associated with contact lens use largely occurred in the graft center, perhaps related to central corneal hypoxia under the contact lens. Despite the frequent concurrence of sutures, the cause of infection in these cases did not appear to involve an increased risk of suture exposure with contact lens wear. The distinct sites of involvement in suture-related and contact lens-related graft infections imply differing pathogenetic mechanisms. Although the use of aphakic contact lens correction may be diminishing after PK because of combined intraocular lens insertion, the complications of bandage contact lens wear will continue to be problematic. The risk of a persistent epithelial defect developing can be minimized by caring for the corneal epithelium during surgery, using preservative-free solutions to avoid medication toxicity, and aggressive lubrication. 14 Conservative occlusive therapy, consideration of early tarsorrhaphy, and the prophylactic correction of preexisting factors such as exposure, sicca, trichiasis, and external ocular inflammation are all important. 14,IB Clinical and experimental trials to assess the efficacy of topical fibronectin and epidermal growth factor to enhance epithelial healing l9-21 and botulinum toxin A-induced protective ptosis22 promise further therapeutic improvements. Fungal infections were uncommon (6%) in our series, reflecting the low incidence of fungal infection in the northeastern United States; a high incidence of fungal graft infection is recorded in some,I,21 but not in all,3 studies from the southern United States. All fungal infections were suture related: an important clinical point is that fungal infection may recede from the surface into the suture tract. Long-term prophylactic antibiotics were used in 59% of the corneal transplantation patients with microbial keratitis, but did not prevent infection. An important finding was that almost two thirds of the graft infections complicating the use of prophylactic antibiotics involved bacterial strains resistant to those antibiotics. Similar findings have been reported in other PK keratitis studies. 3,14,23 The possibility that the long-term use of topical antibiotics may alter the normal flora, select resistant strains of bacteria, or locally induce antimicrobial resistance requires urgent attention. Studies should be conducted to delineate the risk-benefit ratio of longterm topical antibiotic and/or corticosteroid regimens in ocular surface disease. Topical corticosteroid usage was a common factor (84%) in the graft infections, although the concurrence of multiple factors makes it difficult to interpret its specific role. However, corticosteroids impair host-defense and healing mechanisms and promote microbial superinfection24 ; they also may reduce the symptoms and signs of keratitis (as may poor graft reinnervation 25 ), encouraging more extensive disease to develop. There is
FONG et al
•
CORNEAL GRAFT INFECTIONS
no general consensus on the duration and intensity of postoperative steroids after corneal transplantation. 26 The contamination of topical medications also can be important. 27 At the time of the study (January 1978-December 1985), 2006 penetrating keratoplasties were performed at the Massachusetts Eye and Ear Infirmary. Sixty-six corneal grafts (recurrent infections in 2 grafts) were complicated by microbial keratitis over this period. An approximate incidence of 3.3% can therefore be calculated, which, because of delayed infection and the progressively increasing frequency of PK surgery at this institution, should probably be considered as a minimal estimate. The figure is comparable with previous reports,I,2,23 if somewhat less than another. 14 Although the incidence is low, the toll of ocular morbidity is disproportionately high. 2,3 Steroid-induced local immunosuppression and delayed wound healing,IO,28 compounded by the high incidence of suture-related infections, may lead to a loss of structural integrity and wound dehiscence (24%), perforation (12%), and the high rates for emergency regrafting (19%) and enucleation/evisceration (9%). The data suggest a worse prognosis for suture-related than contact lens-related graft infections, possibly related to their proximity to the graft edge and also to their occurrence earlier in the wound healing process. Endophthalmitis developed in a very large proportion of perforated PK infections as noted previously.29 This study has shown a major association of postkeratoplasty microbial keratitis to suture-related problems, which emphasizes the need for meticulous surgery, prompt patient recognition of new symptoms, and the urgent attention to exposed, loose, and broken sutures. The identification of patients at special risk, with persistent epithelial defects, contact lens wear, failed grafts, and graft rejection episodes also can improve preventive management. Only increased vigilance and patient education can prevent the disaster of microbial keratitis complicating corneal transplantation. Because of the high risk of visual loss, these patients should be admitted to the hospital, with the exception of patients with a small infection who are highly compliant; these individuals can be treated with subconjunctival or intensive topical antibiotics, or both, and seen every day. The inadequacy oflong-term antibiotics in preventing infections and the risk of inducing an antibiotic-resistant conjunctival flora necessitate reappraisal of the optimal use of postoperative and prophylactic antibiotics.
REFERENCES 1. Lamensdorf M, Wilson LA, Waring GO, et al. Microbial keratitis after penetrating keratoplasty. Ophthalmology 1982; 89(Sept. Suppl): 124. 2. Tuberville AW, Wood TO. Comeal ulcers in comeal transplants. Curr Eye Res 1981; 1:479-85. 3. Driebe WT Jr, Stem GA. Microbial keratitis following comeal transplantation. Comea 1983; 2:41-5. 4. Jones DB, Liesegang TJ, Robinson NM. Laboratory diagnosis of
ocular infections, Cumitech 13. Washington, DC: American Society for Microbiology, 1981. 5. Lennette EH, Balows A, Hausler WJ Jr, Truant JP. Manual of Clinical Microbiology. 3rd ed. Washington, DC: American Society for Microbiology, 1980. 6. National Committee for Clinical Laboratory Standards. Performance Standards for Antimicrobial Disc Susceptibility Tests (Approved Standard, M2A2). 2nd ed. Villanova, PA: The Committee, 1979. 7. Davison PF, Galbavy EJ. Connective tissue remodeling in comeal and scleral wounds. Invest Ophthalmol Vis Sci 1986; 27:1478-84. 8. Maurice DM. The biology of wound healing in the comeal stroma. Castroviejo Lecture. Comea 1987; 6:162-8. 9. Binder PS, Abel R Jr, Polack FM, Kaufman HE. Keratoplasty wound separations. Am J Ophthalmol1975; 80:109-15. 10. Binder PS. Selective suture removal can reduce postkeratoplasty astigmatism. Ophthalmology 1985; 92:1412-6. 11. Forstot SL, Abel R Jr, Binder PS. Bacterial endophthalmitis following suture removal after penetrating keratoplasty. Am J Ophthalmol 1975;80:509-12. 12. Weiss JL, Nelson JD, Lindstrom RL, Doughman DJ. Bacterial endophthalmitis following penetrating keratoplasty suture removal. Cornea 1984/1985; 3:278-80. 13. Gelender H. Bacterial endophthalmitis following cutting of sutures after cataract surgery. Am J Ophthalmol1982; 94:528-33. 14. Smith SG, Lindstrom RL, Nelson JD, et al. Comeal ulcer infiltrate associated with soft contact lens use following penetrating keratoplasty. Comea 1984; 3:131-4. 15. Ormerod LD, Smith RE. Contact lens-associated microbial keratitis. Arch Ophthalmol1986; 104:79-83. 16. Alfonso E, Mandelbaum S, Fox MJ, Forster RK. Ulcerative keratitis associated with contact lens wear. Am J Ophthalmol 1986; 101 :429-33. 17. Cohen EJ, Laibson PR, Arentsen JJ, Clemons CS. Comeal ulcers associated with cosmetic extended wear soft contact lenses. Ophthalmology 1987; 94:109-14. 18. Nelson JD. Epithelial problems. In: Brightbill FS, ed. Comeal Surgery: Theory, Technique, and Tissue. St Louis: Mosby, 1986; 292-304. 19. Singh G, Foster CS. Epidermal growth factor in alkali-bumed comeal epithelial wound healing. Am J Ophthalmol1987; 103:802-7. 20. Holmes V, Singh G, Jumblatt MM, et al. Comeal epithelial cell attachment to alkali-bumed keratectomy and basement membrane corneal substrates: a new model. ARVO Abstracts. Invest Ophthalmol Vis Sci 1986; 27(Suppl):31. 21. Nishida T, Nakagawa S, Manabe R. Clinical evaluation of fibronectin eyedrops on epithelial disorders after herpetic keratitis. Ophthalmology 1985; 92:213-6. 22. Adams GGW, Kirkness CM, Lee JP. Botulinum toxin A induced protective ptosis. Eye 1987; 1:603-8. 23. Harris DJ JR, Stulting RD, Waring GO III, Wilson LA. Late bacterial and fungal keratitis following comeal transplantation. Ophthalmology 1987; 94(Final Program Suppl):75. 24. Dannenberg AM Jr. The antiinflammatory effects of glucocorticosteroids. A brief review of the literature. Inflammation 1979; 3:329-43. 25. Rao GN, John T, Ishida N, Aquavello JV. Recovery of corneal sensitivity in grafts following penetrating keratoplasty. Ophthalmology 1985; 92:1408-11. 26. Sullivan WR. Routine postoperative management. In: Brightbill FS, ed. Comeal Surgery: Theory, Technique, and Tissue. St Louis: CV Mosby, 1986; 283-5. 27. Templeton WC III, Eilerman RA, Snyder JW, et al. Serratia Keratitis transmitted by contaminated eyedroppers. Am J Ophthalmol 1982; 93:723-6. 28. Gasset AR, Dohlman CH. The tensile strength of comeal wounds. Arch Ophthalmol 1968; 79:595-602. 29. Guss RB, Koenig S, De La Pen a W, et al. Endophthalmitis after penetrating keratoplasty. Am J Ophthalmol1983; 95:651-8.
1275