Table 5 summarizes the preoperative and the postoperative visual acuity. There was no significant difference in preoperative visual acuity between the groups. However, compared with the non-PCED group, the PCED group had a higher incidence of a final visual acuity worse than 20/200 (76.2% vs. 48.8%, P = 0.024).
Pars plana vitrectomy may offer the opportunity for visual recovery in many vitreoretinal diseases that threaten vision. However, corneal complications, such as corneal ED, corneal edema, and superficial punctate keratopathy, are major early morbidities after PPV. The overall rate of corneal complication was 22.4% in our study. In most of the cases, early corneal complications can usually be reversed if recognized early and properly treated; however, 4.6% of the eyes developed PCED in this study. In multivariate analysis, the risk factors associated with PCED were diabetes, assistance by a first-year resident, and intravitreal tamponade with C3F8.
The reported incidence of PCED after PPV using a contact viewing lens ranges from 5.1% to 11.3%,1,6,7 which is slightly higher than in this study (4.6%). A possible explanation is that there was a lower rate of intraoperative epithelial debridement in this study. In previous reports, 2.3% to 32% of patients needed intraoperative epithelial debridement to improve visualization during PPV.1,6–8 However, Virata et al7 found that 6 of 14 (42.9%) of corneal ED due to epithelial debridement progressed to PCED without healing within 2 weeks. Because of the awareness of our surgeons and improvements in instrumentation and surgical techniques, we rarely perform intraoperative epithelial debridement.
Assistance by a first-year resident was associated with a higher risk of PCED in this study. This may be because the junior residents had less experience in holding or manipulating the contact viewing lens, which may have caused more corneal trauma during surgery. In our department, one resident or one fellow would be assigned to the vitreoretinal team every month. This resident or fellow was responsible for holding the contact lens viewing system in all of the vitreoretinal surgeries within this month, regardless the duration of the surgery. Neither nurse nor other vitreoretinal specialist would help to hold the contact lens viewing system. A noncontact lens viewing system is not available in our hospital. Furthermore, we also performed a statistic analysis comparing the operative time between first-year-resident assisting cases and non–first-year-resident assisting cases. There was no significant difference found (mean operative time: 128.80 minutes vs. 121.55 minutes, P = 0.552). Thus, the association between first-year-resident assistant and PCED was unlikely caused by different operative time between first-year residents and senior residents.
Direct intraoperative injury was a major cause of PCED in our study. We found that most cases of PCED arose from acute-onset (17/21 eyes, 81%) rather than delayed-onset (4/21 eyes, 19%) ED. Apart from the eyes that underwent intraoperative epithelial debridement, the other cases of acute-onset ED were likely caused by direct intraoperative mechanical corneal trauma by the contact viewing lens. Virata et al7 noted that the incidence of corneal ED after PPV can be reduced by avoiding the use of a contact viewing lens. Therefore, we suggest that when training a junior resident in a teaching hospital, a noncontact viewing system may be better to protect corneas in high-risk patients.
Diabetes mellitus has been reported to be a risk factor for corneal ED in previous studies.1,6–9 Our study also demonstrated that diabetes mellitus was significantly associated with the development of PCED. This makes sense mechanistically, because diabetic corneas have previously been reported to have delayed wound healing,10,11 impairment in epithelial barrier function,12–14 abnormalities in epithelial basement membrane,15–17 and abnormalities in the shape of epithelial and endothelial cells.18–26 Thus diabetic eyes may be more vulnerable to corneal injury during surgery. Someone might suspect that increased operative time for more complicated surgeries in patients with diabetes could also increase the risk of PCED. So, we further analyzed our data; there was no significant difference in mean operative time between patients with diabetes and patients without diabetes (120.45 minutes vs. 127.49 minutes, P = 0.495). There was also no significant difference between PCED and non-PCED groups in the mean operative time (120.84 minutes vs. 134.67 minutes, P = 0.278) (Table 2). When we put operative time into the multivariate logistic regression model, it did not show statistical significance (Table 3). Thus, this might imply that the PCED may be more likely to associate with the compromised corneal condition in diabetes rather than longer operative time in these patients.
Our results demonstrated that the occurrence of PCED was significantly greater in the eyes using C3F8 tamponade. However, other tamponade substances such as air, SF6, and silicone oil were not associated with PCED. In subgroup analysis, we found that the association of C3F8 and PCED was only significant in the aphakic or pseudophakic patients (P < 0.001), but not in the phakic patients (P = 0.089). Intraocular tamponade of long-acting expansile gasses in the presence of an open anterior lenticular capsule may induce corneal endothelial cell toxicity22,23,27–31 and corneal abnormalities,6,7 such as corneal edema and corneal opacity. We hypothesize that PCED may result from the direct toxicity of C3F8 when in contact with the cornea. The loss of corneal endothelial cells has also been reported to be significantly greater in eyes with C3F8 than in those with SF6.27 Intraocular C3F8 gas tamponade was a strong predictor of PCED in our study. Taken together, these results indicate that the use of C3F8 should be more conservative in aphakic and pseudophakic patients.
Elevated intraocular pressure may cause corneal edema and increase the risk of corneal complications.24,25 Glaucoma eye drops may also be toxic to the cornea, and this may contribute to the formation of superficial punctate keratopathy and corneal erosions in some patients.26,32 In our study, postoperative ocular hypertension requiring glaucoma medication was a risk factor for PCED in univariate analysis but not in multivariate analysis. This may be, at least in part, explained by the fact that postoperative ocular hypertension was associated with C3F8 tamponade (P = 0.034) in this study, demonstrating that C3F8 tamponade was a much stronger risk factor for PCED compared with previous studies.
In our study, 11/21 eyes (52%) with PCED failed to heal within 4 weeks after conventional medical treatment. These recalcitrant PCEDs tended to have a much higher incidence of serious late complications, such as corneal opacity, infectious keratitis, endophthalmitis, and even phthisis. This suggests that PCEDs persisting for more than 4 weeks is a critical cutoff point for the development of corneal infections and opacity.
To the best of our knowledge, this is the first report showing a high incidence of herpetic keratitis in recalcitrant PCED after PPV. Among 11 eyes with PCED lasting for longer than 4 weeks, 4 eyes (36%) progressed to geographic herpetic keratitis. After vitrectomy, topical corticosteroids are usually prescribed to control intraocular inflammation, and it has been reported that geographic HSV epithelial keratitis is more likely to occur than HSV dendritic keratitis under topical corticosteroid treatment.33 Thus, it may be more difficult to differentiate post-PPV herpetic keratitis from noninfectious PCED. Geographic HSV epithelial keratitis was not initially recognized in one of our patients. This eye had severe corneal melting and required repeated amniotic membrane transplantations. Early diagnosis and aggressive antiviral treatment of geographic HSV epithelial keratitis is important for many reasons. First, geographic ulcers usually take longer to heal.33 Second, when geographic HSV keratitis is associated with stromal inflammation, it usually results in other complications, such as corneal scarring, secondary glaucoma, and cataracts. Third, it also eliminates the beneficial effects of vitrectomy in visual outcomes. Therefore, we suggest that herpetic keratitis should always be taken into consideration in PCEDs that persist for longer than 4 weeks after PPV.
Prolong use of topical corticosteroid after vitrectomy could also potentially retard the corneal epithelial healing and increase the risk of herpetic keratitis. In our hospital, an antibiotic–steroid combination ointment (Tobradex ophthalmic ointment; Alcon Laboratories, Inc) was used twice daily for 3 to 4 weeks after vitrectomy to control postoperative inflammation. However, in patient with corneal ED for 1 week or more, the corticosteroid would either be early tapered or be used with extremely cautious. Because of the retrospective nature in this study, it is difficult to determine whether the topical corticosteroid increased the corneal complications from our data. In the literature, there is still no consensus on the optimal corticosteroid formula and duration of use after vitrectomy.34,35 Also, it needs further studies to determine whether topical corticosteroid would retard the corneal ED after vitrectomy.36 However, to facilitate corneal epithelial healing and prevent herpetic keratitis, it is reasonable to taper the corticosteroid once the anterior chamber is free of cell and flare.
Our study showed that patients with PCED had worse visual outcome than those without PCED. Persistent corneal epithelial defects may lead to further corneal infections and corneal opacity. Although this is a retrospective study with a relatively small sample size, it provides important information on the risk factors, clinical course, and prognosis of PCED after PPV.
The limitations to this study include its retrospective design and that the diagnosis of geographic herpes simplex keratitis from PCED after PPV in four of five infectious keratitis cases was a clinical diagnosis with a lack of laboratory evidence of corneal HSV infection. Furthermore, information on systemic morbidities such as the presence of diabetes mellitus was obtained by questionnaires in some patients without determining the actual level of hemoglobin A1c level or blood sugar.
In conclusion, the incidence of PCED after PPV was 4.6% in this study, and the eyes with PCED had worse visual outcomes. Diabetes, assistance by a first-year resident, and intravitreal tamponade with C3F8 were the risk factors associated with PCED in multivariate analysis. Our results also revealed that a healing time of PCED of more than 4 weeks was a cutoff point for the development of corneal opacity and a high incidence of geographic HSV epithelial keratitis. Early and aggressive management is necessary for patients presenting with PCED after vitrectomy to prevent poor outcomes.
1. Perry HD, Foulks GN, Thoft RA, et al.. Corneal complications after closed vitrectomy through the pars plana. Arch Ophthalmol 1978;96:1401–1403.
2. Berman M. The pathogenesis of corneal epithelial defects. Acta Ophthalmologica Suppl 1989;192:55–64.
3. Tsubota K, Goto E, Shimmura S, et al.. Treatment of persistent corneal epithelial defect
by autologous serum application. Ophthalmology 1999;106:1984–1989.
4. Randleman JB, Hewitt SM, Song CD. Corneal and conjunctival changes after posterior segment surgery. Ophthalmol Clin North Am 2004;17:513–520, v.
5. Hiraoka M, Amano S, Oshika T, et al.. Factors contributing to corneal complications after vitrectomy in diabetic patients. Jpn J Ophthalmol 2001;45:492–495.
6. Chung H, Tolentino FI, Cajita VN, et al.. Reevaluation of corneal complications after closed vitrectomy. Arch Ophthalmol 1988;106:916–919.
7. Virata SR, Kylstra JA, Singh HT. Corneal epithelial defects following vitrectomy surgery using hand-held, sew-on, and noncontact viewing lenses. Retina 1999;19:287–290.
8. Friberg TR, Ohji M, Scherer JJ, et al.. Frequency of epithelial debridement during diabetic vitrectomy. Am J Ophthalmol 2003;135:553–554.
9. Foulks GN, Thoft RA, Perry HD, et al.. Factors related to corneal epithelial complications after closed vitrectomy in diabetics. Arch Ophthalmol 1979;97:1076–1078.
10. Chikama T, Wakuta M, Liu Y, et al.. Deviated mechanism of wound healing in diabetic corneas. Cornea 2007;26:S75–S81.
11. Schultz RO, Van Horn DL, Peters MA, et al.. Diabetic keratopathy. Trans Am Ophthalmol Soc 1981;79:180–199.
12. Gekka M, Miyata K, Nagai Y, et al.. Corneal epithelial barrier function in diabetic patients. Cornea 2004;23:35–37.
13. Gobbels M, Spitznas M, Oldendoerp J. Impairment of corneal epithelial barrier function in diabetics. Graefes Arch Clin Exp Ophthalmol 1989;227:142–144.
14. Yokoi N, Niiya A, Komuro A, et al.. Effects of aldose reductase inhibitor CT-112 on the corneal epithelial barrier of galactose-fed rats. Curr Eye Res 1997;16:595–599.
15. Taylor HR, Kimsey RA. Corneal epithelial basement membrane changes in diabetes. Invest Ophthalmol Vis Sci 1981;20:548–553.
16. Hatchell DL, Magolan JJ Jr, Besson MJ, et al.. Damage to the epithelial basement membrane in the corneas of diabetic rabbits. Arch Ophthalmol 1983;101:469–471.
17. Azar DT, Spurr-Michaud SJ, Tisdale AS, et al.. Altered epithelial-basement membrane interactions in diabetic corneas. Arch Ophthalmol 1992;110:537–540.
18. Tsubota K, Yamada M. The effect of aldose reductase inhibitor on the corneal epithelium. Cornea 1993;12:161–162.
19. Meyer LA, Ubels JL, Edelhauser HF. Corneal endothelial morphology in the rat. Effects of aging, diabetes, and topical aldose reductase inhibitor treatment. Invest Ophthalmol Vis Sci 1988;29:940–948.
20. Matsuda M, Awata T, Ohashi Y, et al.. The effects of aldose reductase inhibitor on the corneal endothelial morphology in diabetic rats. Curr Eye Res 1987;6:391–397.
21. Hosotani H, Ohashi Y, Yamada M, et al.. Reversal of abnormal corneal epithelial cell morphologic characteristics and reduced corneal sensitivity in diabetic patients by aldose reductase inhibitor, CT-112. Am J Ophthalmol 1995;119:288–294.
22. Friberg TR, Doran DL, Lazenby FL. The effect of vitreous and retinal surgery on corneal endothelial cell density. Ophthalmology 1984;91:1166–1169.
23. Mittl RN, Koester CJ, Kates MR, et al.. Endothelial cell counts following pars plana vitrectomy
in pseudophakic and aphakic eyes. Ophthalmic Surg 1989;20:13–16.
24. Ytteborg J, D C. Corneal edema and intraocular pressure. I. Animal experiments. Arch Ophthalmol 1965;74:375–381.
25. Ytteborg J, D C. Corneal edema and intraocular pressure. II. Clinical results. Arch Ophthalmol 1965;74:477–484.
26. Fraunfelder FW. Corneal toxicity from topical ocular and systemic medications. Cornea 2006;25:1133–1138.
27. Mitamura Y, Yamamoto S, Yamazaki S. Corneal endothelial cell loss in eyes undergoing lensectomy with and without anterior lens capsule removal combined with pars plana vitrectomy
and gas tamponade. Retina 2000;20:59–62.
28. Sternberg P Jr, Hatchell DL, Foulks GN, et al.. The effect of silicone oil on the cornea. Arch Ophthalmol 1985;103:90–94.
29. Benson WE, Diamond JG, Tasman W. Intraocular irrigating solutions for pars plana vitrectomy
. A prospective, randomized, double-blind study. Arch Ophthalmol 1981;99:1013–1015.
30. McDermott M, Snyder R, Slack J, et al.. Effects of intraocular irrigants on the preserved human corneal endothelium. Cornea 1991;10:402–407.
31. Diddie KR, Schanzlin DJ. Specular microscopy in pars plana vitrectomy
. Arch Ophthalmol 1983;101:408–409.
32. Inoue K, Okugawa K, Kato S, et al.. Ocular factors relevant to anti-glaucomatous eyedrop-related keratoepitheliopathy. J Glaucoma 2003;12:480–485.
33. Wilhelmus KR, Coster DJ, Donovan HC, et al.. Prognosis indicators of herpetic keratitis. Analysis of a five-year observation period after corneal ulceration. Arch Ophthalmol 1981;99:1578–1582.
34. Paccola L, Jorge R, Barbosa JC, et al.. Anti-inflammatory efficacy of a single posterior subtenon injection of triamcinolone acetonide versus prednisolone acetate 1% eyedrops after pars plana vitrectomy
. Acta Ophthalmol Scand 2007;85:603–608.
35. Nagpal M, Lambat S, Mehrotra N, et al.. Topical nepafenac 0.1% alone versus prednisolone acetate 1% as postoperative anti-inflammatory agents in small gauge vitrectomy. Indian J Ophthalmol 2014;62:606–609.
36. Yülek F, Ozdek S, Gürelik G, et al.. Effect of topical steroids on corneal epithelial healing after vitreoretinal surgery. Acta Ophthalmol Scand 2006;84:319–322.
Keywords:© 2016 by Ophthalmic Communications Society, Inc.
pars plana vitrectomy; persistent corneal epithelial defect