To report the clinical outcomes of penetrating keratoplasty (PK) after autologous cultivated limbal epithelial transplantation in eyes with limbal stem cell deficiency (LSCD) after ocular surface burns.
Retrospective case series.
This study included 47 patients with unilateral LSCD treated by autologous cultivated limbal epithelial transplantation and PK between 2001 and 2010. PK was performed either along with (single-stage; n = 12) or at least 6 weeks after (2-stage; n = 35) limbal transplantation. The primary outcome measure was corneal allograft survival, and failure was defined clinically as loss of central graft clarity. Secondary outcomes were postoperative Snellen visual acuity and complications.
Most patients were young (mean age, 18 ± 11.4 years) males (76.6%) with LSCD resulting from alkali burns (78.7%) and with visual acuity less than 20/200 (91.5%). The mean follow-up was 4.2 ± 1.9 years. Kaplan-Meier corneal allograft survival rate at 1 year was significantly greater in eyes undergoing 2-stage limbal and corneal transplantation (80 ± 6%; median survival, 4 years) compared with single-stage limbal and corneal transplantation (25 ± 13%; median survival, 6 months; P = .0003). Visual acuity of 20/40 or better was attained by 71.4% of eyes with clear corneal grafts. Allograft failure occurred in 26 (60.5%) eyes as a result of graft rejection (57.7%), graft infiltrate (26.9%), or persistent epithelial defects (15.4%). Recurrence of LSCD was more common after single-stage (58.3%) than 2-stage (14.3%) surgery ( P = .008).
The 2-stage approach of autologous cultivated limbal epithelial transplantation followed by PK successfully restores ocular surface stability and vision in eyes with chronic ocular burns. The single-stage approach is associated with poorer clinical outcomes and should be avoided.
Corneal transparency is an essential prerequisite for optimal vision. In severe ocular surface burns, corneal transparency is lost because of damage to the limbal stem cells, responsible for corneal epithelial renewal and repair, and also because of scarring and vascularization of the corneal stroma. Treatment of limbal stem cell deficiency (LSCD) is possible by direct or cultivated limbal transplantation from donors (allogeneic) or the healthy fellow eye (autologous), and the scarred cornea can be replaced by allogeneic corneal transplantation. Both procedures are necessary to optimize visual recovery, because the clarity of an allogeneic corneal graft is only temporary without a stable ocular surface. However, extensive corneal stromal vascularization predisposes these eyes to a higher risk of corneal allograft failure.
Cultivated limbal epithelial transplantation alone improves vision in only 16% to 54% of eyes with LSCD, and another 18% to 38% of all eyes need penetrating keratoplasty (PK) for visual restoration. Despite this, most large studies on autologous cultivated limbal epithelial transplantation have focused primarily on the outcomes of the limbal graft, and there is insufficient information in the literature regarding the long-term survival of corneal allografts. Similarly, there are no existing guidelines regarding the relative timing of the 2 transplantations and whether they can be combined as a single-stage procedure.
In 2 previous studies, we reported the encouraging early results of PK after autologous or allogeneic limbal transplantation. In this study, we report the corneal allograft survival, visual outcomes, and complications of single-stage and 2-stage autologous cultivated limbal and allogeneic corneal transplantation in a large series of eyes with unilateral LSCD.
At the L. V. Prasad Eye Institute, Hyderabad, India, between April 2001 and February 2010, autologous cultivated limbal epithelial transplantation was performed in 523 eyes of 512 patients clinically diagnosed to have LSCD based on the presence of corneal findings such as dull, irregular epithelial reflex, peripheral superficial neovascularization, conjunctivalization, persistent epithelial defect, and absence of limbal palisades of Vogt. The inclusion criteria for this study were eyes with LSCD that underwent autologous cultivated limbal epithelial transplantation and PK. Exclusion criteria were eyes with LSCD resulting from causes other than ocular surface burns and post-PK follow-up of less than 1 year. This study was approved by the institutional review board and was conducted in strict adherence to the tenets of the Declaration of Helsinki and Indian National Guidelines on Stem Cell Research and Therapy, with written informed consent from patients or from the guardians in case of minor patients.
The medical records of all patients included in this study were reviewed retrospectively for demographics and relevant clinical information ( Supplemental Table , available at AJO.com ). The corneal buttons obtained during PK were retrieved from the archives of the Ophthalmic Pathology Laboratory and the histopathologic results sections of formalin-fixed, paraffin embedded tissues stained with hematoxylin and eosin or periodic acid–Schiff were reviewed.
Patients underwent either single-stage or 2-stage limbal and corneal transplantation. Single-stage surgery was planned when severe corneal scarring was noted before surgery. Otherwise, PK was planned at least 6 weeks after limbal transplantation if the postoperative best spectacle-corrected visual acuity (BSCVA) was worse than 20/200 and was attributed to central corneal scarring.
Technique of Limbal Biopsy and Culture
A 2 × 2-mm biopsy sample was obtained from a healthy part of the limbus; the conjunctiva was excised just behind the pigmented line (palisades of Vogt) and was dissection into 1 mm of clear corneal stromal tissue at the limbus. The limbal tissue that contained epithelial cells and a part of the corneal stroma was excised. The tissue was transported to the laboratory in human corneal epithelium medium, which has been described previously. Briefly, it is composed of modified Eagle medium/F12 medium (1:1) solution containing 10% (vol/vol) autologous serum, 2 mM 1-glutamine, 100 U/mL penicillin, 100 μg/mL streptomycin, 2.5 μg/mL amphotericin B, 10 ng/mL human recombinant epidermal growth factor, and 5 μg/mL human recombinant insulin. In the laboratory under strict aseptic conditions, the donor limbal tissue was shredded into small pieces. These were explanted over the central 10 mm 2 of a 3 × 2.5-cm de-epithelialized and preserved human amniotic membrane. The cells were cultured using human corneal epithelium medium, the growth was monitored daily using an inverted phase contrast microscope, and the human corneal epithelium medium was changed every 2 days. The culture was maintained for 10 to 15 days, by which time a confluent monolayer of the limbal epithelial cells around the implanted tissues was achieved.
Technique of Limbal Transplantation
The lids were separated and any symblepharon was excised to allow for the insertion of the lid speculum. A peritomy was performed, and the vascular pannus was dissected from the corneal surface. The corneal clarity was assessed during surgery, and if severe central corneal scarring was noted, a PK was performed at this stage. Next, the human corneal epithelium medium with the monolayer of cultivated limbal epithelial cells was transferred onto the recipient’s ocular surface and secured with fibrin glue (TISSEEL Kit; Baxter AG, Vienna, Austria).
Technique of Penetrating Keratoplasty
The recipient cornea was excised using a disposable handheld trephine, with 0.5 mm of graft–host disparity. Lensectomy, anterior vitrectomy, and intraocular lens insertion were performed, depending on the clinical situation in each case. The graft was secured by 10-0 nylon interrupted sutures with knots buried on the donor side. At the end of the surgery, a subconjunctival injection of dexamethasone sodium phosphate (4 mg/mL) and gentamicin sulfate (20 mg/mL) was given. The excised fibrovascular pannus, recipient corneal button, or both were sent for histopathologic examination in 10% formaldehyde.
All patients underwent a comprehensive ophthalmic examination of both eyes at each follow-up visit. After limbal or corneal transplantation, or both, all patients were treated with 1% prednisolone acetate eye drops 8 times daily tapered to once daily in 5 to 6 weeks. Patients were prescribed 0.3% ciprofloxacin hydrochloride eye drops 4 times daily for 1 week or until the epithelial defect healed. The patients were seen on postoperative day 1, week 1, week 6, and at 3-month intervals thereafter for the first year. Any event of endothelial graft rejection was treated with an hourly instillation of 1% prednisolone acetate eye drops (tapered to the previous dosage in 1 month’s time) and a single dose of 500 mg intravenous methylprednisolone.
The primary outcome measure was corneal allograft survival; allograft failure was defined clinically as loss of central graft clarity for at least 3 months. Secondary outcome measures were BSCVA after PK and occurrence of complications. Recurrence of LSCD was diagnosed clinically based on the presence of corneal findings like peripheral superficial vascularization, conjunctivalization, and recurrent or persistent epithelial defects. Endothelial rejection was defined clinically as acute onset of anterior uveitis with keratic precipitates on the graft endothelium and overlying graft edema.
All statistical analysis was carried out using MedCalc software version 11.0.4 (MedCalc Software, Mariakerke, Belgium). Means (with standard deviations) were reported for all parametric data and medians (with ranges) were reported for nonparametric data. Kaplan-Meier survival analysis was performed to estimate the probability (reported as percentage with standard error) of corneal allograft survival. Univariate (log-rank test) analysis was carried out to identify the variables associated with graft survival, which subsequently were fitted in a multivariate model (Cox proportional hazard regression with stepwise elimination) to estimate their relative effect on graft survival (after checking all the assumptions and interactions of the model). Baseline features and recurrence of LSCD were compared between eyes undergoing simultaneous and sequential surgery by using the Fisher exact test. A P value of less than .05 was considered to be statistically significant.
Demographic and Baseline Data
This study included 47 eyes of 47 patients who underwent autologous cultivated limbal epithelial transplantation and PK for unilateral total LSCD. Of them, 12 eyes underwent single-stage limbal and corneal transplantation and 35 eyes underwent 2-stage limbal and corneal transplantation. Table 1 shows that eyes in both groups were comparable at baseline. Most patients were young (mean age, 18 ± 11.4 years; range, 3 to 47 years) males (76.6%) with LSCD resulting from alkali burns (78.7%) and vision less than 20/200 (91.5%). The mean follow-up after PK was 4.2 ± 1.9 years (range, 1 to 8 years). The duration between injury and PK ranged from 4.5 months to 12 years (median, 1 year). In 35 eyes having 2-stage surgery, the duration between limbal and corneal transplantation ranged from 6 weeks to 4 years (median, 6 months). At the time of PK, there was no clinically evident LSCD in the fellow eyes of any patients.
|Characteristics||Single-Stage Procedure (n = 12)||2-Stage Procedure (n = 35)||P Value|
|Age (y) at penetrating keratoplasty, no. (%)|
|≤ 8||4 (33.3)||9 (25.7)||.51|
|8 to 16||3 (25)||3 (8.6)|
|> 16||5 (41.7)||23 (65.7)|
|Cause of limbal stem cell deficiency, no. (%)|
|Alkali||10 (83.4)||26 (74.2)||.82|
|Acid||1 (8.3)||8 (22.9)|
|Thermal||1 (8.3)||1 (2.9)|
|Previous ocular procedures, no. (%) a|
|None||6 (50)||21 (60)||.58|
|AMG||3 (25)||11 (31.4)|
|LK||3 (25)||1 (2.9)|
|SR + CAG||2 (16.6)||3 (8.7)|
|Entropion correction||1 (8.3)||2 (5.8)|
|Duration (mos) between injury and penetrating keratoplasty, no. (%)|
|≤ 6||2 (16.6)||13 (37.1)||.33|
|> 6||10 (83.4)||22 (62.9)|
|Best-corrected visual acuity, no. (%)|
|≥ 20/200||1 (8.3)||3 (8.7)||.56|
|< 20/200||11 (91.7)||32 (91.3)|
|Extent (clock hrs) of limbal involvement, no. (%)|
|≤ 9||0||1 (2.9)||.57|
|> 9||12 (100)||34 (97.1)|
|Symblepharon, no. (%)|
|Absent||3 (25)||10 (28.6)||.89|
|Present||9 (75)||25 (71.4)|
|Donor trephine size (mm), no. (%)|
|≤ 8||3 (25)||2 (5.8)||.18|
|> 8||9 (75)||33 (94.2)|
|Endothelial cell density of donor cornea (cells/mm 2 ), no. (%)|
|2250 to 2750||8 (66.7)||24 (68.5)||.81|
|> 2750||4 (33.3)||11 (31.5)|
|Additional surgery during penetrating keratoplasty (%)|
|None||10 (83.3)||30 (85.7)||.78|
|ECCE + PCIOL||2 (16.7)||5 (14.3)|
The overall corneal allograft survival rate at 1 year was 66 ± 7%, with a median survival of 2.5 years ( Figure 1 ). Although the 2 groups were comparable at baseline ( Table 1 ), the corneal allograft survival rate at 1 year in eyes undergoing 2-stage surgery was 80 ± 6% (median survival, 4 years) compared with 25 ± 13% (median survival, 6 months) in eyes undergoing single-stage surgery ( P = .0003; Figure 1 , Bottom). Both univariate and multivariate analysis ( Table 2 ) showed that single-stage surgery carried a significantly higher risk of corneal allograft failure (hazard ratio, 5.2; P = .0009) compared with 2-stage surgery. Among eyes with 2-stage surgery, timing of PK (6 weeks to 6 months or after 6 months) did not affect corneal allograft survival ( P = .3).
|Factor||No.||Failure||Survival Probability (%) at 1 y ± SE||Univariate P Value||Hazard Ratio (95% CI)||Multivariate P Value|
|Age at penetrating keratoplasty (y)|
|< 16||19||9||57.9 ± 0.11||.45|
|≥ 16||28||17||71.4 ± 0.09||1.2 (0.4 to 3.4)||.79|
|Cause of limbal stem cell deficiency|
|Alkali||36||18||63.9 ± 0.08||.16|
|Others||11||8||72.7 ± 0.15||2.9 (0.94 to 8.2)||.052|
|Previous corneal surgery|
|None||29||15||62.1 ± 0.09||.85|
|≥ 1||18||11||72.2 ± 0.11||0.8 (0.34 to 2)||.84|
|Duration between CLET and PK (mos)|
|> 6||19||9||78.9 ± 0.09||.0026|
|1.2 to 6||16||7||81.2 ± 0.09||0.6 (0.18 to 1.7)||.3|
|Single-stage||12||10||25 ± 0.13||5.2 (1.7 to 15.2)||.0032|
|Endothelial graft rejection|
|Absent||28||12||67.9 ± 0.08||.09|
|Present||19||14||63.2 ± 0.11||2 (0.8 to 5.3)||.15|
|Recurrence of LSCD after CLET and PK|
|Absent||35||21||65.7 ± 0.08||.49|
|Present||12||5||66.7 ± 0.14||0.5 (0.2 to 1.3)||.16|
|Postoperative suture-related problems (loose sutures and/or infiltrates)|
|Absent||35||18||71.4 ± 0.07||.12|
|Present||12||8||50 ± 0.14||1.7 (0.6 to 4.7)||.3|
The visual outcomes after PK are summarized in Table 3 . Among the 21 eyes that maintained a clear graft at last follow-up, 15 (71.4%) eyes had BSCVA of 20/40 or better, whereas 6 (28.6%) eyes had BSCVA of less than 20/40. Of these 6 eyes, 4 (66.6%) eyes were amblyopic, whereas 2 (33.3%) eyes had posterior capsular opacification ( Figure 2 ).
|Characteristic||1 y||2 y||3 y||4 y||5 y|
|≥ 20/40||17 (36.2)||14 (31.7)||10 (29.4)||8 (33.3)||4 (25)|
|20/50 to 20/200||7 (14.9)||5 (12.2)||5 (14.7)||4 (16.7)||3 (18.8)|
|< 20/200||23 (48.9)||23 (56.1)||19 (55.9)||12 (50)||9 (56.2)|
|Cause of BCVA less than 20/40|
|Graft edema||11 (36.7)||13 (46.4)||13 (54.2)||9 (56.2)||7 (58.3)|
|Infiltrate/scar||7 (23.3)||6 (21.4)||6 (25)||3 (18.8)||2 (16.7)|
|Amblyopia||5 (16.7)||4 (14.3)||4 (16.7)||4 (25)||3 (25)|
|Astigmatism > 6 D||3 (10)||1 (3.6)||0 (0)||0 (0)||0 (0)|
|LSCD ± PED||4 (13.3)||3 (10.7)||1 (4.2)||0 (0)||0 (0)|
|Cataract/PCO||0 (0)||1 (3.6)||0 (0)||0 (0)||0 (0)|
Graft Failure and Complications
In 26 (55.3%) of the 47 eyes, the corneal allografts failed over the study period at various time points. Of these, 15 (57.7%) failed after 1 or more episodes of endothelial rejection ( Figure 2 ), 7 (26.9%) failed as a result of central graft infiltrates ( Figure 2 ), and 4 (15.4%) failed because of recurrence of LSCD with central persistent epitheleal defects. Twelve eyes (25.5%) had a recurrence of LSCD (8 partial, 4 total; Figure 2 ). Recurrence of LSCD was more common after single-stage surgery (7/12; 58.3%) than after 2-stage surgery (5/35; 14.3%; P = .008). Three eyes underwent repeat limbal transplantation for recurrence of LSCD, whereas 2 eyes underwent emergency therapeutic PK for graft infiltrate and melt. All 7 graft infiltrates were loose-suture related, occurred within 1 year of PK, and were caused by coagulase-negative Staphylococci in 4 eyes and α-hemolytic Streptococci in 3 eyes ( Figure 2 ). Glaucoma occurring after PK in transplanted eyes or iatrogenic LSCD occurring after PK in donor eyes was not noted in the follow-up period.
On histopathologic analysis, corneal buttons obtained during PK from 35 eyes undergoing 2-stage surgery showed a normal corneal epithelium of 3 to 5 layers. Corneal buttons from 12 eyes undergoing single-stage surgery showed presence of a superficial fibrovascular pannus with areas of epithelial denudation. The Bowman membrane was absent in all corneas, and the stroma showed thinning, irregular scarring involving 30% to 90% of the stromal depth and peripheral vascularization. The Descemet membrane and endothelium appeared to be normal in all specimens.