Overall graft survival (Kaplan-Meier method) of 69 consecutive anterior lamellar keratoplasties (ALKs) and 69 penetrating keratoplasties (PKs). Graft survival was higher in the ALK group, but the difference did not reach statistical significance (log-rank test, χ² = 1.18; P = .19).
Confocal microscopy of the corneal epithelium (A, C, and E) and anterior stroma (B, D, and F) after anterior lamellar keratoplasty (ALK). A and B, Complete dissection (big-bubble technique). C and D, Noncomplete (manual) dissection. E and F, Automated dissection. No significant differences in keratocyte density in the anterior stroma were observed among the 3 subgroups of ALK.
Confocal microscopy of the corneal middle (A, C, and E) and posterior (B, D, and F) stroma after anterior lamellar keratoplasty. A and B, Complete dissection (big-bubble technique). C and D, Noncomplete (manual) dissection. E and F, Automated dissection. Myofibroblasts and bright elements can be seen in the posterior stroma after automated dissection (F). The highest keratocyte density was obtained after complete dissection, and the lowest was obtained after automated dissection.
Twelve-month spectacle-corrected visual acuity (logarithmic scale) after anterior lamellar keratoplasty according to corneal keratocyte density. Visual acuity increased with keratocyte density (Spearman ρ = 0.44; P = .002). Each dot may correspond to several eyes.
Borderie VM, Werthel A, Touzeau O, Allouch C, Boutboul S, Laroche L. Comparison of Techniques Used for Removing the Recipient Stroma in Anterior Lamellar Keratoplasty. Arch Ophthalmol. 2008;126(1):31-37. doi:10.1001/archophthalmol.2007.12
To compare 3 techniques used for removing the recipient stroma during anterior lamellar keratoplasty (ALK): the “big-bubble” technique, manual dissection using a crescent blade and slitlamp operating microscope, and microkeratome lamellar cut.
Retrospective comparative cohort study of 69 consecutive ALKs and 69 consecutive penetrating keratoplasties (PKs). Manifest refraction, slitlamp examination, Goldmann tonometry, ultrasound pachymetry, specular microscopy, and confocal microscopy findings were recorded.
The 12-month graft survival estimate was 98.5% in the ALK group and 94.1% in the PK group (P = .19). Higher endothelial cell density was found after ALK (P < .001). At 12 months (before suture removal), 53% of eyes that underwent ALK and 44% of eyes that underwent PK had 20/40 or better spectacle-corrected visual acuity (P = .24). In keratoconic eyes, these values were 83% and 69%, respectively (P = .18). Significant differences in visual acuity, corneal central thickness, and keratocyte density among ALK subgroups were found, with the best results obtained using the big-bubble technique and the worst results obtained using the microkeratome. In eyes that underwent ALK, visual acuity increased with keratocyte density.
Better results were obtained after ALK vs PK, and the big-bubble technique seemed to provide the best results.
A current trend among corneal surgeons toward replacing penetrating keratoplasty (PK) with various types of lamellar techniques that aim to remove only damaged tissue is widely shared. Deep anterior lamellar keratoplasty (ALK) is currently considered to be the first-choice surgical procedure in patients with corneal diseases not involving the endothelial layer, including keratoconus, stromal scars, and lattice dystrophy.1,2 Removal of the damaged corneal stroma may be achieved using various techniques, including manual dissection with a surgical blade, fluid injection, and microkeratome-assisted lamellar cut. Descemet membrane detachment from the corneal stroma can be achieved using air injection (“big-bubble” technique), hydrodelamination through a sclerocorneal flap, or sodium hyaluronate injection.3- 5 The objective of the present retrospective study was to compare 3 different techniques used for dissecting the recipient damaged stroma: the big-bubble technique, manual dissection using a crescent blade and slitlamp operating microscope, and microkeratome lamellar cut.
We retrospectively studied 69 consecutive ALK procedures performed by 1 of us (V.M.B.) in 69 patients with corneal diseases not (or moderately; no corneal edema on slitlamp examination) involving the corneal endothelium (ie, keratoconus, scars after infectious keratitis, stromal dystrophies, and trauma) between April 17, 2002, and November 30, 2005. For the control group, data from 69 consecutive PKs performed by the same surgeon in 69 patients with the same corneal diseases between November 4, 1999, and March 28, 2002, were recorded. All the PKs and ALKs were primary procedures. In accordance with French law, institutional review board and ethics committee approval were not required for this study because no modifications to French standards of treatment or follow-up were made.
All transplantations were performed at Centre Hospitalier National d’Ophtalmologie des XV-XX. Organ-cultured donor tissue was used in all cases (ALKs and PKs). Patients with superficial opacities and corneal thickness of more than 400 μm underwent automated lamellar keratoplasty (automated dissection subgroup) using a microkeratome (ALTK; Moria, Antony, France) with local anesthesia. Patients with keratoconus were not considered for automated lamellar keratoplasty because of low corneal thickness. The microkeratome was used to create a 250- or 350-μm-thick superficial dissection of the recipient central cornea. A donor lamellar graft was prepared using the microkeratome and an artificial anterior chamber. The graft was then trephined using the Hanna device (Moria). The donor trephination size was chosen according to the measured diameter obtained in the recipient cornea. Patients with deep stromal opacities or corneal thickness less than 400 μm underwent deep lamellar keratoplasty with general anesthesia. A first nonpenetrating trephination was performed using the Hanna device in the recipient cornea (trephination depth, 80% of corneal central thickness), and the anterior and middle stromas were removed using a crescent blade. Then air was injected into the posterior stroma. When air injection induced detachment of the Descemet membrane, sodium hyaluronate was injected between the posterior stroma and the Descemet membrane, and the remaining posterior stroma was completely removed using scissors (big-bubble technique, complete dissection subgroup).6,7 When no detachment of the Descemet membrane could be obtained after repeated air injection (60% of patients), deep posterior stromal dissection was performed using a crescent blade (noncomplete dissection subgroup).8 Dissection depth was controlled using a slitlamp operating microscope. Donor tissue used for deep lamellar keratoplasty was trephined from the posterior corneal surface using the Hanna device, and the Descemet membrane was removed.
An ALK was scheduled in 77 eyes during the study period. Significant perforation occurred during stromal dissection in 8 of these 77 eyes (10%), and ALK was converted to PK with no notable difficulties and normal subsequent outcomes (data not shown). Finally, 69 ALKs were performed: 12 (17%) using automated dissection (automated dissection subgroup), 34 (49%) using manual dissection (noncomplete dissection subgroup), and 23 (33%) using the big-bubble technique (complete dissection subgroup). The PKs were performed using the Hanna device as previously described.9 Graft suturing was performed according to standardized methods in all patients (ALK and PK), with mixed suturing (8 interrupted sutures and a 24-bit running suture) used in most patients.10 Sutures were removed after at least 18 months. All the patients were treated with topical dexamethasone phosphate (1 mg/mL) and neomycin sulfate (3400 IU/mL).11 No subconjunctival or systemic corticosteroids were used. This treatment was tapered for several months without standardization of postoperative corticosteroid management. The initial corticosteroid regimen was 1 drop hourly in patients with vascularized corneas and 1 drop every 6 hours in the other patients. Corticosteroid use was never stopped in the former patients, and it was discontinued when all the sutures were removed in the latter.
High-risk recipients were defined as having a vascularized cornea (≥2 quadrants of corneal vascularization). Patients were hospitalized up to the time of graft reepithelialization. They were then prospectively examined at 1 and 2 weeks and 1, 3, 6, 9, 12, 18, 24, 30, and 36 months after surgery. Manifest refraction (with spectacle correction), corneal central thickness (ultrasound pachymetry), intraocular pressure (mean of 2 measurements obtained along the steepest and flattest corneal meridians using the Goldmann applanation tonometer), and graft transparency were recorded at each examination. The criteria for graft failure were irreversible graft stromal edema and corneal opacification. During the second year after keratoplasty, the graft was evaluated using contact wide-field specular microscopy (Tomey, Nagoya, Japan) and confocal microscopy (Heidelberg Retina Tomograph II; Heidelberg Engineering, Dossenheim, Germany) as part of routine care. The PK patients were not examined using confocal microscopy because the device was not available in our institution at that time. For each cornea, 9 to 15 confocal images of the stroma (anterior, middle, and posterior) were used to calculate keratocyte nucleus density. Only bright and clear oval images were considered to be nuclei. Nuclei not fully included in the confocal image were counted only for 2 of the 4 sides of the image.
The mean (SD) follow-up of patients was 22.8 (11.1) months in the ALK group and 48.1 (21.3) months in the PK group. At 12 months postoperatively, 67 eyes (97%) were available for follow-up and 2 (3%) were lost to follow-up in the ALK group. These figures were 66 (96%) and 3 (4%), respectively, in the PK group.
For statistical analysis of visual acuity, astigmatism, corneal thickness, intraocular pressure, keratocyte density, and endothelial cell density, only eyes with clear grafts were included. Visual acuity was converted to log MAR units before statistical analysis. For quantitative variables, unpaired t tests were used to compare the ALK group with the PK group. Comparison of the subgroups, with sample sizes less than 30, was performed using Kruskal-Wallis analysis of variance and the Mann-Whitney test. Correlation was assessed using the Spearman regression test. Qualitative variables were analyzed using the χ2 test. Graft survival was assessed using the Kaplan-Meier method, and the log-rank test was used to compare the ALK group with the PK group. Statistical analysis was performed using a software program (Statistica version 6.1; StatSoft France, Maisons-Alfort).
Minimal perforation of the Descemet membrane occurred during surgery in 14 of 69 eyes (20%), and ALK could be performed in these eyes. Two graft failures were observed during the study (1 primary graft failure due to persistent double anterior chamber and 1 case of late bacterial keratitis). Other postoperative complications included reversible double anterior chamber (6 eyes), persistent epithelial defects (2 eyes), increased intraocular pressure (6 eyes), fixed dilated pupil (1 eye), interface hemorrhage (1 eye), cataract progression (5 eyes), lens posterior capsule opacification requiring laser capsulotomy (3 eyes), reversible stromal rejection (2 eyes), reversible epithelial rejection (1 eye), and herpetic keratitis (recurrence, 1 eye). In all the patients, slitlamp examination of the posterior stroma in the first 6 postoperative months showed fine folds in patients with complete dissection, thicker circumferential folds in patients with automated dissection, and intermediate folds in patients with noncomplete dissection. These folds resolved during the second year either spontaneously or after suture removal.
Confocal microscopy showed that the mean (SD) keratocyte density was higher in the anterior stroma (121  cells/mm2) than in the posterior stroma (82  cells/mm2; P < .001) and higher in the middle stroma (130  cells/mm2) than in the posterior stroma (P < .001), whereas the difference between the anterior stroma and the middle stroma was not significant (P = .08). Bright elements were seen in the posterior stroma near the interface in 67% of eyes (44/66). Activated keratocytes (myofibroblasts) were found in the corneal stroma in 33% of eyes (22/66).
Preoperatively, there were no significant differences between groups except that the mean (SD) age of donors in the PK group (70  years) was significantly higher than that of recipients in the ALK group (44  years; P < .001) (Table 1). The donor tissue mean (SD) endothelial density in the PK group (2328  cells/mm2) did not differ significantly from the recipient preoperative mean (SD) endothelial density in the ALK group (2367  cells/mm2; P = .54). At 12 months postoperatively, there were no significant differences in graft survival between groups (Figure 1); however, endothelial cell density was significantly higher in the ALK group, and mean refractive astigmatism (before suture removal) and mean intraocular pressure were significantly lower in the ALK group. Significantly fewer corticosteroids were used in the ALK group. At 12 months (sutures not removed), 53% of patients in the ALK group (35/66) and 44% in the PK group (27/61) had 20/40 or better spectacle-corrected visual acuity, and 6% of patients in the ALK group (4/66) and 15% in the PK group (9/61) had 20/200 or worse visual acuity (P = .24). In keratoconic eyes, 83% of patients in the ALK group (25/30) and 69% in the PK group (22/32) had 20/40 or better visual acuity, and none had 20/200 or worse in both groups (P = .18). No significant correlation between 12-month visual acuity and graft (or recipient) trephination size was observed.
Significant differences in visual acuity, corneal central thickness, and keratocyte density among subgroups were found, whereas no significant differences in intraocular pressure, refractive astigmatism, and postoperative endothelial cell density were observed (Figure 2, Figure 3, and Table 2). Visual acuity was also affected by preoperative diagnosis (P < .001). Conversely, no significant effect of the preoperative diagnosis was found on corneal central thickness (P = .69), intraocular pressure (P = .22), refractive astigmatism (P = .87), postoperative endothelial cell density (P = .18), or keratocyte density (P = .33). In patients with keratoconus, better visual acuity (P = .01) and higher keratocyte density (P = .006) were obtained with complete dissection than with noncomplete dissection. In patients with corneal scars after infectious keratitis, no significant differences in visual acuity were observed between dissection subgroups. In all the patients who underwent ALK, visual acuity significantly increased with stromal keratocyte density (Spearman ρ = 0.44; P = .002) (Figure 4).
In this retrospective study of patients with keratoconus, corneal scars after infectious keratitis, stromal dystrophies, and trauma, eyes that underwent ALK and eyes that underwent PK had similar 12-month visual acuity. Less astigmatism and higher endothelial cell densities at 12 months postoperatively were obtained in the ALK group. These figures are in accordance with several studies of patients with keratoconus.12- 15 Although not significant, higher graft survival was observed in the ALK group. Longer follow-up of patients is needed to determine whether the significant improvement in endothelial cell survival will improve subsequent end graft survival. If the long-term endothelial cell loss rate remains close to the normal loss rate related to aging, unlimited graft survival could be expected in at least some of these eyes.
In eyes that underwent ALK, we did not observe the extreme temporary thinning that occurs during the first year after PK.16,17 Conversely, corneal central thickness progressively decreased during the first year after ALK (data not shown).
The 3-dimensional keratocyte density of the central normal human cornea is difficult to assess accurately, and results depend on the technique used.18- 20 Mean 2-dimensional keratocyte density was reported to be 480 cells/mm2 in the normal human posterior corneal stroma and 601 cells/mm2 in the posterior stroma 15 months after PK.21,22 From these data, it seems likely that ALK induces keratocyte loss in the graft and in the remaining recipient corneal stroma. In fact, laser in situ keratomileusis has been shown to induce 37% cell loss in the stromal flap, 42% cell loss in the anterior retroablation zone, and 22% cell loss in the posterior stroma 5 years after surgery.23
In eyes that underwent ALK, complete removal of the recipient stroma gave the best visual results and the highest keratocyte densities. However, the big-bubble technique could not be achieved in many eyes (60%), such as those with keratoconus stage IV and those with deep stromal scars involving the posterior stroma and the Descemet membrane. In these eyes, manual dissection of the recipient corneal stroma using a crescent blade and the slitlamp operating microscope to control dissection depth permitted removal of most of the recipient stroma. In fact, the mean (SD) obtained corneal thickness (570  μm) was close to that of the normal cornea. This technique permitted preservation of the recipient endothelium, which seems valuable for young patients with keratoconus who may experience late endothelial failure in the long term and for patients with vascularized corneal scars who are at high risk for endothelial rejection. In a series24 of 127 keratoconic corneas, baring of the Descemet membrane during ALK was achieved in 37%. Patients with no remaining recipient posterior stroma had better visual acuity results. In another series25 of 13 keratoconic corneas, detachment of the Descemet membrane after air injection was achieved in 69%.
Automated lamellar keratoplasty has been used for grafting eyes with complications after laser in situ keratomileusis and photorefractive keratectomy, eyes with granular corneal dystrophy, and eyes with keratoconus.26- 28 The reported results were better than the present results using this technique. The distinction may originate from a difference in preoperative corneal diagnosis. From our data, it is not possible to determine whether automated dissection is actually a less preferred technique than deep manual dissection, even if lower visual acuity and lower keratocyte density were obtained with the former.
In the present series, the percentage of Descemet membrane perforation requiring conversion of ALK to PK was 10%. In the ALK group, at the end of surgery, microperforation had occurred in 20% of the patients. The reported percentage of Descemet membrane perforation during ALK is 9% to 39%.24,29- 33 The recipient preoperative corneal condition may affect the likelihood of Descemet membrane perforation during ALK. Very thin corneas and corneas with deep corneal scars are associated with a high risk of perforation.24,34 The present series includes 47% of eyes with scars after infectious keratitis or trauma and a high percentage of keratoconus stage IV.
A double anterior chamber occurred after ALK in 9% of patients (6/69), but in only 1 patient did this complication lead to graft failure. In the other 5 patients, it resolved either spontaneously or after graft resuturing combined with air injection in the anterior chamber, as previously reported by Tu et al.35 One main advantage of ALK compared with PK is the absence of endothelial rejection. However, stromal rejection and epithelial rejection are still possible after lamellar keratoplasty.36,37 Our results are in accordance with these studies.
In the present study, keratocyte density assessed during the second year after ALK significantly affected visual acuity; the higher the former, the higher the latter. In patients with deep ALK performed by means of manual dissection, the interface reflectivity assessed using confocal microscopy was shown to affect visual acuity recovery.38 Differences in visual acuity observed among the 3 subgroups of ALK patients may be at least partially explained by differences in keratocyte density. In rabbit eyes, laser in situ keratomileusis was shown to induce keratocyte apoptosis in the central corneal stroma located anterior and posterior to the lamellar cut.39,40 Keratocytes that die along the lamellar cut after laser in situ keratomileusis are replenished in 2 to 4 days with myofibroblastic cells. We also observed myofibroblastic keratocytes in 33% of ALK eyes close to the interface between the graft and the remaining recipient corneal stroma or the recipient Descemet membrane. Visual acuity after ALK for keratoconus was reported in several studies.7,15,24,25,30,32,41,42 A visual acuity of 20/40 was obtained in at least 80% of the patients in these studies, as in the present study.
Postoperative endothelial cell loss is moderate after ALK. One-year cell loss was reported to be 13% after ALK.43 Morris et al44 reported a mean endothelial cell density of 2417 cells/mm2 an average of 3 years after ALK (n = 20). Panda et al13 reported a mean endothelial cell density of 2220 cells/mm2 1 year after ALK. Sugita and Kondo33 reported a mean endothelial cell density of 1937 cells/mm2 2 years after ALK, which corresponded to a 13% postoperative endothelial cell loss. In the present ALK group, mean endothelial cell density assessed during the second postoperative year was more than 2000 cells/mm2. Mean postoperative endothelial cell loss was 9.3% in the 27 eyes with available preoperative endothelial cell counts.
In conclusion, deep ALK seems to be a successful technique in eyes with a normal or moderately damaged endothelium, and it gave better results than PK in the present nonrandomized retrospective study. When technically possible, the recipient stroma should be completely removed. When Descemet membrane detachment cannot be achieved, the surgical technique should aim to minimize the thickness of the remaining recipient stroma. For this purpose, the slitlamp operating microscope is a useful tool to appreciate stromal thickness during surgery. In patients not eligible for deep lamellar surgery, such as those with contraindications to general anesthesia, automated lamellar keratoplasty can be considered when the stromal opacities are superficial. Femtosecond laser–assisted ALK should be addressed in further studies.
Correspondence: Vincent M. Borderie, MD, PhD, Centre Hospitalier National d’Ophtalmologie des XV-XX, 28 rue de Charenton, 75012 Paris, France (firstname.lastname@example.org).
Submitted for Publication: March 7, 2007; accepted May 1, 2007.
Financial Disclosure: None reported.
Funding/Support: This study was supported by Université Pierre et Marie Curie-Paris6.