Retrospective Analysis of a New Intrastromal Dissection Technique Using the Retinal Reflex for Deep Anterior Lamellar Keratoplasty

Abstract

Purpose:

The purpose of this study was to describe a new surgical technique for deep anterior lamellar keratoplasty.

Methods:

All pupils in the recipient eyes were dilated preoperatively. Vertical grooving was performed using a crescent blade with a width of 5 mm and a depth of one-third to half corneal thickness on the temporal side of the limbus. Stromal dissection was performed as close as possible to Descemet membrane by observing the gap between the gold line by retinal reflex and the front edge of the crescent blade. Lamellar dissection was performed along the lamellar plane using corneal dissectors. The ophthalmic viscoelastic device was injected into the intrastromal pocket to separate the anterior and posterior stroma and an anterior corneal lamella was excised. A donor cornea was sutured into the recipient bed.

Results:

In 18 eyes, none of the patients had Descemet membrane rupture during surgery. The mean postoperative residual stromal thickness was 80 ± 31 μm. The mean central corneal thickness after surgery was 660 ± 69 μm. At the last follow-up, the cornea was cleared in all 18 eyes on slit-lamp examination.

Conclusions:

We estimated the residual stromal thickness based on the gap between the gold line by the retinal reflex and crescent blade, and intrastromal lamellar dissection was performed using a smooth corneal dissector. Consequently, the surface of stromal dissection was smooth, and the residual stromal thickness was even.

Deep anterior lamellar keratoplasty (DALK) is a partial-thickness corneal transplantation that involves selective transplantation of the corneal stroma, leaving the recipient's Descemet membrane and endothelium. Because it preserves the recipient's own corneal endothelium, there is no endothelial rejection compared with penetrating keratoplasty; therefore, it has the advantage of a high long-term survival rate.^1,2 There is a disadvantage that the degree of improvement of visual acuity may be lower than that of penetrating keratoplasty because of interface haze.^3–6 However, recently, many articles have reported that the degree of improvement of visual acuity is almost equivalent to that of penetrating keratoplasty.^7–14 Compared with penetrating keratoplasty, the surgical procedure is difficult and takes a long time, and there is a risk of Descemet membrane rupture because it is difficult to estimate the remaining stromal thickness under a surgical microscope during stromal dissection.²

There are several ways to remove the stroma, leaving only Descemet membrane in DALK. First, manual dissection is a traditional method performed layer by layer using a blade¹⁵; yet, the operation takes a long time, and postoperative visual acuity may be slightly poor because the dissection surface is irregular. Recently, the big-bubble technique by Anwar has been widely used.^16,17 It induces separation of the stroma from the pre-Descemet membrane Dua layer by injecting air into the deep stroma. This technique was able to achieve consistently successful Descemet membrane separation in a shorter time than manual dissection in patients with keratoconus.¹⁶ However, there is a disadvantage in that a large bubble is not always successful, and Descemet membrane can be ruptured while making the big bubble. A method using a femtosecond laser has been introduced.^18,19 This technique is used to create a laser-created vertical or lamellar incisions in recipient and donor corneas to facilitate DALK, although this method has the disadvantage of requiring an expensive femtosecond laser machine. Another method is the Melles technique,²⁰ wherein the aqueous humor in the anterior chamber is exchanged with air to visualize the posterior corneal surface (air-to-endothelium interface). Through a 5.0-mm scleral incision, a deep stromal pocket was created across the cornea, using the air-to-endothelium interface as a reference plane for dissection depth. Because stromal dissection uses the natural lamellar structure of the stroma, the surface of the stromal bed is smooth.

Recently, we found that we can estimate the thickness of the remaining stroma during corneal incision in cataract surgery using the retinal reflex (Figs. 1A, B). In this article, we introduce an easier and more stable intrastromal dissection technique in the recipient cornea by visually estimating the remaining stromal thickness using the retinal reflex during DALK.

FIGURE 1.

Gold line around the crescent blade by the retinal reflex during cataract surgery. The gap (two arrows) between the gold line and crescent blade represents the residual stromal thickness. A, Wide gap means shallow dissection and thick remaining stroma. B, Narrow gap means deep dissection and thin remaining stroma.

PATIENTS AND METHODS

Ethics Statements

This study complied with the Declaration of Helsinki and received approval from Yeouido St. Mary's Hospital Institutional Review Board (SC21RISI0058). Informed consent was waived by IRB for this retrospective analysis.

Study Design and Population

This study was a retrospective medical record analysis of 18 eyes (18 patients). Patients who underwent DALK from June 2019 to April 2023 using retinal reflex for intrastromal dissection in the recipient cornea were included in the study. As exclusion criteria, patients with keratoconus who had a scar after hydrops and patients with deep stromal opacity for whom manual dissection from the beginning is necessary were excluded.

Before surgery, best-corrected visual acuity was measured as the logarithm of the minimum angle of resolution (logMAR). Counting fingers was defined as 1.90 logMAR, and hand movement was defined as 2.30 logMAR, according to previous literature.²¹ The refractive error, corneal power, and corneal astigmatism were measured using an auto refractometer (KR-1; Topcon, Tokyo, Japan). Central corneal thickness was measured using anterior segment optical coherence tomography (OCT) (DRI OCT Triton; Topcon), and endothelial cell density was measured using a specular microscope (SP-8000; KONAN, INC, Hyogo, Japan).

Surgical Technique

All surgeries were performed with patients under general anesthesia.

Recipient Cornea

All pupils in the recipient eyes were dilated preoperatively. The circle to be trephinated was marked on the cornea with a marking pen. Vertical grooving was performed using a crescent blade with a width of 5 mm and a depth of one-third to half corneal thickness on the temporal side of the limbus (Fig. 2A). By holding 1 side of the corneal incision site with Colibri forceps, stromal dissection to Descemet membrane was performed with the same crescent blade (Fig. 2B). At this time, we could observe the gold line around the crescent blade because of the retinal reflex. The gap between the gold line and crescent blade represents the residual stromal thickness. Stromal dissection was performed as close as possible to Descemet membrane by observing the gap between the gold line around the crescent blade and the front edge of the crescent blade (Fig. 2C). A small amount of ophthalmic viscoelastic device (OVD; hyaluronic acid 1%; Hyalu Inj., Hanmi Pharm. Co., Seoul, Korea) was injected into the stromal pocket for lubrication, and lamellar dissection was performed along the lamellar plane to the pupil using a straight corneal dissector (K2-3660; Katena, Parsippany, NJ). The remaining area was dissected using a curved corneal sector (Fig. 2D). Dissection was performed 1 to 2 mm outside the trephination mark (Fig. 2E). Marking of recipient's cornea was performed with an 8-prong radial marker (Fig. 2F). After stromal dissection, a 1-bite suture was placed at the inlet of the stromal pocket. Then, the OVD (hyaluronic acid 1%) (Hyalu Inj., Hanmi Pharm. Co.) was injected into the intrastromal pocket to separate the anterior and posterior stroma (Fig. 2G). Trephination was performed slowly using a Barron radial vacuum trephine (Katena; diameter: 7.00–7.50 mm) until the OVD emerged (Fig. 2H). The anterior stroma was cut using Vannas scissors (E3218-R; Bausch Lomb, Bridgewater, NJ; Figs. 2I, J). After removing all the anterior stroma, the thickness of the residual stroma was assessed using a portable slit beam (Eidolon Optival, LLC, Natick, MA). In addition, manual dissection was performed using a crescent blade if the remaining stroma was too thick or if there was significant opacity in front of the pupil that could affect vision. The stromal surface of the recipient bed was sufficiently irrigated with a balanced salt solution to remove all the remaining OVD.

FIGURE 2.

Procedures of intrastromal dissection using retinal reflex during DALK. A, Vertical grooving was performed using a crescent blade with a width of 5 mm and a depth of one-third to half corneal thickness on the temporal side of the limbus. B, By holding 1 side of the corneal incision site with Colibri forceps, stromal dissection of Descemet membrane was performed with the same crescent blade. At this time, we could observe the gold line around the crescent blade because of the retinal reflex. The gap between the gold line and crescent blade represents the residual stromal thickness. C, Stromal dissection was performed as close as possible to Descemet membrane by observing the gap between the gold line around the crescent blade and the front edge of the crescent blade. D, Lamellar dissection was performed along the lamellar plane to the pupil using a straight corneal dissector (K2-3660; Katena). The remaining area was dissected using a curved corneal sector. E, Dissection was performed 1 to 2 mm outside the trephination mark. F, Marking of host cornea was performed with an 8-prong radial marker. G, OVD was injected into the intrastromal pocket to separate the anterior and posterior stroma. H, Trephination was performed slowly using a Barron radial vacuum trephine (Katena) until the OVD emerged. I,J, Anterior stroma was cut using Vannas scissors (E3218-R; Bausch Lomb). K, After removing Descemet membrane from the donor cornea, the donor cornea was placed on the recipient bed and 16 intermittent sutures were placed with 90% stromal-depth radial bites using 10-0 nylon.

Donor Cornea

The donor cornea was placed endothelial side up on a Barron vacuum punch and then punched 0.25 mm larger than the donor cornea or the same size. After removing Descemet membrane from the donor cornea, the donor cornea was placed on the recipient bed and 16 intermittent sutures were placed with 90% stromal-depth radial bites using 10-0 nylon (Fig. 2K). To minimize astigmatism, the suture was adjusted by referring to the light-emitting diode ring of the ophthalmic surgical microscope (OPMI Lumera 700; Carl Zeiss, Oberkochen, Germany) reflected on the surface of the donor cornea. Figure 3 is a schematic diagram of these procedures. Video 1 shows the procedures of intrastromal dissection using retinal reflex during DALK.

FIGURE 3.

Schematic diagram of procedures of new intrastromal dissection using retinal reflex during DALK. A, Stromal dissection was performed as close as possible to Descemet membrane by observing the gap between the gold line around the crescent blade and the front edge of the crescent blade. B, Lamellar dissection was performed along the lamellar plane to the pupil using a straight corneal dissector. The remaining area was dissected using a curved corneal sector. C, OVD was injected into the intrastromal pocket to separate the anterior and posterior stroma. D, Trephination was performed slowly using a Barron radial vacuum trephine until the OVD emerged. E, Anterior stroma was cut using Vannas scissors. F, Donor cornea was placed on the recipient bed and 16 intermittent sutures were placed with 90% stromal-depth radial bites using 10-0 nylon.

Video 1 Procedures of intrastromal dissection using retinal reflex during deep anterior lamellar keratoplasty. Vertical grooving was performed using a crescent blade with a width of 5 mm and a depth of one-third to half corneal thickness on the temporal side of the limbus. By holding one side of the corneal incision site with Colibri forceps, stromal dissection of the Descemet membrane was performed with the same crescent blade. At this time, we could observe the gold line around the crescent blade because of the retinal reflex. The gap between the gold line and crescent blade represents the residual stromal thickness. Stromal dissection was performed as close as possible to Descemet membrane by observing the gap between the gold line around the crescent blade and front edge of the crescent blade. Lamellar dissection was performed along the lamellar plane to the pupil using a straight corneal dissector (K2-3660; Katena). The remaining area was dissected using a curved corneal sector. Dissection was performed 1–2 mm outside the trephination mark. Marking of host cornea was performed with an 8-prong radial marker. The ophthalmic viscoelastic device (OVD) was injected into the intrastromal pocket to separate the anterior and posterior stroma. Trephination was performed slowly using a Barron radial vacuum trephine (Katena) until the OVD emerged. The anterior stroma was cut using Vannas scissors (E3218-R; Bausch Lomb). After removing Descemet membrane from the donor cornea, the donor cornea was placed on the recipient bed and 16 intermittent sutures were placed with 90% stromal-depth radial bites using 10-0 nylon.

Follow-Up

Postoperatively, the patient's eye was treated with moxifloxacin (Vigamox eye drops 0.5%; Novartis AG, Basel, Switzerland) and prednisolone (Predbell 1% eye drops; Chong Kun Dang, Seoul, Republic of Korea) 4 times per day. We reviewed the recorded videos of the surgery and determined the durations of intrastromal dissection and total surgery. Best-corrected visual acuity, refractive error, corneal power, corneal astigmatism, total central corneal thickness, residual stromal thickness, and endothelial cell density were measured during regular follow-up visits at the outpatient clinic. During the last follow-up, corneal conditions and complications were investigated using a slit-lamp microscope.

RESULTS

Demographic and Clinical Characteristics

Eighteen patients were included in this study: 7 men and 11 women. Their ages ranged from 26 to 77 years, with an average age of 56.4 years (Table 1). The preoperative diagnoses were stromal scar in 11 eyes, lattice corneal dystrophy in 4 eyes, keratoconus in 2 eye, and lipid keratopathy in 1 eye. The mean intrastromal dissection time was 8.9 ± 2.6 minutes, and the mean total operative time was 86.6 ± 19.8 minutes. After intrastromal dissection, additional stromal dissection was not required in 12 eyes, but in 6 eyes, significant opacity remained in the pupil area after removing the anterior stroma; therefore, manual dissection was additionally performed using a crescent blade in that area. None of the patients had Descemet membrane rupture during surgery. There were no cases of postoperative endophthalmitis.

TABLE 1.

Demographics and Clinical Outcomes of the Patients Included in the Study

Patient No.	Age (yr)	Preoperative Diagnosis	Preop			Stromal Dissection Time (min)	Total OP Time (min)	Additional Manual Dissection	Comorbidity	FU (mo)	Postop
			BCVA (Log MAR)	CCT (μm)	ECD (Cells/mm2)						BCVA (Log Mar)	SE (D)	K (D) (Flat/Steep)	CA (D)	CCT (μm)	Residual Stromal Thickness (μm)	ECD (Cells/mm2)	Additional Surgeries	Final Corneal Status
1	62	Lattice corneal dystrophy	1.90	627	1474	8	117	No	None	48	0.70	−6.50	43.75/44.50	0.75	663	115	2313	Total stitch out	Clear
2	43	Keratoconus	1.90	526	2688	9	70	No	None	28	0.40	−7.25	46.75/48.25	1.50	630	95	2573	Total stitch out	Clear
3	77	Stromal scar	1.90	535	1472	9	94	Yes	Amblyopia	31	2.3	+6.50	44.75/47.00	2.25	865	89	1243	Total stitch out	Clear
4	32	Lattice corneal dystrophy	0.80	565	2327	6	70	No	None	10	0.22	−5.50	40.50/43.50	3.00	719	132	2808	Partial stitch out	Clear
5	62	Stromal scar	2.30	604	1890	5	67	No	None	28	1.30	−5.00	43.25/48.50	5.25	632	91	1052	Total stitch out, cataract surgery	Clear
6	59	Stromal scar	1.00	499	2924	7	89	No	None	30	0.30	−0.25	43.25/45.50	2.25	633	73	2450	Total stitch out, cataract surgery	Clear
7	49	Stromal scar	1.22	616	1044	7	92	Yes	None	24	0.70	−2.50	45.25/52.25	7.00	623	68	876	Total stitch out	Clear
8	63	Stromal scar	0.70	625	1158	6	105	Yes	None	26	0.52	−0.75	42.75/51.00	8.25	605	27	694	Total stitch out, cataract surgery	Clear
9	63	Stromal scar	1.00	502	3424	13	73	Yes	None	23	0.70	−2.50	42.75/47.50	4.75	599	29	3595	Total stitch out	Clear
10	60	Stromal scar	1.00	397	4696	10	72	No	None	22	0.40	+0.50	42.25/45.75	3.50	697	80	2316	Total stitch out	Clear
11	67	Stromal scar	0.70	432	3095	14	90	No	ERM with ME	14	0.70	1.50	42.50/46.00	3.50	676	81	2576	Total stitch out, cataract surgery	Clear
12	68	Stromal scar	1.90	294	2631	14	81	No	None	12	0.52	−1.00	46.00/48.25	2.25	584	65	1785	Total stitch out, cataract surgery	Clear
13	66	Stromal scar	1.90	471	1715	8	111	Yes	Myopic degeneration	12	1.90	−7.00	45.75/52.75	7.00	653	44		Total stitch out, intravitreal injection	Clear
14	61	Stromal scar	1.30	590	2994	8	83	No	RD, PVR	15	1.30	+13.50	38.50/43.25	4.75	576	94	2702	Total stitch out, cataract surgery, pars plana vitrectomy for RD	Clear
15	48	Lattice corneal dystrophy	1.30	566	2070	9	77	No	None	11	0.15	−1.0	42.25/45.25	3.00	757	105	2908	Total stitch out	Clear
16	53	Lipid keratopathy	1.52	576	2120	10	135	Yes	None	9	0.82	−4.75	44.25/53.75	9.50	656	40	1620	Partial stitch out	Clear
17	57	Lattice corneal dystrophy	1.90	651	1632	9	73	No	None	5	0.82	−2.00	44.25/46.25	2.00	632	87	2918		Clear
18	26	Keratoconus	1.16	423	2838	9	60	No	None	0.3	1.30	−7.75	48.00/56.00	8.00	694	130			Clear
Mean ± SD	56.4 ±12.8		1.41 ±0.50	528 ±95	2344 ± 918	8.9 ± 2.6	86.6 ±19.8			19.4 ±11.6	0.64 ±0.34*	−1.8 ±5.3		4.36 ± 2.60	660 ±69	80 ± 31	2152 ±844

*Mean and SD of BCVA except amblyopic eye (patient number 3), myopic eye with degeneration of retina (patient number 13), and PVR eye (patient number 14).

CA, corneal astigmatism; CCT, central corneal thickness; D, diopter; ECD, endothelial cell density; ERM, epiretinal membrane; FU, follow-up; K, keratometry; ME, macular edema; OP, operation; PVR, proliferative vitreoretinopathy; RD, retinal detachment; SE, spherical equivalent.

Outcomes

The mean postoperative follow-up period was 19.4 months (Table 1). The mean best-corrected visual acuity was 0.64 ± 0.34 (range: 0.15–1.30) logMAR of the remaining 15 patients except amblyopic eye(patient number 3), myopic eye with degeneration of retina (patient number 13), and proliferative vitreoretinopathy eye (patient number 14). The mean central corneal thickness after surgery was 660 ± 69 μm. The mean postoperative residual stromal thickness was 80 ± 31 μm. The mean endothelial cell density values were 2344 cells/mm² preoperatively and 2152 cells/mm² postoperatively. At the last follow-up, the cornea was cleared in all 18 eyes on slit-lamp examination.

Representative Cases

A 43-year-old patient (patient number 3) was diagnosed with keratoconus in OU and underwent DALK using this technique in the left eye. The preoperative best-corrected visual acuity was counting fingers (1.90 logMAR). On slit-lamp examination before surgery, no hydrops or hydrops scars were found in the cornea. The refractive error before surgery was −3.50 Dsph: −4.25 Dcyl × Axis 170 degrees. Preoperatively, the endothelial cell density was 2688 cells/mm², and the central corneal thickness was 526 μm. The recipient trephine size was 7.5 mm. The total operative time was approximately 70 minutes, and stromal dissection took 9 minutes. No complications, such as Descemet membrane perforation, occurred intraoperatively. The best-corrected visual acuity after surgery was 0.22 logMAR. On slit-lamp microscopy, the cornea was clear and no interface haze was observed (Fig. 4A). The refractive error after surgery was −4.75 Dsph: −3.00 Dcyl × Axis 10 degrees. Postoperatively, the endothelial cell density was 2881 cells/mm² (Fig. 4B), central corneal thickness was 518 μm, and residual stromal thickness on anterior segment OCT was 84 μm. The interface between the recipient and donor corneas was very smooth (Fig. 4C). At 11 months postoperatively, the cornea remained clear.

FIGURE 4.

Representative case. A 43-year-old patient (patient number 2) was diagnosed with keratoconus in OU and underwent DALK using this technique in the left eye. The best-corrected visual acuity after surgery was 0.22 logMAR. A, On slit-lamp microscopy, the cornea was clear and no interface haze was observed. B, Postoperatively, the endothelial cell density was 2881 cells/mm². C, Central corneal thickness was 518 μm, and residual stromal thickness on anterior segment OCT was 84 μm. The interface between the recipient and donor corneas was very smooth.

A 66-year-old patient (patient number 13) was diagnosed with stromal opacity in his right eye and underwent DALK using this technique (Fig. 5). In the clear periphery, stromal dissection to Descemet membrane was performed with a crescent blade. We could observe the gold line around the crescent blade because of the retinal reflex.

FIGURE 5.

Representative case. A 66-year-old patient (patient number 13) was diagnosed with stromal opacity in his right eye and underwent DALK using this technique. A dark arrow indicates a gold line around the crescent blade.

Supplemental Figure, Supplemental Digital Content 1, http://links.lww.com/ICO/B616 shows 2 anterior segment OCT images of the corneas of 2 other patients (patient numbers 6 and 7) who underwent DALK using the intrastromal dissection technique. The residual stromal thicknesses were 63 μm in Supplemental Figure, Supplemental Digital Content 1, http://links.lww.com/ICO/B616 and 60 μm in Supplemental Figure, Supplemental Digital Content 1, http://links.lww.com/ICO/B616. In both cases, the residual stromal thickness was even and the donor–recipient interface was smooth.

DISCUSSION

In DALK, it is important to leave the recipient stroma as thin as possible. When using a general ophthalmic surgical microscope, it is difficult to estimate the depth of stromal dissection in the recipient cornea. If the operating microscope had built-in OCT, it would be possible to estimate the remaining stromal thickness, but not all hospitals have this type of expensive equipment. Therefore, in the Melles technique, after filling the anterior chamber with air, the residual stromal thickness was estimated by the total reflection occurring at the air–cornea interface.²⁰ Unlike the Melles technique, we estimated the residual stromal thickness based on the gap between the gold line by the retinal reflex and crescent blade.

In 2015, Scorcia et al²² reported a red reflex-guided big-bubble DALK. They judged the dissection depth using lines seen in the red reflex of the microscope light source as a reference to guide the insertion of a blunt probe for big-bubble formation. In their technique, starting from the base of corneal trephination, a blunt triangled spatula is inserted into the stroma, creating a curved line ahead of the tip edge. The tip of the instrument is advanced centripetally and tilted downward until the red zone between the spatula tip and the dark line disappears. A blunt cannula is inserted into the same track and air is injected into the stroma until formation of the big bubble is completed. However, we used red reflex to judge the dissection depth to guide the placement of a corneal dissector for intrastromal dissection. In our technique, stromal dissection was performed as close as possible to Descemet membrane by observing the gap between the gold line around the crescent blade and front edge of the crescent blade. And lamellar dissection was performed along the lamellar plane using a corneal dissector. Trephination was performed, and the anterior stroma was cut using Vannas scissors.

Descemet membrane perforation is a common and serious complication that can occur during DALK, occurring in 11.7% of cases.² The incidence of microperforation of Descemet membrane during DALK using the big-bubble technique is reported to be 9% to 23%.^16,23–27 The incidence of microperforation of Descemet membrane using the Melles technique has also been reported to be 12%.²⁰ However, there was no perforation in 18 eyes when using our technique.

Endothelial cell density was not significantly reduced after DALK using our method. In some patients, the endothelial cell density after surgery was found to increase compared with before surgery, but this is believed to be because the endothelial cell density before surgery was inaccurate.

The first advantage of our technique over the Melles technique is that it does not inject air into the anterior chamber. In the Melles technique, endothelial cell damage is possible because of the air in the anterior chamber. As our technique does not fill the anterior chamber with air, there is no risk of endothelial cell damage due to air. In this study, the mean preoperative endothelial cell count did not decrease postoperatively. In addition, it is difficult to observe a mirror image of a blade if several small air bubbles are formed instead of 1 large air bubble. Because there is no air injection or air–fluid exchange, there is no need to control the anterior pressure, especially during lamellar dissection. Second, there was no possibility of endophthalmitis because there was no procedure in which the eyeball was punctured. In this study, there were no cases of postoperative endophthalmitis. Third, because the technique is simple and easy, it has the advantage of shortening the time compared with other techniques. In particular, lamellar dissection took only 8.9 minutes on average (5–14 minutes). After reaching the optimal depth, stromal dissection was performed with a smooth corneal dissector, according to the natural lamellar plane of the stroma. Therefore, the residual stromal thickness with our technique was even compared to that with manual dissection using a sharp crescent blade. Our study also confirmed that the thickness of the recipient stroma was even on anterior segment OCT after surgery in all patients.

However, this technique also has some disadvantages. First, the residual stromal thickness was greater than that obtained using the other techniques. In the case of DALK using manual stromal dissection, the residual stromal thickness was reported to be 30.50 ± 27.60 μm.²⁸ In the case of DALK using the big-bubble technique, the average residual stromal thickness was 7.0 μm (range, 2.6–17.4 μm).²⁹ Melles et al did not report residual stromal thickness,²⁰ but according to another study in which DALK was performed using the Melles technique, residual stromal thickness values were <20 μm in 6 eyes, 20 to 80 μm in 5 eyes, and >80 μm in 6 eyes among 17 eyes.³ In this article, the thickness of the remaining stroma ranged from 27 to 132 μm, with an average of 80 μm. The residual stromal thickness and central corneal thickness (660 ± 69 μm) were thicker than that of other techniques; however, the residual stromal thickness was even in the anterior segment OCT, and the postoperative visual acuity was good. There were no cases of Descemet membrane rupture. As the surgeon gains more experience, the thickness will become thinner. Second, the recipient cornea must have a clear area at the periphery at least 2 clock hour to use retinal reflex. In the clear area, the gold line by the red reflex is visible, but it is not as visible as in keratoconus without stromal opacity (Fig. 5). If the entire cornea is opaque and there is no clear area, conventional manual dissection is inevitable because we cannot use the red reflex for stromal dissection of Descemet membrane. However, the Melles technique using total internal reflection at the air–cornea interface has the same disadvantage. If the opacity of the peripheral cornea is not severe, the Melles technique is more recommended than our technique because the mirror image of the crescent blade is more definite than the gold line because of the red reflex. Third, if the pupil is not fully dilated, our technique cannot be used because the red reflex is not possible in the peripheral cornea. In these cases, we have to use the Melles technique, conventional manual dissection, or Anwar big-bubble technique. Fourth, if the stromal scar is too deep, stromal opacity may persist after DALK using this technique, resulting in poor vision. If deep stromal opacity is confirmed preoperatively, stromal dissection should be performed as close as possible to Descemet membrane. If stromal opacity remains in the residual stroma in front of the pupil, additional manual dissection should be performed. Fifth, it is difficult to use this technique in keratoconus with severe ectasia and severe thinning. It is very difficult to perform intrastromal dissection to a very steep and ectatic cornea following the curved corneal profile to provide a final even thickness of the residual stromal bed. There is an increased risk of Descemet membrane perforation in this situation. In this case, we have to use conventional manual dissection. Therefore, we could include only 2 patients with moderate keratoconus in this study.

The following points should be noted during DALK using this technique. First, the eccentricity of the pupil should be assessed preoperatively without mydriasis. An eccentric pupil about the center of the graft cornea decreases vision postoperatively. However, it is impossible to determine the eccentricity of the pupil intraoperatively because the operation is performed with the pupil dilated. Second, if the initial vertical grooving is too shallow or if it starts far from the limbus (ie, too close to the center), when the crescent blade reaches the desired depth, the tip of the crescent blade may be inside the circular marks that the surgeon originally planned to remove (see Supplemental Figure, Supplemental Digital Content 2, http://links.lww.com/ICO/B617). Therefore, vertical grooving of the limbus should be sufficiently deep (depth of 1/3–1/2 corneal thickness) and be located close to the limbus.

In conclusion, we estimated the residual stromal thickness based on the gap between the gold line by the retinal reflex and crescent blade, and intrastromal lamellar dissection was performed using a smooth corneal dissector. Consequently, the surface of stromal dissection was smooth, and the residual stromal thickness was even. The duration of stromal dissection was short, and the postoperative visual acuity was good without complications, for example, Descemet perforation.

Supplementary Material