Secondary Logo

Journal Logo


Big-bubble technique to bare Descemet's membrane in anterior lamellar keratoplasty

Anwar, Mohammed FRCSa; Teichmann, Klaus D. MD*,b

Author Information
Journal of Cataract & Refractive Surgery: March 2002 - Volume 28 - Issue 3 - p 398-403
doi: 10.1016/S0886-3350(01)01181-6
  • Free


Lamellar keratoplasty (LKP) was developed more than 150 years ago.1 The first attempt to dissect near Descemet's membrane was by Hallermann.2 Anwar3 and Archila4 were the first to describe complete baring of Descemet's membrane in the recipient cornea. This technique promised to result in less interface opacity and hence improved visual acuity postoperatively, yielding vision comparable to that resulting from penetrating keratoplasty (PKP). However, few surgeons continued pursuing this goal5–12 because the rate of intraoperative perforation was high (39.2%)10 when one attempted to remove all corneal stroma. In addition, failure to bare Descemet's membrane was the outcome in a high proportion of cases.7,8,10 The crucial step, namely gaining access to the tissue plane immediately anterior to Descemet's membrane (the pre-Descemet's anatomic and potential surgical plane), presented the major difficulty in preparing an extremely deep lamellar bed6 and thus in performing maximum-depth anterior LKP.11

We present a technique that facilitates exposure of Descemet's membrane, considerably shortens the duration of the operation, lowers the risk of perforation, and has the potential to make maximum-depth LKP a popular procedure.

Surgical Technique

Partial-thickness trephination of the cornea is performed to a depth between 60% and 80% using a calibrated guided trephine system (Krumeich, Rhein Medical) that can be set to any depth. The Hanna suction trephine (Moria) is also accurate but can be set in 100 μm steps only. The Hessburg-Barron vacuum trephine (Jedmed Instrument Co., Katena Products Inc.) is less precise13 but easy to use and disposable.

A convenient point in the circumference of the cut is chosen as the entry site for the needle. If the trephine groove at this site is too shallow, manual deepening is performed.

A 27- or 30-gauge needle is attached to a 1 to 3 mL air-filled syringe. The needle is bent approximately 5.0 mm from its tip so that the terminal segment angles up approximately 60 degrees while the bevel faces down. The tip is introduced, bevel down, into the central corneal stroma at the chosen entry site, deep in the trephination groove.

Under direct visual control, the needle is carefully advanced into the partially severed central button, aiming in a direction halfway between a tangential and a radial one until the bevel is completely buried (Figure 1). In general, the tip of the needle reaches about 3.0 to 4.0 mm from the entry point in the trephination groove. The oblique direction of advancement avoids the central part of the cornea, which is usually the thinnest area, particularly in keratoconus. The bevel of the needle is kept strictly facing posteriorly so that the risk of puncturing Descemet's membrane is reduced and air entry into the pre-Descemet's plane is encouraged.

Figure 1.:
(Anwar) A 27-gauge needle is positioned bevel down in the paracentral area.

The plunger of the air-filled syringe is pressed until an effect is noted. This takes 1 of 2 forms. Most commonly, after initially infiltrating part of the central corneal disk, the air suddenly forms a large bubble with a more or less circular outline between Descemet's membrane and the deepest stroma (ie, in the pre-Descemet's plane). This is the desired result and is indicated by the sudden easing of resistance of the plunger of the syringe as well as the “explosive” appearance of a white, semiopaque disk (Figure 2). The edge of the disk may, but usually does not, coincide exactly with the trephination groove. The size of the bubble depends on the amount of air injected. Pressure on the plunger is immediately released, and the needle is withdrawn.

Figure 2.:
(Anwar) A white, semiopaque disk with a near-circular outline develops after air injection. This indicates the formation of a big bubble, which detaches the central Descemet's membrane, even though the bubble is not visible.

Less frequently, the air keeps infiltrating the central corneal disk without a discernible bubble appearing. In this case, the somewhat denser white opacity lacks the sharp, circular outline of the bubble. By the time the air extends peripherally to the trephine groove, bubble formation is unlikely and the surgeon should stop injecting air. This preserves some areas of clear corneal tissue (ie, in regions where the air has not penetrated) and prevents air from entering the anterior chamber through the trabecular meshwork.8

The aim should always be to achieve a large bubble. If the first attempt yields a local, fuzzy region of white opaque cornea, air injection is stopped, the needle is withdrawn, and the surgeon may repeat the procedure, starting at another point on the perimeter of the trephine groove where the cornea is still relatively clear. The needle tip should be visible at all times to avoid perforation. As each attempt leaves a part of the button opaque, more than 3 or 4 tries are usually not possible. When a big bubble does form, the appearance is unmistakable (Figure 2)

If a large bubble is not generated even with additional air injections, an anterior lamellar keratectomy can be performed in the traditional manner. The remaining stroma is then hydrated and considerably thickened by injecting balanced salt solution (BSS®) via a blunt cannula as described by Sugita and Kondo.10 After a few moments, the initially white cornea becomes semitransparent and further attempts to create a big air bubble can be performed as described above, with some chance of success.

When a big bubble is generated, a paracentesis is performed at a site peripheral to the edge of the large air bubble, but no fluid is drained at this stage. The surgeon should be very careful not to broach the big bubble when doing the paracentesis. If it collapses prematurely, entry into the pre-Descemet's plane will be more difficult.

A partial-thickness anterior keratectomy is then performed by dissecting with a #69 Beaver blade (Becton Dickinson & Co.), leaving a layer of corneal stroma in place anterior to the air bubble. (When a big bubble is achieved only after a keratectomy is made, no additional stromal tissue need be removed.) After the keratectomy is completed, some intraocular fluid is drained through the paracentesis. However, for good optical control, the surgical field is kept dry.

With the sharp tip of a pointed blade (eg, 30-degree Alcon ophthalmic knife) held almost parallel to the surface, the remaining layers of corneal stroma are penetrated near the center of the cornea (Figure 3). Under direct visual control, the knife tip is carefully advanced, entering the big air bubble in a nearly tangential manner and creating a small nick in the stromal layers. The surgeon attempts to keep holding the knife in 1 plane; if it is tilted after entering the bubble, air may escape prematurely. If possible, the opening should be made large enough so that later, it can be easily identified and a wire spatula inserted. If air escapes from the bubble, however, the knife must be immediately withdrawn, even if the opening is not yet large enough.

Figure 3.:
(Anwar) After the anterior stroma is removed by partial keratectomy, a small incision is made in the anterior wall of the big air bubble.

Collapse of the big air bubble is indicated by sudden darkening of the disk-shaped opacity (Figure 4). At this point, the eye may be softened further by draining more fluid from the paracentesis.

Figure 4.:
(Anwar) Moments after the incision in the bubble, air escapes and the bubble collapses, as indicated by darkening of the disk-shaped opacity.

An Anwar spatula formed like a blunt wire (K3–2310, Katena Products Inc.) is inserted through the tiny opening made by the sharp-tipped knife. The spatula is carefully advanced in the cleavage plane that was created by the air until its tip approaches the trephination groove. Although the spatula is in direct contact with Descemet's membrane, the risk of perforation is minimal as it is blunt and does not directly push against the membrane.

The spatula is held steadily and firmly while it is gently lifted anteriorly, slightly tenting the residual stromal layers. The layers are incised by filing or rubbing with a #69 Beaver blade (Figure 5). The spatula can also be scraped with a sharp knife (eg, Alcon 30-degree ophthalmic knife) held parallel to the spatula. Any semisharp tool such as a round, rotating smooth burr can be used; however, this seems to exert more pull on the collagen fibers and has a tendency to suddenly slip off the spatula in an uncontrolled manner.

Figure 5.:
(Anwar) An iris spatula is introduced into the space left by the collapsed bubble through the opening made by the point of the knife. The overlying stromal fibers are severed by rubbing over the spatula with the sharp edge of a Beaver blade.

The abraded “cut” can extend from the spatula's point of entry into the bubble to near its tip. It opens a line of entry into the pre-Descemet's plane (Figure 6) without requiring introduction of a sharp tool in the immediate vicinity of Descemet's membrane. However, when the spatula is rubbed with the Beaver blade, care must be taken to ensure that the spatula does not inadvertently slide outside the previously detached region as the spatula tip may perforate Descemet's membrane. The maneuver can be repeated in other directions, either at a 180-degree angle or in any other radial direction, to enlarge Descemet's membrane exposure. The corneal stromal layers are kept dry at all times to facilitate visual control of the ensuing dissection.

Figure 6.:
(Anwar) Two slits are made in the anterior wall of the collapsed bubble by rubbing, or filing, over the spatula as shown in Figure 5.

When 1 or more cuts into the remaining stroma are completed, the deepest stromal layers are circularly excised with a blunt-tipped microscissors (18010, Moria). The air bubble has opened this plane, so no firm adhesions remain, although there may be slight stickiness from drying tissue fluids. Eventually, the cut proceeds along the line of the trephination groove or just central to it. While the circular excision gradually bares increasingly larger areas of Descemet's membrane, the structure is repeatedly moistened by gentle application of a Weck-Cel sponge soaked in BSS to keep it from drying under the operating microscope lights (Figure 7). The corneal stroma is kept dry until the excision is complete.

Figure 7.:
(Anwar) The deepest stromal layers can usually be lifted off Descemet's membrane and safely excised using blunt-tipped scissors. The exposed Descemet's membrane is repeatedly moistened with a wet Weck-Cel sponge.

Finally, the corneal donor button is stripped of Descemet's membrane and endothelium (Figure 8) and sutured onto the bare Descemet's membrane in the eye. This prevents wrinkling of the donor Descemet's membrane at the interface. In many cases, the 2 layers can be detached from the graft in 1 piece. The edge of the corneal button is firmly seized with a fine-toothed forceps, with the surgeon attempting to grasp most of the corneal stroma but not Descemet's membrane and endothelium. A dry Weck-Cel sponge is applied near the forceps, at the inner edge of the button, to dislodge the innermost layers from the stroma by gently pulling centrally until a small area is detached. This detachment is extended by further pulling with the Weck-Cel sponge or using a tying (nontoothed) forceps. If the peeling layers tear, each piece is removed separately. Some grafts do not yield to this technique. When the attachment between Descemet's membrane and deep stroma is very firm, vigorous rubbing with dry Weck-Cel sponges tears the inner membranes and the layers are removed piecemeal. This has the disadvantage of leaving roughened stromal areas.

Figure 8.:
(Anwar) The donor button is prepared by stripping Descemet's membrane and endothelium.

In our experience, a big air bubble can be created in 80% to 90% of cases. If all attempts to form a bubble fail, baring Descemet's membrane is more tedious, time consuming, and hazardous. In such cases, an initial anterior lamellar keratectomy is made. The Katena spatula is inserted in a cornea-parallel plane at the bottom of the trephine cut and carefully moved forward in a probing manner. When a portion of stroma is undermined in this way, it is incised over the spatula using the method described above; then it is excised. The process is repeated and stroma removed layer by layer. Intraoperative perforations can usually be managed by intracameral air injection, and conversion to PKP may be required.

Indications and Precautions

Optical LKP requires a functioning endothelium. The techniques described work best in cases of keratoconus, stromal corneal dystrophies, and corneal scars that do not reach Descemet's membrane. If there is a history of hydrops, air injection into the deep stroma invariably leads to entry of air into the anterior chamber through the break(s) in Descemet's membrane; a big bubble will fail to form. In these cases, air injection is not advisable and complete baring of Descemet's membrane is not possible. Some stroma must be left behind to cover the area of the ruptured Descemet's membrane, a technique called near-full-thickness dissection. When firm scars extend to Descemet's membrane, baring the membrane is not possible or is very hazardous, and perforation is a likely outcome.


In 181 cases of keratoconus, an intraoperative perforation occurred in 16 eyes (9%). In 1 case, the procedure was converted to PKP.


One of us (M.A.) discovered that stromal air injection sometimes creates localized detachment of Descemet's membrane. This greatly facilitated baring of Descemet's membrane, so a method was developed and continuously refined until a big air bubble could be achieved in nearly all cases.

Initially, air injection into the corneal stroma was proposed by Archila.4 Later, it was used by Price,7 Chau and coauthors,8 and Morris and coauthors.14 All injected air before the partial-thickness trephination. Morris and coauthors14 also report cleaving Descemet's membrane from stroma by injecting air. They did not, however, reach this goal consistently, perhaps as a result of several aspects of their technique. Specifically, they injected air before doing the trephination. The needle was not advanced deep enough in the stroma, nor was it introduced bevel down. Entry of air into the anterior chamber, which was common, may have hampered the formation of a large central Descemet's detachment.

Sugita and Kondo10 injected saline to facilitate dissection but also did not try to create a large, central Descemet's detachment. They did, however, pioneer the use of a fine spatula to help separate Descemet's membrane from stroma. Others15,16 have injected viscoelastic substances for this purpose and report success in a small number of cases, but with a significant risk of perforation.16 The intracorneal depth location method developed by Melles and coauthors17 could prove useful during the advancement of the sharp, disposable needle before air injection, especially for less experienced surgeons. We have not used this technique yet because we want to avoid making a paracentesis before injecting the air into the stroma. The presence of a paracentesis can shunt air into the anterior chamber instead of creating a big bubble.

We have used the big-bubble method for more than 2 years, during which the technique has been continually refined. Used in the manner described here, it represents a reliable, safe way to gain access to the pre-Descemet's plane, making wide exposure of Descemet's membrane a realistic goal.

Lamellar keratoplasty has many advantages over PKP. These include maintenance of globe integrity, the possibility of using older graft material, a significant reduction in quantity and duration of postoperative steroid therapy, nearly complete absence of graft rejection, and a low rate of acute and chronic endothelial cell loss.10,14 The full benefits of LKP will be realized only with long-term follow-up.

Although there is a learning curve for surgeons embarking on this technique, major problems should not be encountered. If the surgeon fails to achieve a big bubble, he or she is left with a cornea that has been subjected to stromal air injection.4,8,14 Rupture of Descemet's membrane has been described with forceful injection of viscoelastic material16 and might also occur with air injection. We have not observed this. Perforation of the cornea during insertion of the needle is another possible complication that rules out generating a big bubble but does not preclude completing a successful near-Descemet's membrane dissection by other means. Our overall rate of perforation, even including surgeries during the learning curve and development phase, is significantly lower than that previously reported.10 Preliminary visual results appear comparable to those of PKP performed in Saudi Arabia.18,19 Because Descemet's membrane and endothelium are removed from the donor button, there is no visible wrinkling and interface haze is minimal or absent. The striae in the host Descemet's membrane seen in eyes with keratoconus tend to fade over time.

In conclusion, baring Descemet's membrane by removing all corneal stroma in the recipient bed combines the advantages of PKP and LKP while avoiding the disadvantages of both. The big-bubble technique is a significant advance in realizing this goal.


1. Mühlbauer FX. Ueber Transplantation der Cornea; Gekrönte Preisschrift. München, Jos. Lindauer, 1840. (abstract) In: Schmidt CC, ed, Jahrbücher der in und ausländischen gesamten Medizin. Leipzig, Otto Wigand, 1842; 267-268
2. Hallermann W. Verschiedenes über Keratoplastik. Klin Monatsbl Augenheilkd 1959; 135:252-259
3. Anwar M. Technique in lamellar keratoplasty. Trans Ophthalmol Soc UK 1974; 94:163-171
4. Archila EA. Deep lamellar keratoplasty dissection of host tissue with intrastromal air injection. Cornea 1984; 5
5. Amayem AF, Anwar M. Fluid lamellar keratoplasty in keratoconus. Ophthalmology 2000; 107:76-79
6. Naumann GOH, Sautter H. Chirurgie der Kornea. In: Mackensen G, Neubauer H, eds, Augenärztliche Operationen. Berlin, Heidelberg, Springer, 1988; vol 1:508
7. Price FW Jr. Air lamellar keratoplasty. Refract Corneal Surg 1989; 5:240-243
8. Chau GK, Dilly SA, Sheard CE, Rostron CK. Deep lamellar keratoplasty on air with lyophilised tissue. Br J Ophthalmol 1992; 76:646-650
9. Sugita J, Kondo J. [Lamellar keratoplasty and deep lamellar keratoplasty]. [Japanese] Folia Ophthalmol Jpn 1994; 45:1-3
10. Sugita J, Kondo J. Deep lamellar keratoplasty with complete removal of pathological stroma for vision improvement. Br J Ophthalmol 1997; 81:184-188
11. Teichmann KD. Lamellar keratoplasty—a comeback? Middle East J Ophthalmol 1999; 7:59-60
12. Tsubota K, Kaido M, Monden Y, et al. A new surgical technique for deep lamellar keratoplasty with single running suture adjustment. Am J Ophthalmol 1998; 126:1-8
13. Schor P, Rocha E, Cheema DP, Miller D. Calibration of trephine blade depth. J Refract Surg 1997; 13:89-90
14. Morris E, Kirwan JF, Sujatha S, Rostron CK. Corneal endothelial specular microscopy following deep lamellar keratoplasty with lyophilised tissue. Eye 1998; 12:619-622
15. Manche EE, Holland GN, Maloney RK. Deep lamellar keratoplasty using viscoelastic dissection. Arch Ophthalmol 1999; 117:1561-1565
16. Melles GRJ, Remeijer L, Geerards AJM, Beekhuis WH. A quick surgical technique for deep, anterior lamellar keratoplasty using visco-dissection. Cornea 2000; 19:427-432
17. Melles GRJ, Rietveld FJR, Beekhuis WH, Binder PS. A technique to visualize corneal incision and lamellar dissection depth during surgery. Cornea 1999; 18:80-86
18. Mahmood MA, Wagoner MD. Penetrating keratoplasty in eyes with keratoconus and vernal keratoconjunctivitis. Cornea 2000; 19:468-470
19. Mahmood MA, Wagoner MD. Penetrating keratoplasty for keratoconus: long-term results in 38 eyes with and 202 eyes without vernal keratoconjunctivitis. Middle East J Ophthalmol 1999; 7:26-35
© 2002 by Lippincott Williams & Wilkins, Inc.