DALK, hydrops
& decision algorithm
When transparency is lost, we rebuild the stroma while keeping the healthy endothelium: that is DALK, whose entire difficulty lies in the pre-Descemetic plane. Then the synthesis: a single logical thread, from screening to grafting.
DALK — deep anterior lamellar keratoplasty
Deep anterior lamellar keratoplasty is indicated in advanced keratoconus and PMD when scars or stromal opacities degrade vision while the endothelium remains healthy — that is, as long as there is no extensive cicatricial hydrops having compromised Descemet's membrane. Its value rests on a simple principle: by replacing the pathological stroma while preserving the recipient's Descemet and endothelium, it eliminates the risk of endothelial rejection, the main cause of late failure in penetrating keratoplasty. The graft is thus safer in the long term, at the cost of a technically more demanding surgery, whose whole difficulty concentrates on dissecting the pre-Descemetic plane.
The cleavage plane and the "big bubble"
Anwar's technique transformed DALK by making this plane reproducible. By injecting air (or a fluid) into the deep stroma, one detaches Descemet from the residual stroma and creates a bubble exposing a bare membrane on which the lamellar graft will rest. Dua's work then showed that this bubble does not always form in the same plane, and this distinction has direct surgical consequences.
| Bubble | Location / extension | Wall | Diameter | Cleavage plane | Risk |
|---|---|---|---|---|---|
| Type 1 | Central, centrifugal extension | Thick (Dua layer + Descemet) | up to ~8.5 mm | Above the pre-Descemetic layer | Low — the desired bubble |
| Type 2 | Peripheral, crown-shaped | Thin (Descemet only) | up to ~10.5 mm | Between Dua layer and Descemet | High — ruptures easily |
| Mixed | Coexistence of both | Heterogeneous | variable | Two simultaneous planes | The thin type-2 wall may give way when manipulating type 1 |
Manual techniques
In its manual form, big-bubble DALK chains a partial trephination, the insertion of a cannula or needle to two-thirds or three-quarters of stromal depth, then air injection that creates the bubble. It is then punctured, the stromal roof is cut, and dissection is completed down to bare Descemet, if needed under viscoelastic protection. When the bubble does not form — a frequent case in heavily scarred corneas — one falls back on layer-by-layer dissection, longer and sometimes leaving a less regular interface. It is precisely the depth uncertainty and the "blind" nature of needle insertion that expose to perforation, and that femtosecond assistance seeks to reduce.
Femtosecond assistance and double docking (DD-DALK)
The femtosecond laser first served to standardise the geometry of the donor-recipient junction, drawing interlocking side profiles — zigzag, mushroom, top-hat — that improve apposition and biomechanical strength of the scar. But the central difficulty of DALK is not the peripheral cut: it is entering the deep stroma and opening the bubble roof without injuring Descemet.
This is the logic of the double docking (DD-DALK) technique we described (Guindolet, Nguyen, Doan, Cochereau, Gabison), then evaluated on a consecutive series of 37 eyes published in Cornea in 2023. Rather than performing everything in a single applanation, the procedure is split into two successive dockings framing the air injection. During the first docking, under control of the laser-integrated intraoperative OCT — which allows the cuts to be visualised and calibrated directly on a cornea of irregular thickness — the anterior lamellar cut and the circular side cut are performed to remove the anterior stroma. One then undocks and injects air per Anwar into the residual posterior stroma to detach Descemet and obtain the big bubble. Finally comes the second docking: the residual posterior stroma is brought back into the docked position, and the laser performs a penetrating vertical cylindrical cut that opens the bubble roof in a controlled way, where the manual gesture would be freehand. The benefit is twofold: the precision of the stromal cuts is entrusted to the laser, and OCT guidance of the air injection increases the bubble-formation rate. In the 37-eye series, a big bubble was obtained in about two-thirds of cases, the remaining third requiring manual dissection; bubble failure was associated with the presence of stromal scars — a useful reminder that even a standardised technique remains dependent on the state of the stroma it crosses.
| Technique | Principle | Strengths | Limits |
|---|---|---|---|
| Manual big-bubble (Anwar) | Needle air injection at 2/3-3/4 of the stroma | Fast, simple instruments, historical reference | "Blind" insertion, perforation, operator-dependent |
| Layer-by-layer dissection | Progressive manual delamination if the bubble fails | Always-available fallback | Long, sometimes irregular interface |
| Femto "junction" | Interlocking femto side cut (zigzag/mushroom/top-hat) | Better apposition, better-controlled astigmatism | The big bubble stays manual; does not secure deep entry |
| DD-DALK (double docking + iOCT) | Two dockings framing air injection; bubble roof opened with a cylindrical cut | OCT-calibrated cuts, controlled opening, reproducibility | Femto + iOCT platform required; bubble success limited by scars |
Complications and management
Descemet perforation remains the feared complication. A micro-perforation usually allows the procedure to continue, tamponading with an anterior-chamber air bubble and avoiding any over-pressure; a macro-perforation mandates conversion to penetrating keratoplasty. Postoperatively, a Descemet detachment presents as a double anterior chamber, corrected by air re-injection. The other pitfalls concern the interface — fibrosis, haze, folds, irregularities, especially after manual dissection — and the possibility of stromal or epithelial rejection, bearing in mind that the absence of endothelial rejection remains the cardinal advantage of the technique.
Sutures and astigmatism management
Closure follows the same principles as penetrating keratoplasty. Interrupted sutures — classically sixteen 10-0 nylon sutures — are adjustable sector by sector and removed selectively according to topography; the running suture, single or double, in 10-0 or 11-0, distributes tension more harmoniously; the two are often combined. Sutures are passed at about 90 % depth, straddling the donor-recipient junction, without transfixing Descemet. Astigmatism reduction then plays out over time: early keratometry and topography guide loosening or selective removal of tight sutures on the steep meridian, with complete removal staged over twelve to eighteen months.
Acute corneal hydrops
- Mechanism: acute rupture of Descemet + endothelium at the cone apex → aqueous influx into the stroma → massive acute stromal oedema, abrupt visual loss, pain/photophobia.
- Background: very advanced cones, Down syndrome, VKC, rubbing.
Management
Treatment now relies on posterior stromal sutures (pre-Descemetic compression sutures), most often combined with an intracameral gas injection (SF6 20 % or C3F8 14 %): they appose the edges of the Descemet tear and markedly accelerate resorption of the oedema, shortening the course and limiting extensive scarring. This has become the mainstay of modern management, ahead of a purely conservative approach.
The residual stromal scar is sometimes flattening (paradoxical "self-treatment" of the cone), but may secondarily indicate DALK/PKP if the central opacity is disabling. Prevention: stop rubbing, manage atopy, CXL at the progressive stage before evolution to hydrops.
Synthesis decision algorithm
Diagnosis → Background (rubbing/atopy/familial) + Placido + Pentacam (posterior elevation + pachymetry + BAD-D) + Corvis TBI + epithelial mapping (doughnut) → ABCD staging.
| Situation | Management |
|---|---|
| Early stage, non-progressive, good vision | Glasses / soft lenses, stop rubbing, monitoring. |
| Documented progression | CXL (if pachy ≥ 400 µm; otherwise hypo-osmolar riboflavin) ± partial topo-guided PTK same session (Athens) if bothersome irregularity. |
| Symptomatic irregular astigmatism, workable cornea | ICRS/CAIRS (symmetric if symmetric cone; single/asymmetric segment if coma/decentred cone) ± CXL ± t-PTK. Prefer CAIRS if thin/steep cornea. |
| Lens intolerance / rehabilitation | Scleral lenses (often the best non-surgical option, PMD +++). |
| Stromal scar, healthy endothelium | DALK (type-1 big bubble sought; femto/DD-DALK for the junction and deep entry). |
| Extensive cicatricial hydrops / endothelial involvement | Penetrating keratoplasty. |
| PMD | Scleral lenses / band CXL / asymmetric inferior segment / crescentic lamellar surgery · SILK (sclerocorneal intrastromal) in advanced forms. |
A single logic runs through the whole spectrum: a matrix that loses its integrity (collagen/LOX genetics + MMP proteolysis + oxidative stress + defective cross-linking, triggered by rubbing) → a biomechanical deformation that we detect (tomography + biomechanics + epithelium), stabilise (CXL), regularise (ICRS/CAIRS, topo-guided PTK) and reconstruct (DALK/PKP) once transparency is lost.
Glossary & sources
- Anwar M, Teichmann KD. Big-bubble technique to bare Descemet's membrane in anterior lamellar keratoplasty. J Cataract Refract Surg 2002.
- Dua HS et al. Human corneal anatomy redefined: a novel pre-Descemet's layer (Dua's layer). Ophthalmology 2013.
- Guindolet D, Nguyen TT, Doan S, Cochereau I, Gabison EE. Double-docking femtosecond laser-assisted DALK. Cornea 2023.
- Jacob S et al. Corneal allogenic intrastromal ring segments (CAIRS) in keratoconus. J Refract Surg 2018.
- Belin MW, Duncan JK. Keratoconus: the ABCD grading system. Klin Monbl Augenheilkd 2016.
- Khaled ML, Bykhovskaya Y, … Rabinowitz YS, Liu Y. PPIP5K2 and keratoconus. Sci Rep 2019 · Akoto T et al. IOVS 2024.
- Kanellopoulos AJ. Athens protocol — same-day topography-guided PRK + high-fluence CXL.
- Guindolet D, Petrovic A, Doan S, Cochereau I, Gabison EE. Sclerocorneal intrastromal lamellar keratoplasty (SILK) for pellucid marginal degeneration. Cornea 2016;35(6):900-3.
- Theranostic-guided CXL with real-time riboflavin measurement and adaptive UV-A fluence: randomised "ARGO protocol"; early one-year real-world data (progressive keratoconus).