Treatment: three targets, one escalation

Management pursues three objectives simultaneously:

1. Limit inflammation and iatrogenic injury — the first step is always one of withdrawal.
2. Promote epithelial healing — the cellular side (growth factors) and the matrix side (matrix therapies).
3. Limit stromal melting — antiproteases and interruption of epithelial-stromal interactions.

Mandatory starting point: identify the etiology and stop any potentially toxic medication. Look for dryness, surface disease, lagophthalmos, an eyelid abnormality. Stage according to Mackie — it is the stage that guides what follows.

What is actually prescribed: a snapshot of the series

At the first consultation, 344 eyes (97.2%) were already receiving medical treatment.

Among the specific treatments (autologous serum, RGTA, amniotic membrane), 87 eyes received a single modality and 42 a combination.

Antiproteases

Oral doxycycline, topical azithromycin (anti-collagenolytic effect independent of its antibacterial action), N-acetylcysteine. Preferred indication: progressive thinning, keratolysis, associated surface disease. Autologous serum also contributes here through its α2-macroglobulin and TIMPs.

Autologous serum and blood derivatives

Autologous serum provides growth factors (EGF, NGF, IGF-1), α2-macroglobulin and TIMPs at once — it therefore acts simultaneously on targets ② and ③. Documented alternatives: umbilical cord blood serum (Vajpayee, BJO 2003; Yoon, Ophthalmology 2007; Sharma, IOVS 2011; Erdem 2014) and plasma and platelet lysates (Alió, Ophthalmology 2007).

Two practical consequences: start it early (Guadilla et al.), but do not persist with it — a stage I under autologous serum that fails to improve must be reassessed and escalated, not prolonged. In the series, the main indication was a stage I persisting despite non-specific treatment, or a history of stage II-III.

Matrix therapy: RGTA / Cacicol

RGTA is a heparan sulfate mimetic that reconstitutes a transient matrix and protects heparan-sulfate-dependent growth factors from degradation (D. Barritault and J.-P. Caruelle, CRRET laboratory, Paris XII / OTR3). Efficacy was initially reported in neurotrophic ulcers (Aifa, Muraine et al., IOVS 2012: 11 patients, healing in 8.7 weeks, good tolerance; Arvola et al., Cornea 2016), and confirmed for post-crosslinking healing (Kymionis Cornea 2015; Bata JAMA Ophthalmol 2016; Gumus J Refract Surg 2017).

This dual correlation echoes the pre/post-ganglionic framework of Chapter 13: where the peripheral axon is preserved (central causes), matrix therapy has a functional target; where it is destroyed (herpes), it has less purchase.

Amniotic membrane transplantation

Amniotic membrane combines matrix + growth factors + antiproteases — it covers all three targets at once. Three modalities (Sippel et al., Curr Opin Ophthalmol 2001): inlay (same size as the ulcer, basement membrane facing up, integrated into the cornea), multilayer filling (deep stromal ulcerations), overlay (a true biological lens covering the cornea, sutured at a distance, acting as a dressing).

Series results: 145 procedures at stages II-III, 83 successes (57.2%), mean healing time 15.0 ± 14.9 days (< 7 days: 31%; 7-14 days: 40%; > 14 days: 29%), 1.44 ± 0.70 grafts per eye (1 to 5). The healing time under AMT is the shortest observed, with no significant difference from RGTA (p = 0.70).

Distribution by stage: the majority of eyes receiving an inlay were at stage II (27.9%) or III (69.8%); likewise for the overlay (36.4% and 57.1%). Six eyes initially at stage I required an overlay, and 12 eyes at stage II required overlay and inlay due to rapid deterioration.

The etiopathogenic treatment: recombinant NGF

The story begins with the discovery of NGF by Rita Levi-Montalcini and Stanley Cohen (Nobel Prize 1986), continues with recombinant murine NGF (Lambiase et al., NEJM 1998; Bonini et al., Ophthalmology 2000), and culminates in recombinant human NGF — cenegermin.

Randomized, double-masked, vehicle-controlled phase II trial, stages II-III (Bonini et al., Ophthalmology 2018):

It is the only treatment that directly addresses the cause — restoring the trophic signal and promoting nerve regeneration — rather than its epithelial consequences. In the same physiological logic, eye drops combining substance P and IGF-1 exploit the neuro-trophic synergy.

Eyelid protection

Essential in case of malocclusion or lagophthalmos: occlusive dressing (12.4% of eyes), partial or total tarsorrhaphy (7.3%), botulinum toxin injection into the levator (2.0%) (Cochrane Review 2013). Scleral lenses (7.1%) offer a non-surgical alternative for continuous protection and hydration.

Salvage surgery

Debridement of hyperplastic edges — a simple, often decisive gesture. Then, depending on extent and depth: conjunctival flap, cyanoacrylate glue, central "plug" graft (Gabison et al., Cornea 2011), therapeutic lamellar or penetrating keratoplasty. When the defect is too extensive for an amniotic membrane, grafting becomes the treatment of choice, urgently if necessary.

The way forward: corneal neurotization

Rather than supplementing trophic support, reinnervate the cornea. The principle consists of transferring a healthy sensory nerve — typically the contralateral supra-orbital and supratrochlear nerves, directly or via a nerve graft (sural nerve) — to the limbus of the anesthetic eye. Published results show objective recovery of sensitivity and reinnervation on confocal microscopy (Terzis et al., Plast Reconstr Surg 2009; Elbaz, Ali et al., JAMA Ophthalmol 2014;132:1289-95; Benkhatar et al., Cornea 2018). This is today the option to discuss in definitive corneal anesthesia in children and young adults, before embarking on a succession of grafts doomed to fail on a non-innervated background.

Practical strategy: escalation guided by stage

Key messages

1. Faced with any punctate keratitis, any epithelial defect, any ulcer: test corneal sensitivity. This is the gesture most often missed — 37.9% of diagnoses were delayed by a mean of 38.8 days.
2. The level of the lesion relative to the trigeminal ganglion predicts severity: pre-ganglionic or partial ganglionic → SPK; post-ganglionic or total ganglionic → ulcer.
3. Delayed healing = 7 days; persistent epithelial defect = 14 days. Delayed healing is the main risk factor for stromal melting.
4. The cause is multifactorial one time in three (34.2%): finding one etiology does not permit stopping the search.
5. The first treatment is a withdrawal: stopping iatrogenic causes. Fortified antibiotics were unnecessary in 16.4% of eyes, versus 6.5% of documented superinfections.
6. Hyperplastic edges are debrided. Fluorescein seeping under the edges suggests herpes.
7. Adapt to the stage: autologous serum at stage I (early, and without persisting — 21.7% success in 85 days, 17.9% progression); RGTA and amniotic membrane at stages II-III (63.6% and 57.2%); RGTA to be avoided at stage I (0%).
8. Stage I heals more slowly than stages II-III (78 vs 35 days, p < 0.001) — not due to severity, but to the poverty of the treatment arsenal. Do not mistake slowness for benignity.
9. Visual prognosis depends on three variables: initial acuity, initial Mackie stage, age. Neither etiology nor lagophthalmos determines it.
10. NK is an emergency. Treated and closely monitored, it heals in 79.5% of cases; untreated, a stage II-III heals spontaneously in only 13.7% of cases at 4 weeks.