Highlight
• A 2025 Cochrane review of seven randomized controlled trials (n=494) reports low- to very low-certainty evidence for most interventions in neurotrophic keratopathy (NK).
• Recombinant human nerve growth factor (rhNGF; 20 μg/ml) showed a probable increase in partial corneal re-epithelialization (RR 1.88, 95% CI 1.37–2.58) but evidence for complete healing, visual acuity, and sensitivity is uncertain.
• Non-biologic topical agents (e.g., CACICOL20, RGN-259) and a single RCT of amniotic membrane transplantation provided insufficient or very uncertain benefit over controls.
Background: clinical context and unmet need
Neurotrophic keratopathy (NK) is a degenerative corneal disease caused by impairment of trigeminal corneal innervation. Loss of sensory and trophic neural input leads to decreased epithelial cell proliferation, reduced tear production, impaired wound healing and, in advanced stages, persistent epithelial defects (PEDs), stromal ulceration, melting and perforation. Etiologies are heterogeneous and include herpetic infection, surgical or traumatic trigeminal injury, diabetes mellitus, topical drug toxicity, long-standing contact lens use, and intracranial surgery or tumors.
Although NK is uncommon, consequences are potentially sight-threatening and management is challenging because pathophysiology targets corneal nerves rather than inflammation alone. Historically, treatment has been empirical, combining lubrication, prevention of trauma (tarsorrhaphy, therapeutic contact lenses), and surgical rescue (amniotic membrane, conjunctival flaps). The past decade has seen development of mechanism-based topical biologics (notably recombinant human nerve growth factor, rhNGF) and matrix-regenerating agents; however, comparative randomized data have been limited.
Study design and scope of the evidence
The Cochrane review by Kruoch et al. (2025) systematically searched multiple databases to 10 January 2025 and included only parallel-group randomized controlled trials (RCTs) comparing medical or surgical interventions against no treatment, placebo/vehicle, standard care (e.g., artificial tears, bandage contact lens) or alternative treatments for NK.
Seven RCTs (total n = 494; sample sizes 18–156; mean ages 25–68 years; 60–77% female) met inclusion criteria. Follow-up ranged from 28 days to 18 months. Six trials compared medical treatments against vehicle/artificial tears; medical agents included two biologics (rhNGF at 20 μg/ml in two RCTs; recombinant bovine basic fibroblast growth factor [rb-bFGF] in one trial) and three non-biologic topicals (0.1% RGN-259 [thymosin β4], Citicoline + vitamin B12 [Cit-B12], and CACICOL20/T4020 [RGTA-type matrix therapy]). One surgical RCT compared amniotic membrane transplantation (AMT) with tarsorrhaphy or bandage contact lens (BCL).
Primary outcomes were corneal re-epithelialization (variously defined), visual acuity, corneal sensitivity, disease progression/relapse, and adverse events. Outcomes were generally assessed at 1–3 months for healing, with disease progression evaluated at ≥6 months when available. Risk of bias was assessed with RoB 2 and certainty graded using GRADE.
Key findings — efficacy
rhNGF (recombinant human nerve growth factor, 20 μg/ml)
Two RCTs evaluated rhNGF versus vehicle/artificial tears (combined n ≈ 148 for some outcomes). Meta-analysis found that rhNGF may increase the proportion of participants achieving partial corneal re-epithelialization (defined in the review as <0.5 mm of corneal staining): RR 1.88 (95% CI 1.37–2.58), I2 = 0%; low-certainty evidence. For complete re-epithelialization (<0.1 mm staining), pooled effect favored rhNGF (RR 2.75, 95% CI 1.82–4.16) but heterogeneity was substantial (I2 = 72%); the certainty remained low.
On visual acuity, pooled analyses of two trials showed no clear improvement with rhNGF (MD −0.07 logMAR, 95% CI −0.58 to 0.71; I2 = 0%; low-certainty evidence), reflecting small or inconsistent effects on vision despite effects on epithelial closure. Two studies also reported lower recurrence rates of epithelial breakdown with rhNGF compared with vehicle (narrative reports), but follow-up durations and event numbers were limited.
Interpretation: rhNGF appears biologically plausible (restores trophic signaling) and likely improves epithelial healing rates in the short term; however, evidence is limited by small samples, industry sponsorship in several trials, and variable outcome definitions.
Recombinant bovine basic fibroblast growth factor (rb-bFGF)
One trial reported improvements in corneal sensitivity with rb-bFGF (mean difference 8.43 mm on Cochet–Bonnet esthesiometry; 95% CI 1.90–14.97; 60 participants) compared with control; this result was judged very low-certainty due to imprecision and indirectness. Data on re-epithelialization and vision were limited.
Non-biologic topicals: CACICOL20, RGN-259, Cit-B12
CACICOL20 (a matrix-regenerating agent) was evaluated in one RCT and did not significantly increase the proportion of participants attaining re-epithelialization versus control (RR 0.89, 95% CI 0.66–1.19; 148 participants; low-certainty evidence).
A small RCT of 0.1% RGN-259 (thymosin β4) reported an imprecise and non-significant effect on healing (RR 9.00, 95% CI 0.57–141.88; 18 participants; very low-certainty evidence) — the wide CI reflects extreme imprecision from a tiny sample. Citicoline + B12 had sparse randomized data and no clear superiority over vehicle in the reviewed trials.
Surgical interventions: amniotic membrane transplantation (AMT) versus tarsorrhaphy/BCL
Only one randomized trial evaluated a surgical approach (AMT vs tarsorrhaphy or BCL; n = 30). AMT may have little to no effect on the proportion achieving re-epithelialization at the evaluated time point (RR 1.10, 95% CI 0.69–1.76; very low-certainty evidence) and no clear benefit in visual acuity (RR 1.40, 95% CI 0.57–3.43). Reporting was limited and follow-up short, precluding strong conclusions.
Safety
Adverse events (AEs) were generally similar across groups in the trials that reported them. For CACICOL20, one trial found little to no difference in risk of AEs versus control (RR 0.98, 95% CI 0.63–1.52; 152 participants; low-certainty). Meta-analysis of two rhNGF trials showed no clear increase in investigator-reported AEs (RR 1.08, 95% CI 0.62–1.86; I2 = 0%; 151 participants; low-certainty). No major safety signals such as systemic toxicity were observed in the RCTs, but numbers were small and trials were typically industry-funded.
Expert commentary: interpreting the evidence and clinical implications
Strengths of the evidence base include randomized designs and inclusion of mechanism-based agents. However, the overall certainty of the evidence is low to very low for most outcomes because of imprecision (small trials and few events), indirectness (heterogeneous patient populations and outcome definitions), risk of bias concerns (some trials industry-sponsored and with varying blinding), and inconsistency (heterogeneity in some pooled estimates).
Clinical takeaways:
- rhNGF (cenegermin) shows the most consistent randomized efficacy signal for epithelial closure in NK and is mechanistically rational; it is the most evidence-backed disease-modifying topical therapy to date. Regulatory approvals for rhNGF (e.g., Oxervate/cenegermin in some jurisdictions) reflect these data, but clinicians should counsel patients on the limited evidence for functional outcomes such as vision and sensitivity beyond epithelial healing.
- Other topical agents (matrix therapies, thymosin β4, neurotrophic vitamin combinations) have not demonstrated reliable benefits in RCTs to date; small negative or inconclusive trials do not exclude benefit but demand cautious interpretation.
- Surgical measures—AMT, tarsorrhaphy, and BCL—remain important tools for mechanical protection and rescue in advanced disease; randomized evidence comparing them is minimal, and choice should be individualized based on the lesion, ocular surface status, patient preferences and resources.
Mechanistic insight supports early intervention to restore trophic signaling and protect the ocular surface while nerves regenerate. Clinicians must weigh costs, availability, and patient comorbidities (e.g., diabetes, ocular surface disease) when selecting interventions.
Limitations of the trials and gaps for future research
Important limitations highlighted by the review include small sample sizes, variable and sometimes undefined endpoints (differences in what constitutes “complete” epithelial healing), short follow-up for long-term outcomes (recurrence, vision loss), and limited reporting of objective nerve recovery measures (in vivo confocal microscopy or standardized esthesiometry). Trial heterogeneity in NK etiology and disease stage complicates pooling and generalizability.
Future trials should prioritize larger, multicenter randomized designs with standardized staging (e.g., Mackie stages), harmonized outcome definitions (time to durable complete epithelial closure, standardized visual function metrics), longer follow-up to capture recurrence and vision, objective nerve metrics, and health-economic evaluations given the high cost of some biologics. Independent funding or pre-specified strategies to mitigate sponsor bias will strengthen confidence in findings.
Conclusion
The current randomized evidence for interventions in neurotrophic keratopathy is limited. Recombinant human nerve growth factor (20 μg/ml) has the most evidence from randomized trials to support improved epithelial healing compared with vehicle/artificial tears, but the certainty of evidence for many clinically important outcomes (complete healing, visual acuity, corneal sensitivity, long-term disease control) remains low to very low. Other medical and surgical approaches are insufficiently supported by RCT data to draw firm conclusions. Until larger, harmonized trials are available, clinicians should individualize NK management using the best available evidence, consider mechanism-based agents where accessible and appropriate, and use protective surgical strategies for advanced lesions.
Funding and clinical trials registration
The Cochrane review reported no internal funding support; external support included the US National Eye Institute (NIH), Public Health Agency UK, Queen’s University Belfast and Birmingham Health Partners. The systematic review protocol is registered (doi: 10.1002/14651858.CD015723) .
References
1. Kruoch Z, Choo AY, Kemp A, Gonzales M, Yim TW, McCann P, Liu SH, Ting DSJ, Kuo IC. Medical and surgical interventions for neurotrophic keratopathy. Cochrane Database Syst Rev. 2025 Dec 5;12(12):CD015723. doi: 10.1002/14651858.CD015723.pub2 . PMID: 41347649 ; PMCID: PMC12679690 .
2. Oxervate (cenegermin-bkbj) [prescribing information]. U.S. Food and Drug Administration; 2018. (Product label and approval documents describe clinical trial data supporting epithelial healing in NK.)
3. Sacchetti M, Lambiase A. Diagnosis and management of neurotrophic keratitis. Clin Ophthalmol. 2014;8:571–579. doi: 10.2147/OPTH.S45915 .