Medical and surgical interventions for neurotrophic keratopathy

Alterations to the corneal nerves can lead to neurotrophic keratopathy (NK), which is marked by breakdown of the corneal epithelium and impaired healing. If untreated, NK can cause severe corneal damage and significant visual impairment. Etiologies include infection, inflammation, chemical or mechanical trauma, systemic disease (e.g. diabetes mellitus), drug toxicity, contact lens overuse, and ophthalmic, neurosurgical, or otolaryngologic surgery. Treatments are often clinician-dependent and include an array of topical medications and surgical techniques to promote re-epithelialization and preserve vision. There is no consensus on the "gold standard" treatment.

To examine the efficacy and safety of medical and surgical interventions when compared with no treatment, placebo, standard care, or an alternative treatment, for people with neurotrophic keratopathy.

We searched CENTRAL, MEDLINE, Embase, PubMed, LILACS, and trial registries on 10 January 2025.

We included randomized controlled trials (RCTs) in which medical or surgical interventions were compared with no treatment, placebo (e.g. ophthalmic vehicle, normal saline), standard care (e.g. artificial tears, bandage contact lens (BCL)), or an alternative treatment.

Our outcomes of interest were corneal re-epithelialization, visual acuity, corneal sensitivity, worsening or relapse of the disease, and adverse events as assessed by investigators. We analyzed most outcome data at one to three months post-intervention; we assessed disease progression at six months or longer, and non-serious and serious adverse events at the longest follow-up time point.

Using Cochrane's risk of bias tool RoB 2, we assessed the risk of bias for outcomes reported in our summary of findings table.

We performed separate comparisons for medical and surgical interventions. In addition to the qualitative synthesis of included studies, where possible, we conducted fixed-effect meta-analyses, calculating risk ratios (RRs) with 95% confidence intervals (CI) for dichotomous outcomes and mean differences (MDs) and 95% CIs for continuous outcomes. We checked continuous data for skewness. We assessed the certainty of evidence using the GRADE approach.

We included seven parallel-group RCTs with a total of 494 participants (study sample size: 18 to 156 participants; mean age: 25 to 68 years; proportion of female participants: 60% to 77%). Follow-up duration in the studies ranged from 28 days to 18 months. Four studies (57.1%) were multicenter RCTs in the USA and Europe; three studies were conducted in Europe or Asia. Five studies (71.4%) that reported funding sources were industry-sponsored. Six studies compared medical interventions versus opthalmic vehicle or artificial tears; the medical interventions included two topical biologic interventions (20 μg/ml recombinant human nerve growth factor (rhNGF; two studies), and recombinant bovine basic fibroblast growth factor (rb-bFGF)), and three non-biologic interventions (0.1% RGN-259, topical citicoline and vitamin B12 (Cit-B12), and CACICOL20 (T4020)). One trial compared amniotic membrane transplantation with tarsorrhaphy or BCL for NK.

Six studies evaluated medical interventions versus vehicle or artificial tears. Meta-analysis of two studies showed that rhNGF may slightly improve corneal re-epithelialization (defined as < 0.5 mm of corneal staining) (RR 1.88, 95% CI 1.37 to 2.58; I² = 0%; 148 participants; low-certainty evidence). The same two studies assessed complete corneal re-epithelialization (defined as < 0.1 mm of corneal staining) and found similar results but with substantial heterogeneity in the meta-analysis (RR 2.75, 95% CI 1.82 to 4.16; I² = 72%; 148 participants; low-certainty evidence). One trial of CACICOL20 found that it may not increase the proportion of participants with corneal re-epithelialization (RR 0.89, 95% CI 0.66 to 1.19; 148 participants; low-certainty evidence), while one trial of 0.1% RGN-259 found the same (RR 9.00, 95% CI 0.57 to 141.88; 18 participants; low-certainty evidence). Meta-analysis of two studies showed that rhNGF may not improve visual acuity (RR 0.07, 95% CI -0.58 to 0.71; I² = 0%; 151 participants; low-certainty evidence). Meta-analysis of two studies suggested that medical interventions, specifically rb-bFGF, may improve corneal sensitivity (MD 8.43, 95% CI 1.90 to 14.97; I² = 0%; 60 participants; very low-certainty evidence), but the evidence is very uncertain. One trial showed that rhNGF may not improve corneal sensitivity (MD 1.08, 95% CI -0.32 to 2.48; 33 participants; very low-certainty evidence), but the evidence for this result is also very uncertain. Two studies narratively reported lower rates of disease recurrence with rhNGF compared with the opthalmic vehicle. One trial of CACICOL20 found that it may result in little to no difference in risk of adverse events compared with control groups (RR 0.98, 95% CI 0.63 to 1.52; 152 participants; low-certainty evidence); similar results were found in meta-analysis of two studies of rhNGF (RR 1.08, 95% CI 0.62 to 1.86; I² = 0%; 151 participants; low-certainty evidence). Only one trial evaluated surgical interventions, comparing amniotic membrane transplantation versus tarsorrhaphy or BCL. Amniotic membrane may have little to no effect on the proportion of participants with corneal re-epithelialization (RR 1.10, 95% CI 0.69 to 1.76; 30 participants; very low-certainty evidence) or improvements in visual acuity (RR 1.40, 95% CI 0.57 to 3.43; 30 participants; very low-certainty evidence), but the evidence is very uncertain. No other prespecified outcomes were reported in this comparison.

Neurotrophic keratopathy is a rare condition with an array of etiologies, which poses challenges for research in terms of study design, participant recruitment, and consensus on objective outcome measures. Our review found low or very low certainty of evidence regarding the effects of medical or surgical interventions on corneal re-epithelialization, visual acuity, and corneal sensitivity. We downgraded the certainty of the evidence mostly because of imprecision, followed by indirectness, risk of bias, and inconsistency. Given the current evidence and lack of universal guidelines, clinicians should individualize treatment based on clinical judgment and available resources. We expect future studies examining a wider range of interventions will be able to offer higher quality evidence and produce more conclusive assessments.

This review had no internal source of support. External sources: National Eye Institute, National Institutes of Health, USA; Public Health Agency, UK; Queen's University Belfast, UK; Birmingham Health Partners, UK.

Protocol available via doi.org/10.1002/14651858.CD015723.