Treatment of Neurotrophic Keratitis With Topical Insulin

Neurotrophic keratitis (NK) is a degenerative condition characterized by slowed healing of the corneal epithelium resulting from partial or complete loss of corneal sensory innervation.¹ NK has an estimated prevalence of 1.6 to 4.2 cases per 10,000 people.² The condition can be commonly overlooked and quite recalcitrant to treatment, ultimately resulting in significant vision loss due to corneal scarring or a perforated globe.^1,3 The Mackie classification system sorts NK into various stages (Table).⁴ Persistent epithelial defect (PED) is an umbrella term for any nonhealing corneal epithelial defect that fails to show rapid re-epithelialization despite 10 to 14 days of standard treatments.⁵

The two go-to options for the treatment of PED are cenegermin-bkbj ophthalmic solution 0.002% 20 mcg/mL (Oxervate, Dompé) and autologous serum eye drops. However, there has been a recent influx of case reports and series touting the low cost and effectiveness of topical insulin (TI) in healing PEDs associated with neurotrophic keratitis in both diabetic^6-8 and nondiabetic^9,10 patients. This article attempts to address whether there is enough solid evidence of TI’s efficacy in treating NK relative to autologous serum and cenegermin.

CENEGERMIN, AUTOLOGOUS SERUM, AND TOPICAL INSULIN

Two phase 2 double-blind, vehicle-controlled randomized controlled trials (RCTs) have demonstrated the efficacy of cenegermin 0.002% 20 mcg/mL eye drops in helping re-epithelialize neurotrophic corneas.^11,12 Although there is a lack of RCT data, several studies have helped establish autologous serum drops as another standard treatment for recalcitrant PEDs.^13-17 However, cenegermin is costly if not covered by insurance. Arguable cons of autologous serum include risk of microbial contamination during preparation and storage that a compromised ocular surface may then be exposed to and the need to undergo a significant blood draw (likely multiple times a year). It is also good practice to rule out blood-borne pathogens and viral infections such as syphilis, HIV, and hepatitis before ordering serum tears.¹⁸ During my felllowship at the University of Chicago’s Illinois Eye Hospital, screening for inflammatory cytokines in the serum was completed before initiating serum drops in the Dry Eye and Ocular Graft vs Host Disease clinic run by Sandeep Jain, MD. For example, if wanting to be comprehensive, a 13-cytokine panel can be ordered consisting of serum IL-1B, IL-2, IL-12, IFN γ, IL-4, IL-5, IL-8, IL-10, IL-13, IL-17, TNF-α, and another marker of immune activation, IL-2 receptor (CD25).

The reports and studies involving TI for NK have largely centered on moderate NK, particularly PEDs recalcitrant to conventional initial treatments (eg, nonpreserved tears, bandage lenses, etc) (Table). Some investigators have even begun recommending TI as first-line therapy for refractory PEDs—above cenegermin and autologous serum.¹⁹ TI does not require a blood draw or screen and is relatively inexpensive to compound, with one source quoting $60 to $80 per bottle.²⁰ Topical insulin 1 IU/mL dissolved in polyethylene glycol 400 for a total bottle size of 10 mL to be used within 1 month (typically at 4 times daily) could be requested of the compounding pharmacy.²⁰

DOES TI HAVE A PLAUSIBLE MOA IN NK?

Nerve growth factor sustains corneal epithelial stem cells, aids re-innervation and healing after injury, and stimulates epithelial cell proliferation and differentiation.²¹ Meanwhile, autologous serum has epidermal and transforming growth factor, vitamin A and C, glucose, and natural antimicrobials (surface IgA, defensins, lysozyme), while also containing proteins involved in wound healing, such as fibronectin.²² It is thus logical that serum drops and nerve growth factor might facilitate corneal re-epithelialization. How, though, might TI help in the case of a neurotrophic cornea?

Insulin-like growth factor (IGF)-1 and IGF-2, along with insulin, are thought to cross-bind to their respective receptors and trigger signaling pathways inside corneal epithelial cells’ cytoplasm. For example, insulin binding to the insulin receptor may activate the Ras/extracellular signal-regulated kinase (ERK) pathway, promoting epithelial proliferation, differentiation, and migration. Meanwhile, insulin cross-binding to IGF-1 may activate both the Ras/ERK pathway and the phosphatidylinositol 3-kinase pathway, which promotes epithelial cell survival/anti-apoptosis.²¹ Insulin may also improve corneal nerve sensitivity by influencing corneal re-innervation.²² Additionally, TI may aid stromal keratocyte wound healing and prevent proteoglycan degradation.²³ Particularly helpful in diabetes, insulin binding to the insulin receptor may promote cytoplasmic accumulation of Β-catenin. This might ultimately activate the Wnt signaling pathway, promoting transcription and translation of genes and repair of corneal epithelial nerves.²⁴

SORTING THROUGH THE STUDIES

TI Versus Control Groups

Diaz-Valle et al published a retrospective study comparing 1 IU/mL fast-acting TI four times daily with autologous serum control in the treatment of refractory PEDs.²⁵ The size of their NK subgroup using autologous serum was n = 9, while the number using TI was n = 21. More eyes in the NK subgroup achieved complete re-epithelization with TI (P = .005). Eyes on TI ultimately required less amniotic membrane transplantation. Of note, the autologous serum concentration was also only 20%, markedly lower than the 50% we typically use at our institution.

At the 2024 Association for Research in Vision and Ophthalmology (ARVO) conference, Graue-Hernandez et al presented retrospective evidence on TI.²⁶ NK eyes treated with unspecified type 4 IU/mL TI had significantly (P = .000) less mean time to re-epithelialization (32.25 ± 4.40 days) compared with the control group treated with conventional treatment (82.48 ± 6.84 days). There were 27 eyes in the TI group versus 25 in the control group. The final defect area in incompletely healed corneas was significantly smaller in the insulin group. There were no differences in recurrences, complications, or subsequent surgical interventions between groups (P > 0.05). The frequency of insulin dosing was not shared in the meeting abstract, nor was the diabetes status of any patient.

Diabetic NK Studies With Control Groups

Diabetic (neurotrophic) keratopathy is an under-recognized complication of diabetes. NK can result from diabetic hypoesthesia/reduced nerve density and alterations in the corneal epithelium, basement membrane, and stroma. It can also result from panretinal photocoagulation-induced thermal injury to the long ciliary nerve branches as they enter the suprachoroidal space.²⁷

Eleiwa et al examined 50 IU/mL fast-acting TI in post-vitrectomy diabetic NK eyes that did not require intraoperative epithelial debridement. This group was retrospectively compared with a control group treated with preservative-free artificial tears and topical antibiotics. The test and control groups consisted of 18 and 19 eyes, respectively. About 75% of patients in each group had stage 2 NK; the rest had stage 3. Mean time until epithelial healing in patients with stage 2 NK was 8.4 ± 3.2 days via TI, significantly (P = .001) shorter than the 21.7 ± 11.1 days in the control group. For those with stage 3 NK, healing time was again significantly (P < .0001) shorter in the TI group (18.6 ± 2.5 days) versus the control group (48.0 ± 3.6 days). Two of the patients in the control group ultimately required amniotic membrane transplantation versus zero patients in the TI study group.

Compared with 0.15% sodium hyaluronate alone, Quiroz-Mendoza et al showed 0.5 IU/mL fast-acting TI significantly shortened resolution of epithelial defects in an RCT consisting of 36 patients with diabetes.⁸ Note, however, that these were patients who required epithelial debridement to enhance the view during vitrectomy surgery. Therefore, although diabetic, they did not have confirmed NK (as in the study by Eleiwa et al) or PEDs. This same issue holds in the 38-patient RCT by Dasrilsyah et al showing favorable corneal epithelial healing time via fast-acting 0.5 IU/mL TI four times daily in patients with diabetes undergoing intraoperative corneal debridement during vitreoretinal surgery.⁶ TI was associated with faster corneal epithelial healing compared with 0.18% sodium hyaluronate artificial tears four times daily.

Concentrations of TI

Although higher concentrations of TI up to 100 IU/mL were used in earlier decades, these doses do not appear to confer greater treatment efficacy. Interestingly, Zeisberg et al showed that higher doses of insulin were associated with more VEGF-A upregulation (and, therefore, risk of angiogenesis).²⁸ A TI concentration of 1 IU/mL is likely the most commonly used concentration in the literature,^{5,10,25,29,30} followed by 0.5 IU/mL.^6,8,9

Fai et al’s double-blind RCT consisted of 32 patients with corneal epithelial defects after corneal debridement in the context of diabetic vitreoretinal surgery.⁹ Patients were given insulin drops at three different concentrations (0.5 IU/mL, 1 IU/mL, and 2 IU/mL) four times daily. The 0.5 IU/mL concentration was associated with better healing (mean healing rate from maximum to minimum defect size) at a significance level of P = .036.

TI Without Control Groups

Krolo et al showcased four cases of NK of various (nondiabetic) etiologies with PEDs recalcitrant to lubricant and antiviral or bandage contact lens. The 1 IU/mL short-acting TI dosed four to eight times daily led to epithelial wound healing by 1 to 4 weeks in all four cases without adverse event or recurrence during a 2-week to 5-month follow-up period.²³

Soares et al showed that 1 IU/mL regular TI dosed four times daily helped achieved complete re-epithelialization in 19 of 21 patients with stage 2 or 3 NK and resolved refractory epithelial defects between 7 and 45 days. Four of these patients had diabetes.²⁹

Wang et al highlighted six cases of recalcitrant neurotrophic PEDs, which achieved re-epithelialization after a range of 7 to 25 days on 1 IU/mL regular TI dosed two to three times daily.³⁰ In addition to the small sample size and lack of control group, there was perhaps heterogeneity in the insulin drop creation, as drops were prepared by the pharmacy, the patient’s providers, or the patient themselves by injecting regular insulin into a new bottle of artificial tears with a polyethylene glycol and propylene glycol base. Although the healing effect of TI appears impressive (after amniotic membranes in multiple patients failed), it is difficult to establish with certainty the healing effect from TI alone, given the previously tried measures and the fact that it is not explicitly shared whether patients are concurrently on other topical medications. One patient was described as diabetic.

ARVO 2024 also featured a meeting abstract from Zeisberg et al who presented 20 eyes with refractory neurotrophic ulcers treated with 25 U/mL insulin, all of which healed after the addition of TI. Neither healing time nor diabetic status of these patients was described in the abstract.²⁸

Abdi et al examined TI in 23 patients with PEDs who did not respond to “intensive lubrication, bandage soft contact lenses, or other therapies within 2 weeks.”³¹ Patients with infectious keratitis, trauma, or acute chemical/thermal injury were excluded. Seven patients had diabetes. NK had various etiologies, including diabetes (n = 7), herpetic keratitis (n = 5), anesthetic topical drug abuse (n = 2), and facial nerve paralysis (n = 2). Seventy percent of patients (n = 16) achieved complete wound healing within 50 days. No adverse effects from TI were noted.

Individual Case Reports

Tong et al presented on a 55-year-old male with poorly controlled diabetes and bilateral corneal ulcers with severe hypoesthesia.³² He had failed nonpreserved tears, topical moxifloxacin, a tapering dose of topical prednisolone 1%, oral valacyclovir, bandage contact lenses, and tarsorrhaphy with no improvement in corneal ulcers with epithelial defects measuring 7.5 mm × 4.5 mm OD and 5 mm × 3 mm OS with rolled edges. A 25 IU/mL dose of TI (unspecified type) was administered six times daily in each eye and resulted in bilateral re-epithelialization within 1 week.

Galvis et al reported on a 48-year-old female with corneal ulcer/PED from ipsilateral facial paralysis status post-acoustic neuroma removal.³³ Despite 0.4% preservative-free sodium hyaluronate, palpebral occlusion during the night, topical antibiotics, topical steroids, autologous plasma enriched with platelet growth factor, oral ascorbic acid and doxycycline, and later bandage contact lens, the epithelial defect persisted. Taking 1 IU/mL of TI four times daily and discontinuing topical 3.1% vancomycin resulted in a reduction in epithelial defect size of about 50% after 5 days of treatment. Ultimately after application of another bandage contact lens and 14 days of TI treatment, the epithelial defect healed. It is unclear how much of an effect discontinuing the epithelial exposure to the toxicity of topical vancomycin 3.1% every hour had relative to the TI in this single-patient report.³⁴

Qasem et al³⁵ and Khilji et al³⁶ presented separate case reports of one patient each showing TI helping to heal herpes simplex keratitis-associated PEDs recalcitrant to earlier treatments. Both patients were post-cataract surgery. Qasem et al used 1 IU/mL regular TI dosed four to five times daily.³⁵ Qasem’s insulin treatment was accompanied by nonpreserved levofloxacin 1.5% ophthalmic solution (Iquix, Santen) as prophylaxis. Khilji used 1 IU/mL regular TI dosed six times daily, along with oral acyclovir 400 mg twice a day and topical acyclovir ointment 5% five times daily.³⁶

Giannaccare et al reported on a 40-year-old male with a recalcitrant, large (7 × 4 mm) central epithelial defect and one-third stromal thinning associated with iatrogenic trigeminal damage.³⁷ Nonpreserved artificial tears, vitamin A ointment, punctal plug, bandage contact lens, and autologous serum had been used for the previous 3 months without success. Topical fast-acting insulin administered 1 IU/mL four times daily with a special contact lens (Hyper-CL, EyeYon Medical) designed to serve as a drug reservoir resulted in healing of the defect by 20 days.

Moreker et al reported on a 55-year-old male with NK associated with right facial nerve palsy.³⁸ Examination showed absent corneal sensations and a central corneal ulcer, thinning and surrounding corneal haze, vascularization despite tarsorrhaphy, and treatment with a lubricant. Fast-acting TI (1 IU/mL) dosed four times daily led to partial healing of the ulcer by day 12 and some further, but still incomplete, healing by day 25.

Note that single case reports in particular may highlight the apparent success of a drug in one individual patient; however, other failed attempts tend not to be published (selective publication).³⁹ Furthermore, the positive effects from other agents used before TI in refractory cases cannot be accounted for and separated from the effect of TI. The unintended positive effect of withdrawing an agent such as a topical antibiotic, which, while important for prophylaxis, can have a toxic effect on the epithelium, or topical steroids, which can slow stromal wound healing, is also potentially confounding in TI case reports.^40,41

THE BOTTOM LINE

High evidence-level RCTs comparing TI in both diabetic and nondiabetic NK with topical standards of care are limited (as in the case of autologous serum)²⁵ or absent (as in the case of cenegermin).³¹ However, the seemingly impressive results from retrospective case series and reports does appear to merit these RCTs, especially when considering the low cost (relative to cenegermin), apparent safety, and plausible physiological healing benefits of TI. Suggested insulin concentration, type, and dosing frequency needs to be further studied and standardized, along with the vehicle the insulin is mixed with.