The inclusion criteria were unilateral Stage 2 or 3 NK refractory to one or more conventional treatments for at least 2 weeks, evidence of decreased corneal sensitivity, and the presence of negative corneal cultures, which were taken at admission to our department before the treatment. Finally, only patients with at least 12 months of follow-up after treatment were included.

The diagnosis of NK was established using the Mackie classification: Stage 2 [persistent epithelial defect (PED)] or Stage 3 [neurotrophic corneal ulcer (NCU)]. According to Mastropasqua et al., 2019, the Stage 3 enclosed Stage 3A (corneal ulceration with stromal thinning ≤50% of the total CT) and Stage 3B (corneal ulceration with stromal thinning ≥50% of the total CT). During the treatment period with cenegermin, therapeutic contact lens application and additional topical treatments were discontinued. In addition, possible impending perforation or surgical management needs, and active ocular infection or inflammation unrelated to NK were considered as the exclusion criteria, as were pregnancy and breastfeeding. The same eligibility criteria were applied to enclose the 16 historical controls.

All patients underwent a complete ophthalmic evaluation before starting the treatment and at each follow-up visit on a weekly basis for 8 weeks. Clinical assessment included slit-lamp (NIKON FS-3V Zoom Photo slit-lamp) examination implemented with corneal epithelial fluorescein staining, epithelialization times weekly monitoring, and AS-OCT (RTVue-100 applied anterior corneal module; Optovue, Fremont, CA). The size (mm) and depth (μm) of NK were recorded. Corneal hypoesthesia was confirmed using a Cochet–Bonnet aesthesiometer. At the first and at each follow-up visit, several AS-OCT scans were performed: pachymetric map, horizontal raster scan centered on the thinnest area of the cornea (17 horizontal scans within a height of 6 mm), cross-line (horizontal and vertical 8.00-mm scans), and line mode (horizontal 8.00-mm scans) on the area of minimum thinning. All scans were centered in the same area at the different visits. We identified the thinnest part of the cornea and measured the corneal and stromal thickness of this portion using the caliper tools that were built into the RTVue-100 system.

In the treatment group, we used a sterile, preservative-free ophthalmic solution containing 0.002% (0.02 mg/ml) of the active ingredient, cenegermin (Oxervate^®, Dompé Farmaceutici Spa, Milan, Italy). It was packaged in a carton of seven multidose vials (1.0 ml). The patients were instructed to use one vial per day before being discarded and to keep the remaining vials in the carton refrigerated between 2°C and 8°C until the time of use. The scheduled treatment was one drop, six times a day at 2-h intervals for 8 weeks.

In the control group, we used 50% AS eye drops. All patients were screened for blood-borne infections, including syphilis, Hepatitis B and C, and HIV serology before the preparation of AS eye drops. Venipuncture was performed at the antecubital fossa under aseptic conditions to collect 100 ml of whole blood into sterile containers. These were left standing for 2 h to ensure complete clotting. The blood was then centrifuged for 15 min at 3000 RPM. The serum was then separated in a sterile manner and diluted to 50% using saline solution. Patients were recommended to keep the serum at 4°C. We suggested using a bottle for 1 week and administering this six times daily for 8 weeks. Additionally, this group received a prophylactic topical antibiotic (Exocin, Allergan), twice a day. The patients were seen regularly every 7 days for 8 weeks.

• Corneal healing (time frame: at Week 8)

The primary outcome of this study was to achieve complete corneal healing of PED or NCU at Week 8, defined as no corneal fluorescein staining in the area of PED or NCU.

• Quantitative analysis and monitoring of the healing process (time frame: at Week 8)

The secondary end point was to assess the CT, stromal, and epithelial changes through AS-OCT (RTVue-100) and to register any adverse reactions to drug or any disease recurrence with at least 12 months of follow-up.

We compared the results of the AS-OCT scans with the biomicroscopic findings. The variables were summarized with mean and standard deviation for continuous variables and absolute value and percentage for frequency of categorical variables. In addition, a univariate linear regression analysis was performed to identify variables associated with the percentage change of CT by AS-OCT at the thinnest point. A standard statistical two-tailed paired t test was used to estimate the statistical significance of the differences between groups. All statistical analyses were performed with Stata version 15.1 (StataCorp LP, College Station, TX). The alpha level (type I error) was set at 0.05 for all analysis.

There were 11 female and five male patients with a mean age of 60.9 ± 21.1 years (range, 7–88 years) in the cenegermin group. In this group, five (31.2%) eyes experienced PED and 11 (68.8%) eyes had NCUs to varying degrees. According to the substages of NK, at Stage 3 of the Mackie's classification suggested by Mastropasqua et al. (2017), seven were Stage 3A and four were Stage 3B.

Affected eyes with PED: one recurrent herpetic epithelial lesion despite antiviral therapy, one epithelial defect after surgical treatment for acoustic nerve neuroma refractory to previously tarsorrhaphy, one PED developed on a previously transplanted cornea, one PED occurred in a child affected by a diffuse midline glioma, and one developed in the context of severe dry eye concomitant to corneal anesthesia.

Affected eyes with NCU: a cornea ulcer subsequent to a severe alkali burn, three recurrent herpetic stromal ulcers, and one NCU developed on a previously transplanted cornea in a diabetic patient. Four NCU occurred in eyes previously underwent surgical retinal detachment repair (Pars Plana Vitrectomy and endolaser). One NCU developed in the context of systemic polyneuropathy, and one occurred after a long-term topical glaucoma medication concomitant to corneal hypoesthesia.

All patients were treated by cenegermin eye drops six times a day. A total of 16 eyes of 16 patients healed completely with minimal scarring. There was no recurrence in these patients during the 12 months of follow-up time. Three eyes affected by PED (60% of PED) experienced epithelial hyperplasia regressed within the following month (Figure 1 and Figure 2). The demographic data, etiologies, stages of NK before the treatment, and percentage increase in CT after cenegermin administration are summarized in Table 1.

In the control group, there were 11 female and five male patients, and the mean age was 67.2 ± 18.4 years (range, 13–86 years). There were five (31.2%) PED and 11 (68.8%) NCU, with only two in Stage 3B.

Affected eyes with PED: three recurrent herpetic epithelial lesions, one PED developed in the context of corneal hypoesthesia, and one occurred after a long-term topical glaucoma medication.

Affected eyes with NCU: one recurrent herpetic stromal ulcer, one NCU developed on a previously transplanted cornea, and one NCU refractory to previously amniotic membrane transplantation. Two NCU occurred in eyes that had previously undergone surgical retinal detachment repair (Pars Plana Vitrectomy and endolaser) and two NCU resulting from post-surgical trigeminal nerve injury. One NCU developed in systemic polyneuropathy, and three NCU developed in the context of severe dry eye disease.

We administered 50% AS eye drops six times a day. A total of 16 eyes of 16 patients healed completely with minimal scarring. The follow-up time was 12 months, and the defect recurred after the suspension of therapy in four (25%) patients during the follow-up period. The demographic data, etiologies, stages of NK, and percentage increase in CT in the AS control group are summarized in Table 1.

As in the literature data (Bonini et al., 2018a; Bonini et al., 2018b), the most common cause of NK in our patients was herpes simplex keratoplasty, followed by ocular surgery, neurosurgical trigeminal damage, and severe dry eye disease (Table 2).

Additionally, three patients reported associated systemic conditions: DM, systemic polyneuropathy, and diffuse midline glioma, respectively.

The epithelialization was screened by fluorescein staining test on slit-lamp examination. The mean time for closure of the corneal lesion was 3.9 weeks ± 0.5 (range: 3–5 weeks) and 5.9 weeks ± 1.9 (range: 3–8 weeks) for the cenegermin group and AS group, respectively. The difference in epithelialization times between the two groups was statistically significant (p < 0.01).

Then in the cenegermin group, we clinically observed that the cornea healing was faster in patients younger than 75 years, but it was not statistically significant (p > 0.05).

In the AS control group, we observed a 25% recurrence rate compared with no recurrence in the cenegermin group.

During the first phase of NK healing, opaque slit-lamp scar tissue with high reflectivity on AS-OCT, looking different from the normal corneal stroma, was detected (Figure 3). At first, corneal epithelial hypertrophy filled the thinning area. Then scar tissue formed, and the corneal stroma thickness gradually improved. Second, when the scar tissue disappeared on slit-lamp examination, AS-OCT still detected an increased reflectivity of the anterior stroma at the corneal lesion site (Figure 4). In three PED, epithelial hyperplasia co-occurred with reepithelialization. It was appreciable with both the slit-lamp (Figure 1D,E and Figure2A) and AS-OCT (Figure 1F and Figure 2C). Finally, within a month, it resolved in all three cases and was no longer detectable (Figure 1G).

AS-OCT scans revealed an average CT in the thinnest point of 276.3 ± 74.1 μm at baseline and an average increase in CT of 176.5 ± 60.3 μm at the end of the cenegermin treatment (Week 8). Linear regression showed that the percentage increase of CT at the end of treatment with cenegermin was associated with the pre-treatment CT (B = −0.15; p = 0.035) (Graph 1). Table 3 shows the results of the regression analysis. No other variables such as demographic factors or the epithelial defect area were associated with changes detected by AS-OCT.

Compared with the control group, AS-OCT data showed that the average percent change in CT detection was 66.6% [95% confidence interval (CI), 20.6–96.2] in the cenegermin treatment group and 14.7% (95% CI, 9.1–23.9) in the control group. The difference of percentage AS-OCT increase in cornea thickness between the two groups was statistically significant (p < 0.02) (Graph 2 and Graph 3).

No side effects were observed in patients who received cenegermin or AS eye drops. Furthermore, there were no serious complications such as infectious keratoplasty that were encountered during the entire study period.