This was a multicenter, retrospective chart review conducted using electronic medical records (EMRs) from 3 tertiary care centers in Toronto, Canada: St. Michael’s Hospital, Toronto Western Hospital, and the Kensington Eye Institute. Adult patients (aged ≥18 years) with a clinical diagnosis of NK who underwent PK with concomitant tarsorrhaphy between January 2017 and June 2025 were eligible for inclusion. Both permanent and temporary tarsorrhaphies performed during the same surgical session as PK were included. The diagnosis of NK was established clinically based on decreased or absent corneal sensation, assessed using a cotton wisp or tissue paper at the slit lamp, in the context of persistent epithelial defects, corneal ulceration, or other supportive findings. A minimum postoperative follow-up of 3 months was required for inclusion. Cases in which amniotic membrane transplantation (AMT) was used concurrently with tarsorrhaphy were excluded, given that AMT may confer an additional advantage in promoting epithelial healing.

This study was conducted in accordance with the Declaration of Helsinki and adhered to the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) guidelines. Institutional research ethics board approval was obtained and the requirement for individual informed consent was waived given the retrospective nature of the study and the use of anonymized data.

All tarsorrhaphies were performed using a standardized technique. After completion of PK, the eyelid margins were anesthetized using 1% lidocaine with epinephrine. The decision to perform a permanent vs temporary tarsorrhaphy was based on surgeon judgment, considering the severity of NK, degree of ocular surface exposure, anticipated duration of epithelial vulnerability, and the need for postoperative access to the graft. Permanent tarsorrhaphy was favored in eyes with severe disease or tectonic indications, such as corneal perforation or melt, where prolonged epithelial protection was anticipated. For permanent tarsorrhaphy, a raw tarsal bed was created to facilitate lasting adhesion between the lids. Using Westcott scissors, approximately 30% to 40% of the lateral upper and lower lid margin epithelium was trimmed at the grey line to expose bare, opposing tarsal edges. The lids were then permanently apposed using 5-0 polypropylene horizontal mattress sutures placed full-thickness through the grey line. Additional interrupted sutures were placed as needed to reinforce closure and maintain stable long-term apposition.

For temporary tarsorrhaphy, no lid margin epithelial scraping or removal was performed and the double armed 5-0 polypropylene suture was passed in a similar horizontal mattress fashion through the grey line of the upper and lower lids and tied tight at the lateral canthus. On occasion, to reduce the risk of the lids spreading open prematurely, 1 cm bolsters cut from 23-gauge butterfly tubing were used with 5-0 nylon passed through the lower and upper lids. The location and extent of closure (lateral, central, and nasal) were determined according to the patient’s exposure pattern, eyelid anatomy, and risk of epithelial breakdown. Lateral tarsorrhaphy was most commonly used to reduce overall palpebral fissure width, while nasal closure was selected in cases with medial exposure related to lower eyelid laxity or abnormal lid margin architecture, including prior eyelid surgery. Central tarsorrhaphy was typically performed as a temporary measure in select cases to provide additional early postoperative support. Care was taken to avoid lash follicle distortion and to maintain symmetric tension to ensure adequate lid apposition without excessive inversion.

At the conclusion of the procedure, topical antibiotic and steroid drops were administered, and the eye was padded and shielded until the following day. Topical antibiotics were discontinued once complete epithelialization of the graft occurred. Immunosuppression consisted of prednisolone acetate 1% every 4 hours while awake, with additional cyclosporine A (0.05%-1%) and/or tacrolimus (0.02% drops or 0.03% ointment) used based on the indication for PK and risk of rejection. Patients with a history of herpetic eye disease also received oral valacyclovir prophylaxis.

Demographic and clinical data were independently extracted from EMRs by 2 authors (R.S.H., M.A.), with discrepancies resolved through review by a third investigator (C.C.C.). Baseline variables included patient age, sex, laterality, lens status, ocular comorbidities, history of prior ocular surgeries, best-corrected visual acuity (BCVA), and indication for PK. Documented ocular comorbidities included glaucoma, retinal disease, autoimmune disease, prior ocular infection, clinically significant ocular surface disease in addition to NK (e.g., limbal stem cell deficiency, meibomian gland dysfunction), and eyelid abnormalities (e.g., eyelid malposition). Prior ocular surgeries recorded included keratoprosthesis implantation, corneal transplantation, glaucoma surgeries (e.g., trabeculectomy or tube shunt), eyelid surgeries, and vitrectomy. BCVA was measured using Snellen charts and converted to logMAR (logMAR) for statistical analysis, with all measurements taken using spectacle correction. Operative details collected included the size of the donor and recipient trephinations, the type (permanent vs temporary) and location (lateral, central, nasal) of tarsorrhaphy, and any concurrent procedures performed at the time of PK. NK severity was classified using the Mackie staging system based on documented clinical examination findings and surgical indications. Stage 1 was defined by epithelial irregularity or punctate keratopathy; Stage 2 by persistent epithelial defects; and Stage 3 by the presence of stromal melt or perforation.

The primary outcome was the proportion of eyes achieving complete epithelial closure following PK and tarsorrhaphy. Complete epithelial closure was defined as full re-epithelialization on slit-lamp examination with mild or no punctate staining. Secondary outcomes included the time to epithelial healing and the change in BCVA from preoperative baseline to final follow-up.

All variables were summarized using descriptive statistics. For continuous data, normality was assessed using the Shapiro-Wilk test. Variables with a normal distribution were expressed as means with SDs (SD), while non-normally distributed variables were presented as medians with IQRs (IQR). Categorical variables were described using frequencies and percentages. Paired t tests were used to evaluate within-subject changes in BCVA from baseline to final follow-up. To compare outcomes between the permanent and temporary tarsorrhaphy groups, independent-samples t tests were applied to continuous variables such as time to epithelial closure and final BCVA. The chi-square test was employed to compare categorical outcomes, including the proportion of eyes achieving complete epithelial healing. A 2-sided P -value of less than.05 was considered statistically significant. All analyses were performed using Stata version 17.0 (StataCorp LLC, College Station, TX).

A total of 85 eyes with NK that underwent PK with concomitant tarsorrhaphy were included in the analysis ( Table 1 ). The mean age was 68.5 ± 14.2 years, and 51.8% of eyes (n = 44) were pseudophakic at baseline prior to surgery. Common ocular comorbidities included a history of infection (61.2%, n = 52), glaucoma requiring chronic topical intraocular pressure–lowering therapy (24.7%, n = 21), and ocular surface disease in addition to NK (17.6%, n = 15). The latter included clinically significant meibomian gland dysfunction or blepharitis (15.3%, n = 13), limbal stem cell deficiency (11.8%, n = 10), and follicular conjunctivitis (2.4%, n = 2). Mechanical eyelid abnormalities were also observed, including lagophthalmos (7.1%, n = 6) and malposition due to ectropion or entropion (4.7%, n = 4). Additionally, autoimmune diseases were present in 8.2% (n = 7), and retinal pathology in 7.1% (n = 6). Nearly half of the eyes (49.4%, n = 42) had undergone prior PK, with smaller proportions having a history of glaucoma surgery (14.1%, n = 12; including 8 tube shunt implantations and 4 trabeculectomies, 2 eyes receiving mitomycin C), vitrectomy (8.2%, n = 7), keratoprosthesis implantation (3.5%, n = 3), or eyelid surgery (2.4%, n = 2). The mean baseline best-corrected visual acuity (BCVA) was 2.23 ± 0.34 logMAR (Snellen equivalent ∼20/3400).

The indications for PK included corneal perforation or melt (40.0%, n = 34), post-infectious corneal scarring (37.6%, n = 32), and failed graft (22.4%, n = 19). NK was most commonly attributed to herpetic keratitis as the primary etiology (44.7%, n = 38), with herpes simplex virus accounting for 78.9% (n = 30) and varicella zoster virus for 21.1% (n = 8) of herpetic cases. Other etiologies causing NK included autoimmune or severe ocular surface disease, such as Stevens-Johnson syndrome, Sjögren syndrome, ocular cicatricial pemphigoid, and severe dry eyes (14.1%, n = 12), postsurgical NK following neurosurgical or ocular procedures causing trigeminal or corneal nerve injury (11.8%, n = 10), exposure-related causes including facial nerve palsy and thyroid eye disease (11.8%, n = 10), diabetes mellitus-related NK (7.1%, n = 6), postinfectious NK following nonherpetic ulcers from Acanthamoeba or Pseudomonas (5.9%, n = 5), chemical injury (3.5%, n = 3), and medication toxicity (1.2%, n = 1). Using the Mackie classification, 15 eyes (17.6%) were classified as Stage 1, 36 eyes (42.4%) as Stage 2, and 34 eyes (40.0%) as Stage 3.